# experience report

An experience report describes the history of a software project, critically reviews experience during development phases, and records lessons learned.

This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles.

## Prioritize, Don't Decide: What We Learned Auditing a Deployed Human-in-the-Loop CV Sorter

DevFeed: [Prioritize, Don't Decide: What We Learned Auditing a Deployed Human-in-the-Loop CV Sorter](<https://devfeed.tech/articles/prioritize-don-t-decide-what-we-learned-auditing-a-deployed-human-in-the-loop-cv-sorter-143383.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.23359>)

Author: Hadi Mohammadi, Jeroen Sparla, Alexandros Paliouras, Georg Krempl, Robert A. Bagheri

Published: 2026-11-19T05:00:00Z

Content type: news

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [human review](<https://devfeed.tech/topics/human-review.md>), [experience report](<https://devfeed.tech/topics/experience-report.md>)

Tags: [academic](<https://devfeed.tech/tags/academic.md>), [applications](<https://devfeed.tech/tags/applications.md>), [auditing](<https://devfeed.tech/tags/auditing.md>), [budget](<https://devfeed.tech/tags/budget.md>), [classifier](<https://devfeed.tech/tags/classifier.md>), [cold-start](<https://devfeed.tech/tags/cold-start.md>), [confidence](<https://devfeed.tech/tags/confidence.md>), [control](<https://devfeed.tech/tags/control.md>), [cross-encoder](<https://devfeed.tech/tags/cross-encoder.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [cs-ir](<https://devfeed.tech/tags/cs-ir.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cv](<https://devfeed.tech/tags/cv.md>), [data](<https://devfeed.tech/tags/data.md>), [database](<https://devfeed.tech/tags/database.md>), [discrimination](<https://devfeed.tech/tags/discrimination.md>), [embedding](<https://devfeed.tech/tags/embedding.md>), [embedding-model](<https://devfeed.tech/tags/embedding-model.md>), [encoder](<https://devfeed.tech/tags/encoder.md>), [human-in-the-loop](<https://devfeed.tech/tags/human-in-the-loop.md>), [llm](<https://devfeed.tech/tags/llm.md>), [load](<https://devfeed.tech/tags/load.md>), [local](<https://devfeed.tech/tags/local.md>)

### AI overview

The CV Priority Sorter proposes a reading order while leaving hiring decisions to the committee; its frozen multilingual encoder feeds a session-specific linear model that retrains after each label and is discarded at session end. In a 300-CV synthetic pool, the order surfaced 95% of relevant candidates after half the pile, but document length alone made precision-style metrics uninformative. Regenerating the pool with a shared length budget removed that shortcut while preserving the reading-depth result. The reported four-level agreement score is a thresholded binary probability, so it reflects classifier confidence rather than ranking and declines as labels accumulate. Uncertainty sampling preserved model quality but increased total committee reading above the no-label baseline by querying CVs that otherwise would be skipped. Under length control, hosted LLM embeddings improved zero-label ordinal discrimination from 0.66 to 0.80, but the gain faded after about two dozen labels; a local cross-encoder matched the cold-start gain without sending data out.

### Source excerpt

arXiv:2611.23359v1 Announce Type: cross Abstract: A single academic vacancy can draw hundreds of applications, and the first screening round consumes most of a committee's time. We describe a CV Priority Sorter deployed at a national platform for academic vacancies and integrated with its vacancy database. It proposes a reading order for a pile of CVs and leaves every decision to the committee, pairing a frozen multilingual transformer encoder with a per-session linear model that retrains after each label and is discarded when the session ends. It works: on a pool of 300 synthetic CVs, a committee reading in the tool's order reaches 95% of the relevant candidates after half the pile, where an arbitrary order needs almost all of it. This paper is about what we found when we audited that claim, rather than about the system itself. Three findings changed the product. The pool turned out to be sortable by document length alone, which makes precision-style metrics on it uninformative; regenerating it under a shared length budget removes that shortcut and leaves the reading-depth result standing. The four-level agreement score such systems report, ours included, is a binary probability cut at fixed thresholds, so it follows the classifier's confidence rather than its ranking and falls as labels accumulate. And uncertainty sampling, the obvious way to implement a "suggest the next CV" button, leaves model quality intact but raises the committee's total reading load above the no-label baseline, because the CVs it asks about are the ones that would otherwise be skipped. A fourth result shows where to look next: a hosted LLM embedding model lifts zero-label ordinal discrimination from 0.66 to 0.80 under length control, then stops mattering once the session model has two dozen labels; a local cross-encoder matches that cold-start gain without sending data out. We report the fixes each finding forced.

## From Brisbane to the rOpenSci Community: Ten Years of Open Research Software

DevFeed: [From Brisbane to the rOpenSci Community: Ten Years of Open Research Software](<https://devfeed.tech/articles/from-brisbane-to-the-ropensci-community-ten-years-of-open-research-software-144781.md>)

Original publisher: [Read original article](<https://ropensci.org/blog/2026/11/19/brisbane-to-community/>)

Author: rOpenSci

Published: 2026-11-19T00:00:00Z

Content type: article

Language: en

Sources: [rOpenSci](<https://devfeed.tech/sources/ropensci-open-tools-for-open-science.md>)

Topics: [Open Source](<https://devfeed.tech/topics/open-source.md>), [experience report](<https://devfeed.tech/topics/experience-report.md>), [Test-driven development](<https://devfeed.tech/topics/tdd.md>)

Tags: [2016](<https://devfeed.tech/tags/2016.md>), [about](<https://devfeed.tech/tags/about.md>), [also](<https://devfeed.tech/tags/also.md>), [australia](<https://devfeed.tech/tags/australia.md>), [but](<https://devfeed.tech/tags/but.md>), [code](<https://devfeed.tech/tags/code.md>), [community](<https://devfeed.tech/tags/community.md>), [confidence](<https://devfeed.tech/tags/confidence.md>), [connect](<https://devfeed.tech/tags/connect.md>), [contributed](<https://devfeed.tech/tags/contributed.md>), [data](<https://devfeed.tech/tags/data.md>), [developing](<https://devfeed.tech/tags/developing.md>), [doing](<https://devfeed.tech/tags/doing.md>), [event](<https://devfeed.tech/tags/event.md>), [experience](<https://devfeed.tech/tags/experience.md>), [first](<https://devfeed.tech/tags/first.md>), [open](<https://devfeed.tech/tags/open.md>), [research](<https://devfeed.tech/tags/research.md>), [ropensci](<https://devfeed.tech/tags/ropensci.md>), [software](<https://devfeed.tech/tags/software.md>)

### AI overview

The author's ten-year rOpenSci story shows how a welcoming, hands-on research software community can help newcomers contribute and grow. At the 2016 Brisbane Unconf, they improved code for downloading Australian weather grids and wrote a vignette; subsequent software peer review helped strengthen GSODR's focus, reproducibility and robustness, and taught them the importance of documentation, testing and ongoing maintenance. They later developed weather and climate packages, learned to test remote APIs with {vcr}, reviewed packages and became an editor. The author says reviewing others' code also improved their own practice. Lessons from rOpenSci informed Open Plant Pathology, which the founders wound down in 2025. They encourage contributions at many levels, from bug reports and documentation to package reviews, and argue that sustained community relationships and constructive feedback matter alongside the software. They note that rOpenSci must adapt to changing tools and growing editorial demands while keeping peer review useful and welcoming.

### Source excerpt

With rOpenSci turning 15 this year, it's hard to believe that I've been part of it for 10 of those years. My involvement started in Brisbane in 2016, at the first Australian rOpenSci Unconference. I'd known about rOpenSci and its work for some time before then. When rOpenSci was founded in 2011, I'd just moved from the US to the Philippines for work. During the five years I lived and worked there, I became familiar with its packages and the people behind them. I was interested in what rOpenSci was doing, but wasn't sure how I might get involved. I did recognise the value of being part of the R community, though, and joined the Manila R User Group. Brisbane 2016: just the beginning I moved to Australia for work in January 2016 and, as luck would have it, was living close enough to Brisbane to attend the rOpenSci Unconf in April. The event was organised by a group of researchers at QUT, including Nick Tierney, Miles McBain and Jessie Roberts. As it happened, my first R package, {GSODR}, had just been accepted onto CRAN. That was the fact I shared when introducing myself to the other attendees. I was an experienced researcher and R user, but still very much a newcomer to developing and publishing R packages. I was nervous about attending. I was new to Australia and had never been to an event quite like it. But the chance to bring Australian R users together to work on practical research problems was exciting, and I soon settled in. The format was very hands-on. We shared unfinished ideas, learned from one another and built things together. I found this a great way to learn, and it helped me gain confidence. I was one of the few attendees with experience working with geospatial data and a particular interest in weather data, which led me to connect with A/Prof Ivan Hanigan. Ivan had some R code for downloading AWAP grids, AWAP_GRIDS, and I was able to help improve it. I also contributed a vignette describing how to work with Australian weather data, which went on to inf

## S3 vs. ScyllaDB for tiny telemetry: What Red Canary learned the hard way

DevFeed: [S3 vs. ScyllaDB for tiny telemetry: What Red Canary learned the hard way](<https://devfeed.tech/articles/s3-vs-scylladb-for-tiny-telemetry-what-red-canary-learned-the-hard-way-140898.md>)

Original publisher: [Read original article](<https://thenewstack.io/scylladb-s3-tiny-telemetry/>)

Author: Cynthia Dunlop

Published: 2026-11-18T15:00:00Z

Content type: article

Language: en

Sources: [The New Stack](<https://devfeed.tech/sources/the-new-stack.md>)

Topics: [scylladb](<https://devfeed.tech/topics/scylladb.md>), [Endpoint Security & XDR](<https://devfeed.tech/topics/endpoint-security-xdr.md>), [Big data infrastructure](<https://devfeed.tech/topics/big-data-infrastructure.md>), [Amazon S3](<https://devfeed.tech/topics/amazon-s3.md>), [experience report](<https://devfeed.tech/topics/experience-report.md>)

Tags: [account](<https://devfeed.tech/tags/account.md>), [aws](<https://devfeed.tech/tags/aws.md>), [batch](<https://devfeed.tech/tags/batch.md>), [batching](<https://devfeed.tech/tags/batching.md>), [buffer](<https://devfeed.tech/tags/buffer.md>), [canary](<https://devfeed.tech/tags/canary.md>), [data](<https://devfeed.tech/tags/data.md>), [databases](<https://devfeed.tech/tags/databases.md>), [detection](<https://devfeed.tech/tags/detection.md>), [detection-and-response](<https://devfeed.tech/tags/detection-and-response.md>), [edr](<https://devfeed.tech/tags/edr.md>), [endpoint-detection-and-response](<https://devfeed.tech/tags/endpoint-detection-and-response.md>), [finops](<https://devfeed.tech/tags/finops.md>), [large](<https://devfeed.tech/tags/large.md>), [linux](<https://devfeed.tech/tags/linux.md>), [post-contributed](<https://devfeed.tech/tags/post-contributed.md>), [red](<https://devfeed.tech/tags/red.md>), [s3](<https://devfeed.tech/tags/s3.md>), [scylladb](<https://devfeed.tech/tags/scylladb.md>), [small](<https://devfeed.tech/tags/small.md>), [sponsor-scylladb](<https://devfeed.tech/tags/sponsor-scylladb.md>), [sponsored-post-contributed](<https://devfeed.tech/tags/sponsored-post-contributed.md>), [storage](<https://devfeed.tech/tags/storage.md>), [telemetry](<https://devfeed.tech/tags/telemetry.md>), [tiny](<https://devfeed.tech/tags/tiny.md>), [vs](<https://devfeed.tech/tags/vs.md>)

### AI overview

Red Canary reduced costs and improved endpoint data visibility by moving Linux EDR health and telemetry storage from S3 and Postgres to ScyllaDB Cloud. Buffering uploads through Firehose and Kinesis cut S3 PUT costs, but imposed a one-megabyte-per-shard limit, delayed visibility and made endpoint debugging harder; Postgres also struggled with write spikes. In ScyllaDB, the team used short TTLs and faster tombstone cleanup to improve reads, and split large customers across numbered buckets to avoid hot partitions. For telemetry, it partitioned by sensor ID and moved large process-memory arrays into an auxiliary table. A fork of the Rust driver enabled token lookup and execution without serializing statements twice, while separating CPU work from I/O; file processing fell from about 150 ms to 50 ms in this use case. Red Canary reports about 20x fewer writes than inserted rows and endpoint retrieval in seconds rather than minutes for its largest customers. The team is exploring further uses, including detection data; the excerpt does not establish results for those future plans.

### Source excerpt

At Red Canary, storing large amounts of small data is a big deal. Their Linux Endpoint Detection and Response (EDR) The post S3 vs. ScyllaDB for tiny telemetry: What Red Canary learned the hard way appeared first on The New Stack.

## Persistent Recursive Worlds Enable Autonomous Software Evolution

DevFeed: [Persistent Recursive Worlds Enable Autonomous Software Evolution](<https://devfeed.tech/articles/persistent-recursive-worlds-enable-autonomous-software-evolution-139568.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2608.10450>)

Author: Beichen Huang, Zhenyu Liang, Bowen Zheng, Ran Cheng

Published: 2026-11-18T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Software Engineering](<https://devfeed.tech/sources/arxiv-software-engineering.md>)

Topics: [Agentic AI](<https://devfeed.tech/topics/what-is-agentic-ai.md>), [long-running](<https://devfeed.tech/topics/long-running.md>), [agent orchestration](<https://devfeed.tech/topics/agent-orchestration.md>), [Spec Driven Development](<https://devfeed.tech/topics/spec-driven-development.md>), [experience report](<https://devfeed.tech/topics/experience-report.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agents](<https://devfeed.tech/tags/agents.md>), [all](<https://devfeed.tech/tags/all.md>), [also](<https://devfeed.tech/tags/also.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [autonomous](<https://devfeed.tech/tags/autonomous.md>), [autonomous-software](<https://devfeed.tech/tags/autonomous-software.md>), [backend](<https://devfeed.tech/tags/backend.md>), [basis](<https://devfeed.tech/tags/basis.md>), [bounded](<https://devfeed.tech/tags/bounded.md>), [c](<https://devfeed.tech/tags/c.md>), [can](<https://devfeed.tech/tags/can.md>), [case](<https://devfeed.tech/tags/case.md>), [changes](<https://devfeed.tech/tags/changes.md>), [code](<https://devfeed.tech/tags/code.md>), [comes](<https://devfeed.tech/tags/comes.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-ma](<https://devfeed.tech/tags/cs-ma.md>), [cs-ne](<https://devfeed.tech/tags/cs-ne.md>), [cs-se](<https://devfeed.tech/tags/cs-se.md>), [evolution](<https://devfeed.tech/tags/evolution.md>), [software](<https://devfeed.tech/tags/software.md>), [worlds](<https://devfeed.tech/tags/worlds.md>)

### AI overview

EvoX Genesis aims to make software development cumulative across short-lived agent sessions by having agents work from accepted repository versions, delegate within local scopes, and submit changes for verification and parent review. Accepted changes, contracts, and tests carry into later episodes. In one study, 1,019 episodes built a C compiler from scratch over 123.4 hours; repairs raised passes on a fixed 100-program Csmith set from 61 to 93, retained by a subsequent revision. A separate lineage continued from a GLM 5.2 snapshot inherited by DeepSeek V4 Flash. Other studies assessed a Rust redevelopment of selected MESA functionality against a Fortran reference across six workloads, and tracked WebGL repairs and regressions against conformance pages, reaching 2,017/2,019. The abstract presents these as evidence that inherited implementations and explicit evaluation criteria can support continued development; it does not establish that all trajectories avoid regressions.

### Source excerpt

arXiv:2608.10450v4 Announce Type: replace Abstract: Software systems accumulate dependencies and design commitments over time, while the agents that modify them operate within bounded sessions. How, then, can development remain cumulative across generations of transient agents? We introduce EvoX Genesis, which organizes development as a persistent recursive software world. Agents enter an accepted repository version, delegate work within local scopes, and submit changes for verification and parent review. Once accepted, those changes pass to later episodes together with the contracts and tests used to assess them. The project thus preserves both earlier work and a basis for evaluating what comes next. We examine this continuity through four studies of formation, continuation and redevelopment. Starting without compiler code, Genesis built a C compiler with 219,676 lines of Rust source through 1,019 agent episodes over 123.4 hours. Within this growing implementation, repairs increased passes on a fixed set of 100 Csmith programs from 61 to 93; a subsequent revision retained all 93 passes. Development also continued in a separate compiler lineage when DeepSeek V4 Flash inherited a GLM 5.2 snapshot. To examine continuity when the implementation itself changes, Genesis redeveloped selected MESA functionality in Rust, with numerical agreement assessed against the Fortran reference across six workloads. The WebGL trajectory, meanwhile, exposed both repairs and regressions as repeated checks recorded progress towards 2,017/2,019 conformance pages passed. A complementary EvoX Compiler case connects program-merging performance to executable graph and backend contracts. Together, these records trace how accepted work becomes a foundation for further development: later agents inherit both usable implementations and concrete ways to assess further revisions.

## Wrong yet Confident: Automated Detection of Silent Failures in Location-Dependent Software

DevFeed: [Wrong yet Confident: Automated Detection of Silent Failures in Location-Dependent Software](<https://devfeed.tech/articles/wrong-yet-confident-automated-detection-of-silent-failures-in-location-dependent-software-139575.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.21615>)

Author: Juan A. Padilla

Published: 2026-11-18T05:00:00Z

Content type: news

Language: en

Sources: [arXiv: Software Engineering](<https://devfeed.tech/sources/arxiv-software-engineering.md>)

Topics: [Test automation](<https://devfeed.tech/topics/test-automation.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>), [experience report](<https://devfeed.tech/topics/experience-report.md>)

Tags: [automated](<https://devfeed.tech/tags/automated.md>), [automated-testing](<https://devfeed.tech/tags/automated-testing.md>), [autonomous](<https://devfeed.tech/tags/autonomous.md>), [availability](<https://devfeed.tech/tags/availability.md>), [ci-cd](<https://devfeed.tech/tags/ci-cd.md>), [ci-cd-pipelines](<https://devfeed.tech/tags/ci-cd-pipelines.md>), [community](<https://devfeed.tech/tags/community.md>), [comparison](<https://devfeed.tech/tags/comparison.md>), [contributed](<https://devfeed.tech/tags/contributed.md>), [covid-19](<https://devfeed.tech/tags/covid-19.md>), [cs-se](<https://devfeed.tech/tags/cs-se.md>), [dataset](<https://devfeed.tech/tags/dataset.md>), [delivery](<https://devfeed.tech/tags/delivery.md>), [deployment](<https://devfeed.tech/tags/deployment.md>), [detection](<https://devfeed.tech/tags/detection.md>), [environments](<https://devfeed.tech/tags/environments.md>), [error](<https://devfeed.tech/tags/error.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [failures](<https://devfeed.tech/tags/failures.md>), [location](<https://devfeed.tech/tags/location.md>), [software](<https://devfeed.tech/tags/software.md>)

### AI overview

Across 32,019 delivery transactions (24,501 unique coordinates) from one USDA food assistance deployment in Bayamón, Puerto Rico, round-trip offset testing exposed silent geocoding errors that resolution-rate evaluation missed. The oracle reverse-geocodes an input coordinate, forward-geocodes the result, and measures the offset without manual correctness labels. Of the coordinates, 10.0% were unresolved by all three platforms. Among 13,221 coordinates Google resolved to civic addresses, 10.0% had offsets over 2 km with no error signal (5.4% of the full corpus), compared with 0.5% for the other commercial geocoder and 0.04% for the community geocoder. The authors report that this method detected every large-offset failure observed in the dataset, while resolution rate detected none. The results come from a single deployment; the abstract presents resolution rate as an availability measure and round-trip offset as a correctness measure, and describes lessons for integrating verification into CI/CD.

### Source excerpt

arXiv:2611.21615v1 Announce Type: new Abstract: We report an experience paper applying round-trip offset---an automated testing oracle for location-dependent software---at deployment scale across 32,019 GPS delivery transactions, corresponding to 24,501 unique coordinates, from a single USDA COVID-19 food assistance deployment in Bayam\'on, Puerto Rico. The oracle tests each geocoder's implicit spatial accuracy claim by measuring the deviation between an input coordinate and the coordinate recovered after reverse-then-forward geocoding, without requiring manual labeling of correctness. We evaluated one community-contributed and two commercial geocoders against this operationally validated corpus. The experience surfaces two distinct failure classes, which we report with their respective denominators because they are not comparable quantities. First, 2,441 of the 24,501 unique coordinates (10.0%) are invisible to all three platforms: an acknowledged failure with no resolution, and a property of the corpus rather than of any one geocoder. Second, and platform-specific: 10.0% of the 13,221 coordinates Google resolves to a civic address exhibit round-trip offsets exceeding 2~km---5.4% of the full corpus---returned with no error signal, against 0.5% for the second commercial platform and 0.04% for the community geocoder. Round-trip offset detects all large-offset failures observed in this dataset while standard resolution-rate evaluation detects none. We document an eight-mode fault taxonomy, a structured comparison of detection methods, and concrete lessons for integrating automated geocoder verification into CI/CD pipelines. The key lesson: resolution rate measures availability; round-trip offset measures correctness. In non-standard addressing environments these two metrics select for opposite behaviors, and only the latter catches the failures that matter for autonomous action.

## Deploying a Control-Logic LLM Assistant in Air-Gapped Power Plants: Lessons in Source Authority and Bounded Retrieval

DevFeed: [Deploying a Control-Logic LLM Assistant in Air-Gapped Power Plants: Lessons in Source Authority and Bounded Retrieval](<https://devfeed.tech/articles/deploying-a-control-logic-llm-assistant-in-air-gapped-power-plants-lessons-in-source-authority-and-bounded-retrieval-137566.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.20398>)

Author: Yiwei Fu, Nicholas Gabriel

Published: 2026-11-17T05:00:00Z

Content type: news

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [AI Integration Strategies](<https://devfeed.tech/topics/ai-integration-strategies.md>), [AI Chat](<https://devfeed.tech/topics/ai-chat.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [AI Agent](<https://devfeed.tech/topics/ai-agent.md>), [LLM security](<https://devfeed.tech/topics/llm-security.md>), [digital sovereignty](<https://devfeed.tech/topics/digital-sovereignty.md>), [experience report](<https://devfeed.tech/topics/experience-report.md>), [versioning](<https://devfeed.tech/topics/versioning.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [air-gapped](<https://devfeed.tech/tags/air-gapped.md>), [bounded](<https://devfeed.tech/tags/bounded.md>), [classification](<https://devfeed.tech/tags/classification.md>), [cloud](<https://devfeed.tech/tags/cloud.md>), [compute](<https://devfeed.tech/tags/compute.md>), [control](<https://devfeed.tech/tags/control.md>), [controllers](<https://devfeed.tech/tags/controllers.md>), [controls](<https://devfeed.tech/tags/controls.md>), [cost](<https://devfeed.tech/tags/cost.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [data](<https://devfeed.tech/tags/data.md>), [design](<https://devfeed.tech/tags/design.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [llm](<https://devfeed.tech/tags/llm.md>), [logic](<https://devfeed.tech/tags/logic.md>), [plants](<https://devfeed.tech/tags/plants.md>), [power](<https://devfeed.tech/tags/power.md>), [power-plants](<https://devfeed.tech/tags/power-plants.md>), [retrieval](<https://devfeed.tech/tags/retrieval.md>), [source](<https://devfeed.tech/tags/source.md>)

### AI overview

At two gas power plants, a control-logic evidence layer uses authoritative engineering exports and deterministic retrieval to reduce uncertainty in an on-premises operator assistant. A VLM reconstruction of 6,979 PDF pages achieved 0.51 connection precision and took 3.1 hours on 4xH200 GPUs; engineers then identified a native structured export. In a benchmark of 40 template-generated questions, explicit plans over eight graph operations used 20x fewer mean tokens and had 3x lower median latency than an LLM tool-calling agent, with identical retrieved node sets across repeated runs. The deployed design loads the export into a versioned knowledge graph and runs bounded graph operations in fixed software, reserving the local LLM for fallback intent classification and evidence-grounded explanations. Safeguards include abstention, entity disambiguation, graph versioning and read-only integration. One site's graph contains 221K nodes across seven controllers.

### Source excerpt

arXiv:2611.20398v1 Announce Type: new Abstract: Deploying Large Language Model (LLM) assistants in industrial operational technology environments poses constraints largely absent from cloud applications: site policies restrict engineering and operational data from leaving the plant, on-premises compute is limited, and incorrect causal attribution can mislead incident response. We report lessons from deploying a control-logic evidence layer that underpins a broader on-premises operator assistance system at two gas power plants. Two prototypes revealed where generative models added avoidable uncertainty and cost. The first used a Vision Language Model (VLM) to reconstruct control topology from PDF printouts routinely exchanged by engineers. Across the full 6,979-page document, extraction reached 0.51 connection precision and required 3.1 hours on 4xH200 GPUs; consultation with controls engineers then revealed a native structured export from the authoring system. The second used an LLM tool-calling agent to plan repeatedly over eight predefined graph operations. In a controlled benchmark of 40 template-generated questions, replacing this agent with explicit execution plans reduced mean token use by 20x and median latency by 3x, with identical retrieved node sets across repeated runs. The deployed design ingests the authoritative export into a versioned knowledge graph, executes bounded graph operations in fixed software, and confines the local LLM to fallback intent classification and evidence-grounded explanation. At one site, the graph contains 221K nodes across seven controllers. Our experience suggests a practical division of labor for high-consequence AI: LLMs interpret natural-language requests and explain results, while deterministic software constructs and retrieves technical evidence grounded in authoritative sources. Abstention, entity disambiguation, graph versioning, and read-only integration provide additional safeguards.

## Lessons from a Literary Clock

DevFeed: [Lessons from a Literary Clock](<https://devfeed.tech/articles/lessons-from-a-literary-clock-134190.md>)

Original publisher: [Read original article](<https://spin.atomicobject.com/lessons-from-a-literary-clock/>)

Author: Amanda Cowell

Published: 2026-11-14T13:00:12Z

Content type: article

Language: en

Sources: [Atomic Object](<https://devfeed.tech/sources/atomic-object.md>)

Topics: [experience report](<https://devfeed.tech/topics/experience-report.md>), [Claude Code](<https://devfeed.tech/topics/claude-code.md>), [3D](<https://devfeed.tech/topics/3d.md>)

Tags: [2](<https://devfeed.tech/tags/2.md>), [3](<https://devfeed.tech/tags/3.md>), [all](<https://devfeed.tech/tags/all.md>), [background](<https://devfeed.tech/tags/background.md>), [birthday](<https://devfeed.tech/tags/birthday.md>), [claude](<https://devfeed.tech/tags/claude.md>), [clock](<https://devfeed.tech/tags/clock.md>), [extracurricular-activities](<https://devfeed.tech/tags/extracurricular-activities.md>), [first](<https://devfeed.tech/tags/first.md>), [i](<https://devfeed.tech/tags/i.md>), [just](<https://devfeed.tech/tags/just.md>), [love](<https://devfeed.tech/tags/love.md>), [mind](<https://devfeed.tech/tags/mind.md>), [my](<https://devfeed.tech/tags/my.md>), [project](<https://devfeed.tech/tags/project.md>), [set](<https://devfeed.tech/tags/set.md>), [tech](<https://devfeed.tech/tags/tech.md>), [things](<https://devfeed.tech/tags/things.md>), [up](<https://devfeed.tech/tags/up.md>), [user-testing](<https://devfeed.tech/tags/user-testing.md>), [want](<https://devfeed.tech/tags/want.md>), [was](<https://devfeed.tech/tags/was.md>), [write](<https://devfeed.tech/tags/write.md>)

### AI overview

The author turned an old B&N Nook Glowlight into a literary clock that displays book quotes containing the current time, with Claude's help. Two approaches failed: a custom screensaver could not keep accurate time, and Opera could not install because the Nook's old Android could not verify its installer. Reading the system log revealed the cause, and the project switched to the device's existing web viewer. Testing the clock in daily use exposed clipping and overflow, prompting rules for quote length and text size; a watchdog restarts the viewer if the display stops changing. Asking about assumed limits also led to hiding the status bar and using editable plain text, though the old system limited font choices to ten built-in options. The author's takeaways include checking confident AI advice, pivoting after failed approaches, testing with real use, planning for failures, and asking about constraints. A friend's request to rotate the display was a final reminder that users may expect interactions the builder did not anticipate.

### Source excerpt

My close friend's birthday was coming up and they're hard to buy for. They buy themself all the things they want, so I set out to give them something I made, something that couldn't just be purchased off a shelf. They have a tech background and love to write, which brought to mind a project [...] The post Lessons from a Literary Clock appeared first on Atomic Spin.

## A Risk in the Clouds: NetClaw on GCP

DevFeed: [A Risk in the Clouds: NetClaw on GCP](<https://devfeed.tech/articles/a-risk-in-the-clouds-netclaw-on-gcp-131812.md>)

Original publisher: [Read original article](<https://automateyournetwork.com/netclaw/netclaw-on-gcp/>)

Author: John Capobianco

Published: 2026-11-12T12:00:00Z

Content type: article

Language: en

Sources: [Automate Your Network](<https://devfeed.tech/sources/automate-your-network.md>)

Topics: [experience report](<https://devfeed.tech/topics/experience-report.md>), [Cloud Networking](<https://devfeed.tech/topics/cloud-networking.md>), [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>), [Ollama](<https://devfeed.tech/topics/ollama.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>), [Pull Request](<https://devfeed.tech/topics/pull-request.md>)

Tags: [1-3](<https://devfeed.tech/tags/1-3.md>), [17](<https://devfeed.tech/tags/17.md>), [2](<https://devfeed.tech/tags/2.md>), [22](<https://devfeed.tech/tags/22.md>), [3](<https://devfeed.tech/tags/3.md>), [ai](<https://devfeed.tech/tags/ai.md>), [article](<https://devfeed.tech/tags/article.md>), [border](<https://devfeed.tech/tags/border.md>), [cloud](<https://devfeed.tech/tags/cloud.md>), [full](<https://devfeed.tech/tags/full.md>), [gcp](<https://devfeed.tech/tags/gcp.md>), [google-cloud](<https://devfeed.tech/tags/google-cloud.md>), [ha](<https://devfeed.tech/tags/ha.md>), [hybrid](<https://devfeed.tech/tags/hybrid.md>), [live](<https://devfeed.tech/tags/live.md>), [members](<https://devfeed.tech/tags/members.md>), [netclaw](<https://devfeed.tech/tags/netclaw.md>), [network-automation](<https://devfeed.tech/tags/network-automation.md>), [one](<https://devfeed.tech/tags/one.md>), [openshell](<https://devfeed.tech/tags/openshell.md>), [region](<https://devfeed.tech/tags/region.md>), [risk](<https://devfeed.tech/tags/risk.md>), [teardown](<https://devfeed.tech/tags/teardown.md>), [with](<https://devfeed.tech/tags/with.md>)

### AI overview

The article reports that NetClaw Borders and Members ran in GCP, on-premises, hybrid setups, and a live three-region Risk, but warns the system is not production-ready because Border URLs have no authentication: anyone with one can use the HUD and Chat and spend the model API key. In the tested design, Members dial out to Borders; multiple Borders share one execution authority, and loss of quorum blocks new protected work rather than authorizing duplicate execution. Multi-region tests exposed latency-sensitive admission windows, slow failover, and a distant two-region quorum that could not admit work. OpenShell with DefenseClaw added audited blocking and sandbox isolation, with measured overhead below half a percent of the 4,500 ms admission window, though one leadership timing issue remained unexplained. The tested workloads were read-only. The article recommends experimenting in a throwaway GCP project, setting budget alerts (which do not cap spending), keeping Border URLs private, and fully deleting resources afterward. Users, SSO, and RBAC are roadmap work, not current features.

### Source excerpt

From one Border to a three-region Risk: the full NetClaw on GCP story, R01-R09C, with cloud and hybrid Members, HA, OpenShell, DefenseClaw, live evidence and lessons from teardown.

## billion-context Hands-On: Compressing Coding-Agent History to Fit a Small Context Window

DevFeed: [billion-context Hands-On: Compressing Coding-Agent History to Fit a Small Context Window](<https://devfeed.tech/articles/billion-context-hands-on-compressing-coding-agent-history-to-fit-a-small-context-window-130184.md>)

Original publisher: [Read original article](<https://hackernoon.com/billion-context-hands-on-compressing-coding-agent-history-to-fit-a-small-context-window?source=rss>)

Author: Ishank Choudhary

Published: 2026-11-12T07:18:45Z

Content type: article

Language: en

Sources: [HackerNoon](<https://devfeed.tech/sources/hackernoon.md>)

Topics: [AI Context Management](<https://devfeed.tech/topics/ai-context-management.md>), [context window](<https://devfeed.tech/topics/context-window.md>), [Agent Harness](<https://devfeed.tech/topics/agent-harness.md>), [AI Agent](<https://devfeed.tech/topics/ai-agent.md>), [experience report](<https://devfeed.tech/topics/experience-report.md>)

Tags: [4k](<https://devfeed.tech/tags/4k.md>), [adapter](<https://devfeed.tech/tags/adapter.md>), [agent](<https://devfeed.tech/tags/agent.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [algorithm](<https://devfeed.tech/tags/algorithm.md>), [anthropic](<https://devfeed.tech/tags/anthropic.md>), [api](<https://devfeed.tech/tags/api.md>), [api-key](<https://devfeed.tech/tags/api-key.md>), [attribution](<https://devfeed.tech/tags/attribution.md>), [coding](<https://devfeed.tech/tags/coding.md>), [coding-agent](<https://devfeed.tech/tags/coding-agent.md>), [context](<https://devfeed.tech/tags/context.md>), [context-engineering](<https://devfeed.tech/tags/context-engineering.md>), [context-window](<https://devfeed.tech/tags/context-window.md>), [cost](<https://devfeed.tech/tags/cost.md>), [developer-tools](<https://devfeed.tech/tags/developer-tools.md>), [gemini](<https://devfeed.tech/tags/gemini.md>), [history](<https://devfeed.tech/tags/history.md>), [llm](<https://devfeed.tech/tags/llm.md>), [open-source](<https://devfeed.tech/tags/open-source.md>), [prompt](<https://devfeed.tech/tags/prompt.md>), [prompt-caching](<https://devfeed.tech/tags/prompt-caching.md>), [request](<https://devfeed.tech/tags/request.md>), [session](<https://devfeed.tech/tags/session.md>), [small](<https://devfeed.tech/tags/small.md>), [window](<https://devfeed.tech/tags/window.md>)

### AI overview

In a 41-request Gemini 3.5 Flash-Lite session, billion-context reduced the final prompt from 38,816 to 13,431 tokens (2.9x smaller) and cut total input tokens by 12.5%. It preserved all 10 tested user-provided facts; recalling details found only in tool output required search or decompression, and then matched the full-history agent's 6/8 score at substantially higher retrieval cost. The proxy adds about 4,060 tokens of tools and instructions per request. With prefix caching, this short session cost about $0.067 through the proxy versus an estimated $0.039 uncompressed, so compression did not save money here. The author sees value for long sessions near a context limit, with auditable, reversible folds. The tested Gemini OpenAI-compatible route failed on multi-turn tool calls because it dropped `thought_signature`; Gemini's native API worked. The excerpt also flags self-updates, license attribution, and the need to test before production use.

### Source excerpt

I ran a 41-request Gemini agent session through it: the final prompt shrank 2.9x and recall held, but on a short session it cost more

## Lessons from 59 Linux Kernel Bug Reports

DevFeed: [Lessons from 59 Linux Kernel Bug Reports](<https://devfeed.tech/articles/lessons-from-59-linux-kernel-bug-reports-129705.md>)

Original publisher: [Read original article](<https://www.microsoft.com/en-us/security/blog/2026/11/11/lessons-from-59-linux-kernel-bug-reports/>)

Author: FORGE Lab and Cen Zhang

Published: 2026-11-11T16:55:00Z

Content type: article

Language: en

Sources: [Microsoft Security](<https://devfeed.tech/sources/microsoft-security-blog.md>)

Topics: [Linux Kernel](<https://devfeed.tech/topics/linux-kernel.md>), [code security](<https://devfeed.tech/topics/code-security.md>), [Linux](<https://devfeed.tech/topics/linux.md>), [Security](<https://devfeed.tech/topics/security.md>), [Memory Safety](<https://devfeed.tech/topics/memory-safety.md>), [experience report](<https://devfeed.tech/topics/experience-report.md>), [SELinux](<https://devfeed.tech/topics/selinux.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [bug](<https://devfeed.tech/tags/bug.md>), [bug-reports](<https://devfeed.tech/tags/bug-reports.md>), [codename](<https://devfeed.tech/tags/codename.md>), [cves](<https://devfeed.tech/tags/cves.md>), [exploits](<https://devfeed.tech/tags/exploits.md>), [first](<https://devfeed.tech/tags/first.md>), [forge](<https://devfeed.tech/tags/forge.md>), [frontier-ai-models](<https://devfeed.tech/tags/frontier-ai-models.md>), [kernel](<https://devfeed.tech/tags/kernel.md>), [lab](<https://devfeed.tech/tags/lab.md>), [linux](<https://devfeed.tech/tags/linux.md>), [linux-kernel](<https://devfeed.tech/tags/linux-kernel.md>), [local](<https://devfeed.tech/tags/local.md>), [microsoft](<https://devfeed.tech/tags/microsoft.md>), [post](<https://devfeed.tech/tags/post.md>), [privilege-escalation](<https://devfeed.tech/tags/privilege-escalation.md>), [reports](<https://devfeed.tech/tags/reports.md>), [research](<https://devfeed.tech/tags/research.md>), [root](<https://devfeed.tech/tags/root.md>), [security](<https://devfeed.tech/tags/security.md>)

### AI overview

Microsoft FORGE Lab's MDASH scans of selected Linux kernel subsystems produced 26,502 raw findings in about two weeks. Screening and QEMU reproduction narrowed these to 621 PoC-backed cases; expert review yielded 59 upstream reports and 33 assigned CVEs, including eight with confirmed local root exploits. Fixing them required months of researcher and maintainer work: 38 reports have merged fixes, and the median time from first public patch to a mainline release was 41 days. The excerpt's CVE-2026-64563 example traces an eight-year-old rhashtable iterator bug to a missing pointer reset on one resume path; a reachable Netlink socket diagnostics caller made it exploitable. The researchers conclude that AI-assisted discovery can expand vulnerability research, but validating, reporting, and fixing findings remains human-intensive. The work covered only a subset of the kernel, and the authors say findings alone do not establish that the system is secure.

### Source excerpt

Microsoft FORGE Lab researchers used codename MDASH to investigate Linux kernel vulnerabilities, resulting in 59 upstream bug reports, 33 assigned CVEs, and 38 merged fixes. The research includes eight demonstrated local root privilege-escalation exploits and examines the challenges of turning AI-generated findings into validated vulnerabilities and upstream fixes. The post Lessons from 59 Linux Kernel Bug Reports appeared first on Microsoft Security Blog.

## Article: The iOS Crash We Could Not Reproduce

DevFeed: [Article: The iOS Crash We Could Not Reproduce](<https://devfeed.tech/articles/article-the-ios-crash-we-could-not-reproduce-128001.md>)

Original publisher: [Read original article](<https://www.infoq.com/articles/ios-crash-sigbus-debugging-experiment/>)

Author: Abhishek Gang Deb

Published: 2026-11-11T09:00:00Z

Content type: article

Language: en

Sources: [InfoQ](<https://devfeed.tech/sources/infoq.md>)

Topics: [iOS](<https://devfeed.tech/topics/ios.md>), [Post Mortem](<https://devfeed.tech/topics/post-mortem.md>), [experience report](<https://devfeed.tech/topics/experience-report.md>)

Tags: [article](<https://devfeed.tech/tags/article.md>), [but](<https://devfeed.tech/tags/but.md>), [cause](<https://devfeed.tech/tags/cause.md>), [change](<https://devfeed.tech/tags/change.md>), [code](<https://devfeed.tech/tags/code.md>), [concurrency](<https://devfeed.tech/tags/concurrency.md>), [consumer](<https://devfeed.tech/tags/consumer.md>), [crash](<https://devfeed.tech/tags/crash.md>), [debugging](<https://devfeed.tech/tags/debugging.md>), [development](<https://devfeed.tech/tags/development.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [investigation](<https://devfeed.tech/tags/investigation.md>), [ios](<https://devfeed.tech/tags/ios.md>), [ios-crash-sigbus-debugging-experiment](<https://devfeed.tech/tags/ios-crash-sigbus-debugging-experiment.md>), [large](<https://devfeed.tech/tags/large.md>), [local](<https://devfeed.tech/tags/local.md>), [mobile](<https://devfeed.tech/tags/mobile.md>), [operating-systems](<https://devfeed.tech/tags/operating-systems.md>), [production](<https://devfeed.tech/tags/production.md>), [report](<https://devfeed.tech/tags/report.md>), [reproduction](<https://devfeed.tech/tags/reproduction.md>), [required](<https://devfeed.tech/tags/required.md>), [root](<https://devfeed.tech/tags/root.md>), [scale](<https://devfeed.tech/tags/scale.md>), [set](<https://devfeed.tech/tags/set.md>), [small](<https://devfeed.tech/tags/small.md>), [testing](<https://devfeed.tech/tags/testing.md>)

### AI overview

An intermittent SIGBUS crash was traced to HTML import through `NSAttributedString` running without a main-thread guarantee during asynchronous screen preparation. The signal and Foundation stack frames narrowed the search but did not prove a cause; an initial hotfix reduced overall crashes while the targeted signature persisted. Once a QA device reproduced the failure after several attempts, repeated builds that removed one view or behavior at a time narrowed it to the HTML conversion. The repair routed all such conversions through a shared `@MainActor` helper, keeping networking and other suitable preparation asynchronous. Validation included retesting the reproducing device and affected views, checking that call sites used the helper, and monitoring the targeted production signature after release. The article recommends treating crash-rate changes separately from evidence that a specific defect is fixed.

### Source excerpt

This article describes the investigation to find the root cause of a production crash in a large-scale iOS consumer app that had no clean stack trace. The crash occurred intermittently and resisted every local reproduction attempt. The final code change was small, but identifying the root cause required turning a vague SIGBUS report into a controlled set of experiments. By Abhishek Gang Deb

## How a designer learned to unsee Will Smith eating spaghetti

DevFeed: [How a designer learned to unsee Will Smith eating spaghetti](<https://devfeed.tech/articles/how-a-designer-learned-to-unsee-will-smith-eating-spaghetti-127635.md>)

Original publisher: [Read original article](<https://www.mux.com/blog/how-a-designer-learned-to-unsee-will-smith-eating-spaghetti>)

Author: Rob Mach

Published: 2026-11-10T22:37:04Z

Content type: opinion

Language: en

Sources: [Mux Blog - Video technology and more](<https://devfeed.tech/sources/mux-blog-video-technology-and-more.md>)

Topics: [UI/UX Design](<https://devfeed.tech/topics/ui-ux-design.md>), [Claude Code](<https://devfeed.tech/topics/claude-code.md>), [design to code tools](<https://devfeed.tech/topics/design-to-code-tools.md>), [Figma](<https://devfeed.tech/topics/figma.md>), [browser](<https://devfeed.tech/topics/browser.md>), [generative-ai-tools](<https://devfeed.tech/topics/generative-ai-tools.md>), [experience report](<https://devfeed.tech/topics/experience-report.md>)

Tags: [2](<https://devfeed.tech/tags/2.md>), [2026](<https://devfeed.tech/tags/2026.md>), [3](<https://devfeed.tech/tags/3.md>), [about](<https://devfeed.tech/tags/about.md>), [account](<https://devfeed.tech/tags/account.md>), [after](<https://devfeed.tech/tags/after.md>), [browser](<https://devfeed.tech/tags/browser.md>), [build](<https://devfeed.tech/tags/build.md>), [claude](<https://devfeed.tech/tags/claude.md>), [claude-code](<https://devfeed.tech/tags/claude-code.md>), [code](<https://devfeed.tech/tags/code.md>), [company](<https://devfeed.tech/tags/company.md>), [creative](<https://devfeed.tech/tags/creative.md>), [custom](<https://devfeed.tech/tags/custom.md>), [design](<https://devfeed.tech/tags/design.md>), [designer](<https://devfeed.tech/tags/designer.md>), [figma](<https://devfeed.tech/tags/figma.md>), [review](<https://devfeed.tech/tags/review.md>), [skin](<https://devfeed.tech/tags/skin.md>), [using](<https://devfeed.tech/tags/using.md>), [video-js](<https://devfeed.tech/tags/video-js.md>)

### AI overview

A brand designer with no development background built and deployed a custom Video.js v10 player skin by pairing Claude Code with Figma and browser review. After installing Node.js and using the Video.js skill, an initial prompt produced a working player on Vercel in about 30 minutes. The ongoing process was iterative: generate a crude starting point, inspect and refine it in the browser, return to Figma for considered design work such as icons, then have Claude implement those details and review again. Figma remained the place for creative exploration, the browser for judging results, and Claude Code connected design to implementation. The author says the player still needs refinement before meeting Mux's standards; the account describes one designer's experience, not a general guarantee.

### Source excerpt

How Mux's Creative Director designed and built a custom Video.js v10 skin himself, using Figma to design, Claude Code to build, and the browser to review.

## Mismeasuring the Life of a Line of Code: Lessons Learned the Hard Way

DevFeed: [Mismeasuring the Life of a Line of Code: Lessons Learned the Hard Way](<https://devfeed.tech/articles/mismeasuring-the-life-of-a-line-of-code-lessons-learned-the-hard-way-127234.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.10530>)

Author: Pavel Gurov

Published: 2026-11-10T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Software Engineering](<https://devfeed.tech/sources/arxiv-software-engineering.md>)

Topics: [experience report](<https://devfeed.tech/topics/experience-report.md>), [development-process](<https://devfeed.tech/topics/development-process.md>), [AI-assisted coding](<https://devfeed.tech/topics/ai-assisted-coding.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [cs-se](<https://devfeed.tech/tags/cs-se.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [typescript](<https://devfeed.tech/tags/typescript.md>)

### AI overview

Across 32.5 million line births in 120 TypeScript repositories, choices in measuring code-line lifetimes substantially changed the findings. Counting moved or rewritten lines as dead reduced median life from 3.0 years to 1.3; treating any change as death lowered within-project concordance from 0.71 to 0.56 and concealed predictable local revision. Regressing observed lifetimes discarded three quarters of lines, while treating lines as independent made nine of fifteen effects appear significant only under that assumption. After corrections, models could barely rank removal within a project, though they ranked local revision at 0.71 concordance; the top 1% of new lines were revised at 2.3 times the base rate. The abstract says the paper provides a checklist of symptoms for these measurement pitfalls.

### Source excerpt

arXiv:2611.10530v1 Announce Type: new Abstract: Studies of how long lines of code live, including lines written by AI agents, rest on choices that look technical: when a line counts as gone, what to do with lines still alive, whether to pool projects, and how to evaluate a model. We studied 32.5 million line births in 120 TypeScript repositories and got several of these choices wrong at first. For ten of them, we compare the result as we first computed it, or as commonly done, with the result after correction. All ten change conclusions, or would have. Six were our own mistakes. Counting moved and rewritten lines as dead shortens a line's median life from 3.0 years to 1.3. Counting any change as a death hides an outcome that can be predicted: within-project concordance falls from 0.71 to 0.56. Regressing observed lifetimes discards three quarters of the lines. Keeping rewritten lines at risk makes protective effects grow with age. Lines are born and removed together, so treating them as independent misleads twice: nine of fifteen effects look significant only under that assumption, and a split by file makes a barely predictable outcome look good (within-project concordance 0.71 instead of 0.61). After the corrections, removal depends on a line's age and project, but our models could barely rank it within a project. Local revision they could rank (0.71). Obvious cases aside, the top 1% of a project's new lines are revised at 2.3 times the base rate. Each pitfall leaves a symptom; we give a checklist.

## How I Learned to Use Copilot Sessions 400 Million Tokens Later

DevFeed: [How I Learned to Use Copilot Sessions 400 Million Tokens Later](<https://devfeed.tech/articles/how-i-learned-to-use-copilot-sessions-400-million-tokens-later-126487.md>)

Original publisher: [Read original article](<https://code.dblock.org/2026/11/10/how-i-learned-to-use-copilot-sessions-400-million-tokens-later.html>)

Author: Daniel Doubrovkine (dblock@dblock.org)

Published: 2026-11-10T00:00:00Z

Content type: tutorial

Language: en

Sources: [Daniel Doubrovkine](<https://devfeed.tech/sources/daniel-doubrovkine.md>)

Topics: [GitHub Copilot CLI](<https://devfeed.tech/topics/github-copilot-cli.md>), [AI Context Management](<https://devfeed.tech/topics/ai-context-management.md>), [experience report](<https://devfeed.tech/topics/experience-report.md>), [Tool](<https://devfeed.tech/topics/tool.md>)

Tags: [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [context](<https://devfeed.tech/tags/context.md>), [context-switch](<https://devfeed.tech/tags/context-switch.md>), [copilot](<https://devfeed.tech/tags/copilot.md>), [development](<https://devfeed.tech/tags/development.md>), [github-copilot-cli](<https://devfeed.tech/tags/github-copilot-cli.md>), [linter](<https://devfeed.tech/tags/linter.md>), [local](<https://devfeed.tech/tags/local.md>), [logs](<https://devfeed.tech/tags/logs.md>), [markdown](<https://devfeed.tech/tags/markdown.md>), [model](<https://devfeed.tech/tags/model.md>), [open-source](<https://devfeed.tech/tags/open-source.md>), [restore](<https://devfeed.tech/tags/restore.md>), [sessions](<https://devfeed.tech/tags/sessions.md>), [source-code](<https://devfeed.tech/tags/source-code.md>), [switch](<https://devfeed.tech/tags/switch.md>), [tokens](<https://devfeed.tech/tags/tokens.md>), [trees](<https://devfeed.tech/tags/trees.md>), [unrelated](<https://devfeed.tech/tags/unrelated.md>), [using](<https://devfeed.tech/tags/using.md>), [waiting](<https://devfeed.tech/tags/waiting.md>)

### AI overview

Using one Copilot CLI conversation across unrelated work led to about 4,322 model calls in three and a half days, with 11.4 million non-cached input, 2.4 million output, and 411 million cached input tokens. Cached context may be counted repeatedly, so that figure does not represent tokens written or generated and does not directly indicate cost. To reduce unrelated context and the time spent restoring and saving it, the author organizes project checkouts and task worktrees under `~/source`, starts Copilot there, then uses `/resume` to open the session picker, `n` to create sessions, `/rename` to name them after repositories, and `/cwd` to set each session's working directory. Naming a session does not change its directory.

### Source excerpt

I recently built a Markdown linter with GitHub Copilot CLI. In between, I reviewed other projects, edited this blog, and generally treated one conversation as a terminal into which I could dump whatever I was doing next. I really loved the fact that Copilot would context switch, do some work, then go back to what it was doing. It's so smart! Well, that may have burnt some extra tokens. Counting Tokens The local usage logs for the roughly three-and-a-half-day development window showed approximately 4,322 model calls, including delegated agents: Token Type Millions Non-cached input 11.4 Output 2.4 Cached input 411 That's roughly 14 million non-cached input and output tokens, plus more than 400 million cached input tokens. Cached input is previously processed context reused across requests. The same context can be counted repeatedly, so 411 million doesn't mean I wrote 411 million tokens or that Copilot generated that much code. It also doesn't translate directly into dollar spend: cached input has different accounting from non-cached input and output. Still, I had given it a lot of unrelated history to carry around and wasted a lot of time waiting for it to restore and save context. Start a Session, Give It a Name Since then, I've been using Copilot CLI's named sessions more. Here's the workflow that is working for me. My source code lives in ~/source. I group related projects together, keeping my forks and upstream checkouts in the same folder. Git working trees fit into the same layout, named after the project and task. For example, here's how I'd add task-specific working trees alongside my fork of ESLint's Markdown plugin and its upstream checkout: ~/source/ `-- eslint/ `-- markdown/ |-- upstream-markdown/ # upstream checkout |-- dblock-markdown/ # my fork |-- dblock-markdown-parser-fix/ # working tree `-- dblock-markdown-docs/ # working tree I start a single copilot process in ~/source and switch between sessions inside it. cd ~/source copilot --yolo --resume Ins

## We organized Sustainability Month. Here's what we learned. 🌱

DevFeed: [We organized Sustainability Month. Here's what we learned. 🌱](<https://devfeed.tech/articles/we-organized-sustainability-month-here-s-what-we-learned-122217.md>)

Original publisher: [Read original article](<https://www.cncf.io/blog/2026/11/07/we-organized-sustainability-month-heres-what-we-learned-%f0%9f%8c%b1/>)

Author: Elif Samedin - CNCF Ambassador | Diana Todea - CNCF Ambassador, Kepler maintainer, TAG Operational Resilience Lead

Published: 2026-11-07T12:30:00Z

Content type: article

Language: en

Sources: [Cloud Native Computing Foundation](<https://devfeed.tech/sources/cloud-native-computing-foundation.md>)

Topics: [Sustainable Computing](<https://devfeed.tech/topics/sustainable-computing.md>), [Green Software](<https://devfeed.tech/topics/green-software.md>), [experience report](<https://devfeed.tech/topics/experience-report.md>), [eBPF](<https://devfeed.tech/topics/ebpf.md>), [observability](<https://devfeed.tech/topics/observability.md>), [OpenTelemetry](<https://devfeed.tech/topics/opentelemetry.md>), [Hackathon-Organiser](<https://devfeed.tech/topics/hackathon-organiser.md>)

Tags: [blog](<https://devfeed.tech/tags/blog.md>), [carbon](<https://devfeed.tech/tags/carbon.md>), [cloud-native-ecosystem](<https://devfeed.tech/tags/cloud-native-ecosystem.md>), [cncf](<https://devfeed.tech/tags/cncf.md>), [community](<https://devfeed.tech/tags/community.md>), [community-events](<https://devfeed.tech/tags/community-events.md>), [computing](<https://devfeed.tech/tags/computing.md>), [cpu](<https://devfeed.tech/tags/cpu.md>), [data-center](<https://devfeed.tech/tags/data-center.md>), [data-center-efficiency](<https://devfeed.tech/tags/data-center-efficiency.md>), [data-centers](<https://devfeed.tech/tags/data-centers.md>), [ebpf](<https://devfeed.tech/tags/ebpf.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [emissions](<https://devfeed.tech/tags/emissions.md>), [energy](<https://devfeed.tech/tags/energy.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [environmental-sustainability](<https://devfeed.tech/tags/environmental-sustainability.md>), [gpus](<https://devfeed.tech/tags/gpus.md>), [kubernetes](<https://devfeed.tech/tags/kubernetes.md>), [observability](<https://devfeed.tech/tags/observability.md>), [sustainability](<https://devfeed.tech/tags/sustainability.md>)

### AI overview

Cloud Native Sustainability Month showed how environmental sustainability connects to everyday engineering choices, including Kubernetes resource allocation, CPU use, GPU efficiency, carbon observability with OpenTelemetry, eBPF, energy metrics and data center efficiency. The organizers' main takeaway is that engineers need not be sustainability experts to participate: they can question workload resource use, explore projects such as the CNCF incubating project Kepler, and share experiments. Open source contributors can help through code, documentation, testing or community discussion; organizers can start with a lightning talk or meetup discussion. The article presents community events as a way to learn and exchange ideas, while acknowledging that the organizers do not have all the answers.

### Source excerpt

What do Kubernetes resource allocation, eBPF, carbon observability, and GPUs have in common? Quite a lot, actually. Especially when you start looking at them through the lens of environmental sustainability. This October, we brought these conversations...

## Asahi Linux 7.3 progress report

DevFeed: [Asahi Linux 7.3 progress report](<https://devfeed.tech/articles/asahi-linux-7-3-progress-report-122000.md>)

Original publisher: [Read original article](<https://lwn.net/Articles/1105936/>)

Author: corbet

Published: 2026-11-07T09:49:38Z

Content type: news

Language: en

Sources: [LWN.net](<https://devfeed.tech/sources/lwn-net.md>)

Topics: [GNU General Public License](<https://devfeed.tech/topics/gpl.md>), [experience report](<https://devfeed.tech/topics/experience-report.md>)

Tags: [apple-silicon](<https://devfeed.tech/tags/apple-silicon.md>), [code](<https://devfeed.tech/tags/code.md>), [exploit](<https://devfeed.tech/tags/exploit.md>), [forks](<https://devfeed.tech/tags/forks.md>), [gnu](<https://devfeed.tech/tags/gnu.md>), [hijack](<https://devfeed.tech/tags/hijack.md>), [license](<https://devfeed.tech/tags/license.md>), [licensing](<https://devfeed.tech/tags/licensing.md>), [linux](<https://devfeed.tech/tags/linux.md>), [llm](<https://devfeed.tech/tags/llm.md>), [mit](<https://devfeed.tech/tags/mit.md>), [progress](<https://devfeed.tech/tags/progress.md>), [progress-report](<https://devfeed.tech/tags/progress-report.md>), [report](<https://devfeed.tech/tags/report.md>), [repositories](<https://devfeed.tech/tags/repositories.md>)

### AI overview

Asahi Linux developers say attempts to take over and monetize their Apple Silicon Linux work included publishing in-development code from personal forks without attribution. Because MIT licensing allowed this and offered no protection against such use, they have begun moving all repositories to GNU GPL version 2 or later, stating that future Asahi Linux code will use strong copyleft licenses. The excerpt does not describe Linux 7.3 technical progress.

### Source excerpt

The Asahi Linux developers have put out a new report on their progress toward a Linux distribution for Apple devices. Some time ago, we were made aware of multiple attempts to exploit our permissive licensing to hijack our work, with a direct view to replacing the project and aggressively monetising Linux on Apple Silicon. In some extreme cases, in-development feature code was being lifted straight out of our personal forks and published without attribution or lazily rewritten with LLM assistance to obfuscate its origin. The MIT licence explicitly allows this, and has no carveouts to protect against bad faith actors. As such, we made the difficult decision to begin the process of transitioning all of our repositories to the GNU General Public License version 2 or later. Going forward all Asahi Linux code will be licenced under strong copyleft licences only. Those who do not learn from history are doomed to repeat it.

## Godot Showcase - Sovereign Tower

DevFeed: [Godot Showcase - Sovereign Tower](<https://devfeed.tech/articles/godot-showcase-sovereign-tower-122133.md>)

Original publisher: [Read original article](<https://godotengine.org/article/godot-showcase-sovereign-tower/>)

Author: Hugo Locurcio

Published: 2026-11-06T14:00:00Z

Content type: article

Language: en

Sources: [Godot Engine Official](<https://devfeed.tech/sources/godot-engine-official.md>)

Topics: [Godot](<https://devfeed.tech/topics/godot.md>), [Game Development](<https://devfeed.tech/topics/game-development.md>), [experience report](<https://devfeed.tech/topics/experience-report.md>), [Game Jam](<https://devfeed.tech/topics/game-jam.md>)

Tags: [advice](<https://devfeed.tech/tags/advice.md>), [animate](<https://devfeed.tech/tags/animate.md>), [animation](<https://devfeed.tech/tags/animation.md>), [art](<https://devfeed.tech/tags/art.md>), [asset](<https://devfeed.tech/tags/asset.md>), [asset-pipeline](<https://devfeed.tech/tags/asset-pipeline.md>), [challenges](<https://devfeed.tech/tags/challenges.md>), [change](<https://devfeed.tech/tags/change.md>), [community](<https://devfeed.tech/tags/community.md>), [design](<https://devfeed.tech/tags/design.md>), [dev](<https://devfeed.tech/tags/dev.md>), [develop](<https://devfeed.tech/tags/develop.md>), [developer](<https://devfeed.tech/tags/developer.md>), [developing](<https://devfeed.tech/tags/developing.md>), [development](<https://devfeed.tech/tags/development.md>), [experience](<https://devfeed.tech/tags/experience.md>), [games](<https://devfeed.tech/tags/games.md>), [godot](<https://devfeed.tech/tags/godot.md>), [showcase](<https://devfeed.tech/tags/showcase.md>), [tower](<https://devfeed.tech/tags/tower.md>)

### AI overview

Sovereign Tower is a narrative round-table management RPG in which players send knights on quests and can rewind time while retaining the Sovereign's knowledge. That mechanic required two saves per day and careful handling of hundreds of narrative variables, quests, and more than a hundred characters. The team split saved data into precise components and tested extensively; some bugs still reached release and are being fixed. To control localization costs and story scope, the writers cut some narrative variations while aiming to preserve the player experience. Godot's scene system let the art director integrate and adjust assets directly; the team used AnimationPlayers and Tweens, and Godot 4.7 transform offsets made UI transitions easier to animate. The developers describe the workflow as intuitive for their project. The game is available on Steam for Windows and Linux.

### Source excerpt

Paul and Lisa from Wild Wits Games tell us about their experience working on Sovereign Tower.

## From open beta to global GA: What Docusign learned by bringing contract AI workflows to agents

DevFeed: [From open beta to global GA: What Docusign learned by bringing contract AI workflows to agents](<https://devfeed.tech/articles/from-open-beta-to-global-ga-what-docusign-learned-by-bringing-contract-ai-workflows-to-agents-119282.md>)

Original publisher: [Read original article](<https://thenewstack.io/docusign-mcp-server-agents/>)

Author: Carly Page

Published: 2026-11-05T14:50:21Z

Content type: news

Language: en

Sources: [The New Stack](<https://devfeed.tech/sources/the-new-stack.md>)

Topics: [AI Integration Strategies](<https://devfeed.tech/topics/ai-integration-strategies.md>), [agentic workflows](<https://devfeed.tech/topics/agentic-workflows.md>), [human in the loop agent](<https://devfeed.tech/topics/human-in-the-loop-agent.md>), [experience report](<https://devfeed.tech/topics/experience-report.md>)

Tags: [actions](<https://devfeed.tech/tags/actions.md>), [administration](<https://devfeed.tech/tags/administration.md>), [adoption](<https://devfeed.tech/tags/adoption.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agentforce](<https://devfeed.tech/tags/agentforce.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agent](<https://devfeed.tech/tags/ai-agent.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-application](<https://devfeed.tech/tags/ai-application.md>), [ai-workflows](<https://devfeed.tech/tags/ai-workflows.md>), [beta](<https://devfeed.tech/tags/beta.md>), [contract-ai](<https://devfeed.tech/tags/contract-ai.md>), [docusign](<https://devfeed.tech/tags/docusign.md>), [enterprise](<https://devfeed.tech/tags/enterprise.md>), [enterprise-software](<https://devfeed.tech/tags/enterprise-software.md>), [global](<https://devfeed.tech/tags/global.md>), [information](<https://devfeed.tech/tags/information.md>), [model-context-protocol-mcp](<https://devfeed.tech/tags/model-context-protocol-mcp.md>), [open](<https://devfeed.tech/tags/open.md>), [post](<https://devfeed.tech/tags/post.md>), [security](<https://devfeed.tech/tags/security.md>), [sponsor-docusign](<https://devfeed.tech/tags/sponsor-docusign.md>), [sponsored](<https://devfeed.tech/tags/sponsored.md>), [sponsored-post](<https://devfeed.tech/tags/sponsored-post.md>), [workflows](<https://devfeed.tech/tags/workflows.md>)

### AI overview

Docusign's path from MCP open beta in May 2026 to global general availability on September 30 showed that connecting agents to contract tools is easier than making workflows reliable, permission-bound, and clear when requests are ambiguous. Customer feedback expanded agreement workflows, local document uploads in supported MCP Apps clients, and account-level controls. Docusign uses OAuth 2.0 scopes and existing user permissions for tool calls, with account administrators able to enable or disable MCP access; these controls do not determine how third-party AI providers handle received data. A shared server supports multiple AI clients, though client behavior and feature availability can differ. Guided clarification can help agents resolve missing choices before acting. Docusign says MCP complements rather than replaces APIs: APIs suit deterministic processes with defined inputs and outputs, while agents can use MCP to choose tools for broader requests, and both can be combined. Human review and approval remain part of transactions.

### Source excerpt

Connecting an AI agent to enterprise software is relatively easy. Making sure it can find the right information, carry out The post From open beta to global GA: What Docusign learned by bringing contract AI workflows to agents appeared first on The New Stack.

## Lessons learned: Common engineering pitfalls to avoid in the age of AI

DevFeed: [Lessons learned: Common engineering pitfalls to avoid in the age of AI](<https://devfeed.tech/articles/lessons-learned-common-engineering-pitfalls-to-avoid-in-the-age-of-ai-118767.md>)

Original publisher: [Read original article](<https://www.cio.com/article/4244660/lessons-learned-common-engineering-pitfalls-to-avoid-in-the-age-of-ai.html>)

Author: Chris Barnes

Published: 2026-11-05T10:00:00Z

Content type: opinion

Language: en

Sources: [CIO](<https://devfeed.tech/sources/cio.md>)

Topics: [experience report](<https://devfeed.tech/topics/experience-report.md>), [it modernization](<https://devfeed.tech/topics/it-modernization.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Business, Industry, Usability, User Experience, User Research, Web Strategy](<https://devfeed.tech/topics/business-industry-usability-user-experience-user-research-web-strategy.md>)

Tags: [2026](<https://devfeed.tech/tags/2026.md>), [acquisitions](<https://devfeed.tech/tags/acquisitions.md>), [adoption](<https://devfeed.tech/tags/adoption.md>), [age-of-ai](<https://devfeed.tech/tags/age-of-ai.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-coding-agents](<https://devfeed.tech/tags/ai-coding-agents.md>), [code](<https://devfeed.tech/tags/code.md>), [complexity](<https://devfeed.tech/tags/complexity.md>), [contributor](<https://devfeed.tech/tags/contributor.md>), [edition-us](<https://devfeed.tech/tags/edition-us.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [enterprise-architecture](<https://devfeed.tech/tags/enterprise-architecture.md>), [enterprise-architecture-it-leadership-it-management-staff-management](<https://devfeed.tech/tags/enterprise-architecture-it-leadership-it-management-staff-management.md>), [it-leadership](<https://devfeed.tech/tags/it-leadership.md>), [it-management](<https://devfeed.tech/tags/it-management.md>), [language-en](<https://devfeed.tech/tags/language-en.md>), [linkedin](<https://devfeed.tech/tags/linkedin.md>), [platform](<https://devfeed.tech/tags/platform.md>), [products](<https://devfeed.tech/tags/products.md>), [quality](<https://devfeed.tech/tags/quality.md>), [staff-management](<https://devfeed.tech/tags/staff-management.md>), [storytype-opinion](<https://devfeed.tech/tags/storytype-opinion.md>), [team](<https://devfeed.tech/tags/team.md>), [technology](<https://devfeed.tech/tags/technology.md>), [tools](<https://devfeed.tech/tags/tools.md>), [upskilling](<https://devfeed.tech/tags/upskilling.md>)

### AI overview

Veson's ongoing consolidation of Shipfix, VesselsValue and IMOS shows that AI can speed integration work, but also raises risks when teams are still learning the tools and business continuity cannot pause. The author argues that a connected platform needs unified data beneath one experience so AI can reason across products; superficial product stitching is not enough. AI reduced some integration plumbing from months to weeks, while Plandek tracking of speed, quality, predictability and focus is used to spot tradeoffs before they compound. Lessons so far include preserving working systems, embedding AI in trusted workflows, treating continuity as a feature, using one source of truth, spreading adoption peer to peer, and protecting quality gates as engineering shifts toward directing and verifying AI-written code. The author says the work remains in progress.

### Source excerpt

A LinkedIn post about my old company, insightsoftware, popped up the other day. It was lamenting the challenge of turning 29 acquisitions and close to 80 products into one comprehensive platform. I spent years in the middle of that work, and what I know now is how much harder that consolidation is than you might think, and how much of it you only learn by getting it wrong first. So, when I arrived at Veson in November and found a similar landscape, I felt a bit of déjà vu. A bit of "I've done this before, so I know what I'm doing" and a bit of "I've done this before and I know it's no cakewalk." Veson had spent the past several years acquiring a range of products like Shipfix and VesselsValue and now needed to consolidate them onto a platform alongside its flagship IMOS solution. A new variable is also now at play that changes the whole equation: the advancement of AI. While AI may sound like a cheat code for this work, it's very much a double-edged sword... with a grenade taped to it. Yes, we can do some things faster and differently than we ever could a few years ago. But we're also learning as we go and upskilling our team as we go, on a technology that's changing under our feet. The newness of these tools multiplies the complexity in a way I didn't fully appreciate at insightsoftware. Back then, the tools simply didn't exist or were just emerging. Now your platform must transform so that it thrives in the age of AI. It has to be built in the age of AI, which -- depending on whose study you believe -- means anywhere from 2X to 10X faster. And then your team needs to adapt to the age of AI, which again means faster, at higher quality and producing more output. That's a lot of spinning plates, on top of each other, none of which you're allowed to drop. Oh, and one other plate: this is a centuries-old industry that moves billions of dollars in freight a year and does not stop for anyone. You don't get to tie the ship up at the dock while you rebuild the engine. Business

## What we learned as Webflow Campaigns' first customer

DevFeed: [What we learned as Webflow Campaigns' first customer](<https://devfeed.tech/articles/what-we-learned-as-webflow-campaigns-first-customer-121078.md>)

Original publisher: [Read original article](<https://www.webflow.com/blog/how-webflow-uses-campaigns-mcp>)

Author: Webflow Team

Published: 2026-11-05T00:00:00Z

Content type: article

Language: en

Sources: [Webflow Blog](<https://devfeed.tech/sources/webflow-blog.md>)

Topics: [webflow](<https://devfeed.tech/topics/webflow.md>), [experience report](<https://devfeed.tech/topics/experience-report.md>), [human review](<https://devfeed.tech/topics/human-review.md>), [Tool](<https://devfeed.tech/topics/tool.md>)

Tags: [2](<https://devfeed.tech/tags/2.md>), [3](<https://devfeed.tech/tags/3.md>), [a-b-testing](<https://devfeed.tech/tags/a-b-testing.md>), [abm](<https://devfeed.tech/tags/abm.md>), [about](<https://devfeed.tech/tags/about.md>), [account](<https://devfeed.tech/tags/account.md>), [ad-performance](<https://devfeed.tech/tags/ad-performance.md>), [added](<https://devfeed.tech/tags/added.md>), [ads](<https://devfeed.tech/tags/ads.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [airops](<https://devfeed.tech/tags/airops.md>), [all](<https://devfeed.tech/tags/all.md>), [customer](<https://devfeed.tech/tags/customer.md>), [events](<https://devfeed.tech/tags/events.md>), [first](<https://devfeed.tech/tags/first.md>), [inside-webflow](<https://devfeed.tech/tags/inside-webflow.md>), [more](<https://devfeed.tech/tags/more.md>), [search](<https://devfeed.tech/tags/search.md>), [webflow](<https://devfeed.tech/tags/webflow.md>)

### AI overview

Webflow's first Campaigns users report that tailored landing pages helped them reduce page-building handoffs while keeping review and publishing control. For paid search, one template produced seven ad-group variations with distinct URLs and content; adding a page took about 30 minutes. A weekly report uses ad and warehouse data to suggest copy changes, which require approval before staging. For an event, Campaigns used Webflow's live site to establish a design system, and a page drafted from a brief generated 80 warm leads within minutes after a QR code was shown. For ABM, an n8n workflow gathers Salesforce, Gong, opportunity, product-use, and account data; Claude and Campaigns generate account-specific content, while selected sections vary within an approved design. Sales and marketing operations review drafts before publishing. The internal pilot began with 100 accounts. The team describes these as early experiments and says it is still learning how the pages perform.

### Source excerpt

Across paid search, events, and ABM, fewer handoffs are giving us more control over what we launch and when.

## The CFO Who Learned to Code by Shipping: How I Use Hackathons to Keep Up With AI Tools

DevFeed: [The CFO Who Learned to Code by Shipping: How I Use Hackathons to Keep Up With AI Tools](<https://devfeed.tech/articles/the-cfo-who-learned-to-code-by-shipping-how-i-use-hackathons-to-keep-up-with-ai-tools-115707.md>)

Original publisher: [Read original article](<https://hackernoon.com/the-cfo-who-learned-to-code-by-shipping-how-i-use-hackathons-to-keep-up-with-ai-tools?source=rss>)

Author: Cubiczan

Published: 2026-11-03T20:15:00Z

Content type: article

Language: en

Sources: [HackerNoon](<https://devfeed.tech/sources/hackernoon.md>)

Topics: [AI-assisted coding](<https://devfeed.tech/topics/ai-assisted-coding.md>), [Finance](<https://devfeed.tech/topics/finance.md>), [Learning](<https://devfeed.tech/topics/learning.md>), [experience report](<https://devfeed.tech/topics/experience-report.md>), [AI Chat](<https://devfeed.tech/topics/ai-chat.md>), [GitHub](<https://devfeed.tech/topics/github.md>)

Tags: [2026](<https://devfeed.tech/tags/2026.md>), [academy](<https://devfeed.tech/tags/academy.md>), [advice](<https://devfeed.tech/tags/advice.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agentic-ai](<https://devfeed.tech/tags/agentic-ai.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-tools](<https://devfeed.tech/tags/ai-tools.md>), [builder](<https://devfeed.tech/tags/builder.md>), [careers](<https://devfeed.tech/tags/careers.md>), [cfo](<https://devfeed.tech/tags/cfo.md>), [code](<https://devfeed.tech/tags/code.md>), [demo](<https://devfeed.tech/tags/demo.md>), [finance](<https://devfeed.tech/tags/finance.md>), [general-interview](<https://devfeed.tech/tags/general-interview.md>), [hackathons](<https://devfeed.tech/tags/hackathons.md>), [learn](<https://devfeed.tech/tags/learn.md>), [learning](<https://devfeed.tech/tags/learning.md>), [shipping](<https://devfeed.tech/tags/shipping.md>), [stack](<https://devfeed.tech/tags/stack.md>), [tech](<https://devfeed.tech/tags/tech.md>), [tech-interviews](<https://devfeed.tech/tags/tech-interviews.md>), [tech-stack](<https://devfeed.tech/tags/tech-stack.md>), [tools](<https://devfeed.tech/tags/tools.md>), [upskilling](<https://devfeed.tech/tags/upskilling.md>)

### AI overview

After more than 20 years in finance, Shyam Desigan says AI tools and hackathon deadlines helped him learn software by building and shipping projects. Since February 2026, he has submitted more than 35 projects; examples include ChainSight AI, a live Mantle DEX monitor that turns anomalous activity flags into plain-English risk notes and won Mantle's Project Deployment Award, and Council Tower Pipeline, which won a DeveloperWeek New York prize. Other entries did not place, but exposed him to different clouds, models and infrastructure. He treats AI output as a first draft: people still need to review code, test edge cases and own the result. He recommends that non-engineers start with a real problem they understand and learn the stack one layer at a time.

### Source excerpt

A finance executive turned builder on using AI tools and 35+ hackathons as a forced curriculum to learn the tech stack.

## What Rami Ismail learned with his return to game development

DevFeed: [What Rami Ismail learned with his return to game development](<https://devfeed.tech/articles/what-rami-ismail-learned-with-his-return-to-game-development-115615.md>)

Original publisher: [Read original article](<https://www.gamedeveloper.com/design/what-rami-ismail-learned-with-his-return-to-game-development>)

Author: Bryant Francis

Published: 2026-11-03T17:31:00Z

Content type: news

Language: en

Sources: [Game Developer](<https://devfeed.tech/sources/gamedeveloper.md>)

Topics: [experience report](<https://devfeed.tech/topics/experience-report.md>), [Game engine](<https://devfeed.tech/topics/game-engine.md>)

Tags: [action-games](<https://devfeed.tech/tags/action-games.md>), [adapting](<https://devfeed.tech/tags/adapting.md>), [balance](<https://devfeed.tech/tags/balance.md>), [best-practices](<https://devfeed.tech/tags/best-practices.md>), [design](<https://devfeed.tech/tags/design.md>), [design-thinking](<https://devfeed.tech/tags/design-thinking.md>), [developer](<https://devfeed.tech/tags/developer.md>), [developers](<https://devfeed.tech/tags/developers.md>), [effects](<https://devfeed.tech/tags/effects.md>), [freelance](<https://devfeed.tech/tags/freelance.md>), [fun](<https://devfeed.tech/tags/fun.md>), [funding](<https://devfeed.tech/tags/funding.md>), [game-cloning](<https://devfeed.tech/tags/game-cloning.md>), [game-development](<https://devfeed.tech/tags/game-development.md>), [gameplay](<https://devfeed.tech/tags/gameplay.md>), [indie](<https://devfeed.tech/tags/indie.md>)

### AI overview

Rami Ismail returned to game development with Australia Did It, a turn-based strategy game built by a temporary team assembled with Aesthetician Labs and backed by publisher Mystic Forge. He says the industry has become more risk-averse and that no current development model offers stability; for this unusual project, a short-term team made a risky creative direction possible. The team had to establish its communication and process as they went, with producer Sam Magnolia taking production responsibilities so Ismail could focus on creative direction. Switching from fast action games to strategy also changed how he evaluated fun: rather than quick prototypes, the game needed enough content and balance to reveal whether its systems worked, a process that took four to five months to click. Its unusual unit-merging system includes a progression checker intended to help players stay on track and avoid being locked into a losing strategy early. Ismail remains proud of the work but nervous about player reactions.

### Source excerpt

Indie dev advocate Rami Ismail has returned to making games after six years away from active development.

## Getting a Grip: Operating Human-Agent Software Teams Through Governed Messages

DevFeed: [Getting a Grip: Operating Human-Agent Software Teams Through Governed Messages](<https://devfeed.tech/articles/getting-a-grip-operating-human-agent-software-teams-through-governed-messages-113493.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.00689>)

Author: Michael Granitzer, Kanishka Ghosh Dastidar, Andreas Einwiller, Jelena Mitrovi\\'c, Saber Zerhoudi

Published: 2026-11-03T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Software Engineering](<https://devfeed.tech/sources/arxiv-software-engineering.md>)

Topics: [engineering-leadership](<https://devfeed.tech/topics/engineering-leadership.md>), [Multi Agent Systems](<https://devfeed.tech/topics/multi-agent-systems.md>), [agent to agent communication](<https://devfeed.tech/topics/agent-to-agent-communication.md>), [Messaging](<https://devfeed.tech/topics/messaging.md>), [decision-making](<https://devfeed.tech/topics/decision-making.md>), [experience report](<https://devfeed.tech/topics/experience-report.md>), [human review](<https://devfeed.tech/topics/human-review.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [agent-software](<https://devfeed.tech/tags/agent-software.md>), [agents](<https://devfeed.tech/tags/agents.md>), [analysis](<https://devfeed.tech/tags/analysis.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [attention](<https://devfeed.tech/tags/attention.md>), [backdoor](<https://devfeed.tech/tags/backdoor.md>), [coding](<https://devfeed.tech/tags/coding.md>), [coding-agents](<https://devfeed.tech/tags/coding-agents.md>), [context](<https://devfeed.tech/tags/context.md>), [cost](<https://devfeed.tech/tags/cost.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-se](<https://devfeed.tech/tags/cs-se.md>), [decision-making](<https://devfeed.tech/tags/decision-making.md>), [decision-rules](<https://devfeed.tech/tags/decision-rules.md>), [defense](<https://devfeed.tech/tags/defense.md>), [distributed](<https://devfeed.tech/tags/distributed.md>), [distributed-system](<https://devfeed.tech/tags/distributed-system.md>), [failure](<https://devfeed.tech/tags/failure.md>), [messages](<https://devfeed.tech/tags/messages.md>), [software](<https://devfeed.tech/tags/software.md>), [teams](<https://devfeed.tech/tags/teams.md>)

### AI overview

The paper proposes yoars, a relay that governs human-agent software teams through authenticated identities, typed and addressed messages, room-level rules, and an attributable append-only record. Across 899 analyzable controlled runs, changing which messages trigger work and limiting repeated messages coincided with coordination failures falling from 42% to 13%; the adjusted result across nine tests was borderline (p = .057). Role instructions and votes did not reliably enforce process, and conversational defenses did not reliably stop backdoors. Record auditing identified the persuader in all 25 confirmed cases where an attacker recruited a teammate, but also flagged 24 of 40 honest runs. In a four-week deployment involving up to 19 agents and two people, requests mainly waited on the human coordinator and approvals, while agents responded within minutes. The authors identify human attention and context retention as constraints, report instrument failures, and release study materials.

### Source excerpt

arXiv:2611.00689v1 Announce Type: new Abstract: Software is increasingly developed by teams in which several people direct many coding agents from different vendors. Such teams must establish who may request what from whom, which messages start work, which decisions bind, and who caused a change. We propose to govern a human-agent team through its message protocol and present yoars, a relay in which humans and agents are authenticated principals that exchange typed, addressed messages in rooms holding goals, decision rules, routing and an attributable, append-only record. This report gives the architecture, the full controlled study and the field material. In 926 controlled runs (899 analyzable), small agent teams built programs under varied room rules, some with a hostile member planting a backdoor. Revising which messages start work, bundled with limits on repeated messages, was followed by a drop in total coordination failure from 42% to 13% (p = .0063, adjusted p = .057 over nine registered tests). Instructed roles and votes did not reliably produce the intended process, and no defense inside the conversation reliably prevented the backdoor. An auditor read the records of 96 runs with a confirmed backdoor. In the 25 runs where the attacker had persuaded a teammate to write it, the auditor named the persuader in all 25, but it also flagged 24 of 40 honest runs. In a four-week field study, up to nineteen agents and two people developed a distributed system through yoars, producing 7,457 acts. Human attention and decision making emerged as the critical resource: requests waited mainly for the human-run coordinator and for approvals, while agents answered within minutes. Most governance requirements emerged from operation, and holding context dominated token cost. We report twenty-six failures of our own instruments, derive seven lessons and a checklist, and release the corpus, the redacted room ledger and the analysis scripts.

## Stripe on Stripe: What we learned moving $1B in vendor payments to Global Payouts

DevFeed: [Stripe on Stripe: What we learned moving $1B in vendor payments to Global Payouts](<https://devfeed.tech/articles/stripe-on-stripe-what-we-learned-moving-1b-in-vendor-payments-to-global-payouts-109683.md>)

Original publisher: [Read original article](<https://stripe.com/blog/stripe-on-stripe-what-we-learned-moving-1b-in-vendor-payments-to-global-payouts>)

Author: Yash Patni; Hammad Akhtar

Published: 2026-10-30T00:00:00Z

Content type: article

Language: en

Sources: [Stripe](<https://devfeed.tech/sources/stripe-blog.md>)

Topics: [migration](<https://devfeed.tech/topics/migration.md>), [Global Payments](<https://devfeed.tech/topics/global-payments.md>), [experience report](<https://devfeed.tech/topics/experience-report.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>)

Tags: [000](<https://devfeed.tech/tags/000.md>), [2](<https://devfeed.tech/tags/2.md>), [2026](<https://devfeed.tech/tags/2026.md>), [3](<https://devfeed.tech/tags/3.md>), [about](<https://devfeed.tech/tags/about.md>), [account](<https://devfeed.tech/tags/account.md>), [ach](<https://devfeed.tech/tags/ach.md>), [actions](<https://devfeed.tech/tags/actions.md>), [added](<https://devfeed.tech/tags/added.md>), [adopting](<https://devfeed.tech/tags/adopting.md>), [global](<https://devfeed.tech/tags/global.md>), [here-s](<https://devfeed.tech/tags/here-s.md>), [heres](<https://devfeed.tech/tags/heres.md>), [moving](<https://devfeed.tech/tags/moving.md>), [one](<https://devfeed.tech/tags/one.md>), [payments](<https://devfeed.tech/tags/payments.md>), [product](<https://devfeed.tech/tags/product.md>), [setup](<https://devfeed.tech/tags/setup.md>), [stripe](<https://devfeed.tech/tags/stripe.md>), [with](<https://devfeed.tech/tags/with.md>)

### AI overview

Stripe moved vendor payments from separate bank integrations and manual portal entry to Global Payouts, reporting over $1 billion paid to 3,400+ US and UK vendors by October 2026. New-market setup fell from as many as six weeks to one, reconciliation and reporting save an estimated 20-30 hours monthly, and payouts have a 99.9% success rate. The staged migration began with low-value US payments and expanded in four waves, increasing amounts as prior payouts succeeded. The rollout revealed that recipient-data errors caused most failures, so API validation now catches problems before payment; reference numbers and invoice metadata improve sender-side tracking, while customizable statement descriptors help recipients identify deposits. Raising ACH and wire limits to $100 million brought more large payments into automated workflows while retaining approval and reconciliation processes.

### Source excerpt

In early 2026, we began moving our vendor payments to Global Payouts, Stripe's product for paying recipients in 160+ countries. New-market setup fell from six weeks to one, with a 99.9% payout success rate. Here's what we learned.

[Next page](<https://devfeed.tech/topics/experience-report.md?cursor=WyIyMDI2LTEwLTMwVDAwOjAwOjAwKzAwOjAwIiwgIjA0NTU2ODdlLTY3MjQtNGE5OC05YThlLWQxOTkyZDI2NTIzNiJd>)