# GitHub Next

A team of researchers and engineers at GitHub prototyping the tools, workflows, and ideas that will shape the future of software development.

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

## Agent skill: dialing in the work, four knobs at a time

DevFeed: [Agent skill: dialing in the work, four knobs at a time](<https://devfeed.tech/articles/agent-skill-dialing-in-the-work-four-knobs-at-a-time-130607.md>)

Original publisher: [Read original article](<https://githubnext.com/posts/four-knobs/>)

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

Content type: opinion

Language: en

Sources: [GitHub Next](<https://devfeed.tech/sources/github-next.md>)

Topics: [Agent Skills](<https://devfeed.tech/topics/agent-skills.md>), [Developer experience](<https://devfeed.tech/topics/developer-experience.md>)

Tags: [agent-skill](<https://devfeed.tech/tags/agent-skill.md>), [animation](<https://devfeed.tech/tags/animation.md>), [choice](<https://devfeed.tech/tags/choice.md>), [click](<https://devfeed.tech/tags/click.md>), [coding-agents](<https://devfeed.tech/tags/coding-agents.md>), [color](<https://devfeed.tech/tags/color.md>), [compare](<https://devfeed.tech/tags/compare.md>), [control](<https://devfeed.tech/tags/control.md>), [controller](<https://devfeed.tech/tags/controller.md>), [creative](<https://devfeed.tech/tags/creative.md>), [custom](<https://devfeed.tech/tags/custom.md>), [design-tokens](<https://devfeed.tech/tags/design-tokens.md>), [details](<https://devfeed.tech/tags/details.md>), [developers](<https://devfeed.tech/tags/developers.md>), [docs](<https://devfeed.tech/tags/docs.md>), [easing](<https://devfeed.tech/tags/easing.md>), [echo](<https://devfeed.tech/tags/echo.md>), [effect](<https://devfeed.tech/tags/effect.md>), [emotion](<https://devfeed.tech/tags/emotion.md>), [feedback](<https://devfeed.tech/tags/feedback.md>)

### AI overview

Quadknob is an agent skill that helps people shape agent-built interfaces through a small, tactile set of controls instead of facing many design choices at once. It presents four or fewer values per page, can show live previews, and lets users choose the values or have the agent decide. It can use an OP-1 or other MIDI controller, mouse, touch, or keyboard; it also offers illustrations for common value types and can create custom controls while picking up existing design tokens and palettes. The author says it works well for tuning visual details and comparing options, and can make logic such as window manager layouts easier to grasp, though that example is a naive simulation. The article presents this as a way to add human taste incrementally, while acknowledging that four knobs alone will not prevent "slop."

### Source excerpt

You haven't lost control. You're just holding too much at once. On constraints, the OP-1, and dialing in what agents build, four knobs at a time, with Quadknob.

## Central Agentic Ops (CAO)

DevFeed: [Central Agentic Ops (CAO)](<https://devfeed.tech/articles/central-agentic-ops-cao-121418.md>)

Original publisher: [Read original article](<https://githubnext.com/projects/central-agentic-ops/>)

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

Content type: article

Language: en

Sources: [GitHub Next](<https://devfeed.tech/sources/github-next.md>)

Topics: [GitHub Agentic Workflows](<https://devfeed.tech/topics/github-agentic-workflows.md>), [agentic workflows](<https://devfeed.tech/topics/agentic-workflows.md>), [GitHub Actions](<https://devfeed.tech/topics/github-actions.md>)

Tags: [agentic](<https://devfeed.tech/tags/agentic.md>), [agentic-ops](<https://devfeed.tech/tags/agentic-ops.md>), [agentic-workflows](<https://devfeed.tech/tags/agentic-workflows.md>), [monitor](<https://devfeed.tech/tags/monitor.md>), [ops](<https://devfeed.tech/tags/ops.md>), [repositories](<https://devfeed.tech/tags/repositories.md>), [workflows](<https://devfeed.tech/tags/workflows.md>)

### AI overview

Central Agentic Ops (CAO) is an early research prototype for governing shared agentic workflows across repositories. It brings activity, costs, outcomes, and review effort into one view so teams can decide what work is useful and when it should run. CAO governs whether and where work may run; Agentic Workflows, executed through GitHub Actions, govern how it runs. Teams can begin with bounded, read-only monitoring, then run one task in review mode, where proposed outputs go to a configured review destination before live outputs are explicitly authorized. Execution remains constrained by reviewed policy, repository permissions, declared safe outputs, and audit trails; pull requests still require human acceptance. CAO is not an all-purpose agent, and it governs only explicitly integrated workflows. Its recommendations group recurring failures for review, while choosing and dispatching valuable interventions remains research. The prototype supports several credential and hosting setups, but access, API limits, administrator help, GitHub plan limits, and dashboard visibility can constrain them. Its results may be stale, partial, or unavailable. The article describes research directions,

### Source excerpt

Monitor and govern agentic workflows across repositories

## PDD Kit

DevFeed: [PDD Kit](<https://devfeed.tech/articles/pdd-kit-119350.md>)

Original publisher: [Read original article](<https://githubnext.com/projects/pdd-kit/>)

Published: 2026-11-02T12:00:00Z

Content type: article

Language: en

Sources: [GitHub Next](<https://devfeed.tech/sources/github-next.md>)

Topics: [AI-assisted coding](<https://devfeed.tech/topics/ai-assisted-coding.md>), [Spec Driven Development](<https://devfeed.tech/topics/spec-driven-development.md>), [trust](<https://devfeed.tech/topics/trust.md>), [Test automation](<https://devfeed.tech/topics/test-automation.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>), [qualitygates](<https://devfeed.tech/topics/qualitygates.md>), [code comments](<https://devfeed.tech/topics/code-comments.md>)

Tags: [account](<https://devfeed.tech/tags/account.md>), [admin](<https://devfeed.tech/tags/admin.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agentic-development](<https://devfeed.tech/tags/agentic-development.md>), [agents](<https://devfeed.tech/tags/agents.md>), [api](<https://devfeed.tech/tags/api.md>), [app](<https://devfeed.tech/tags/app.md>), [async](<https://devfeed.tech/tags/async.md>), [automation](<https://devfeed.tech/tags/automation.md>), [backend](<https://devfeed.tech/tags/backend.md>), [bottlenecks](<https://devfeed.tech/tags/bottlenecks.md>), [boundary](<https://devfeed.tech/tags/boundary.md>), [browser](<https://devfeed.tech/tags/browser.md>), [browser-agents](<https://devfeed.tech/tags/browser-agents.md>), [code](<https://devfeed.tech/tags/code.md>), [coding](<https://devfeed.tech/tags/coding.md>), [coding-agents](<https://devfeed.tech/tags/coding-agents.md>), [collaboration](<https://devfeed.tech/tags/collaboration.md>), [trust](<https://devfeed.tech/tags/trust.md>)

### AI overview

PDD Kit shifts review of agent-written code toward human approval of intent and specifications, then uses formal proofs and tests to check implementation. In its Rust prototype, it runs inside Chopin: decision cards help clarify requirements, an intent artifact feeds Verus specifications, and agents generate code and proofs that Verus checks. The Pet Store login example shows the distinction: specified properties such as password gating and token roles are proved for all inputs, while database, bcrypt, and JWT behavior sit behind six trusted contracts with planned empirical tests. Tests provide evidence for behavior the proof cannot cover; they are not proofs. Mutation checks test whether each specification clause rules out a targeted bug, and CI reruns proofs alongside tests. The authors say the prototype currently verifies Rust only, proofs still take effort, and guarantees depend especially on trusted password checks and account lookups.

### Source excerpt

Building trust in the code that your agents wrote

## Limit what agents learn about private repos with enclaves

DevFeed: [Limit what agents learn about private repos with enclaves](<https://devfeed.tech/articles/limit-what-agents-learn-about-private-repos-with-enclaves-114844.md>)

Original publisher: [Read original article](<https://githubnext.com/posts/enclaves/>)

Published: 2026-11-02T00:00:00Z

Content type: article

Language: en

Sources: [GitHub Next](<https://devfeed.tech/sources/github-next.md>)

Topics: [GitHub Agentic Workflows](<https://devfeed.tech/topics/github-agentic-workflows.md>), [agentic workflows](<https://devfeed.tech/topics/agentic-workflows.md>), [Multi Agent Systems](<https://devfeed.tech/topics/multi-agent-systems.md>), [GitHub](<https://devfeed.tech/topics/github.md>), [Cryptography](<https://devfeed.tech/topics/cryptography.md>)

Tags: [agents](<https://devfeed.tech/tags/agents.md>), [attack](<https://devfeed.tech/tags/attack.md>), [broker](<https://devfeed.tech/tags/broker.md>), [budget](<https://devfeed.tech/tags/budget.md>), [communication](<https://devfeed.tech/tags/communication.md>), [github](<https://devfeed.tech/tags/github.md>), [github-agentic-workflows](<https://devfeed.tech/tags/github-agentic-workflows.md>), [repositories](<https://devfeed.tech/tags/repositories.md>), [security](<https://devfeed.tech/tags/security.md>), [workflows](<https://devfeed.tech/tags/workflows.md>)

### AI overview

Enclaves let a workflow's primary agent delegate private-repository analysis while receiving only a schema-constrained result. A trusted broker runs the enclave in isolation, validates its response, and charges the schema's maximum information capacity against a shared per-repository budget before execution. For confidential repositories, the example budget is 8 bits per run; status and timing channels also count. Script enclaves run bounded Python over read-only snapshots without network access, while agent enclaves use a configured model provider and may have a narrow GitHub interface. The primary agent never receives the private snapshot or credentials. These controls block unrestricted file copying but do not make every permitted answer harmless or ensure correctness. Agent enclaves still send repository-derived content to the model provider, and workflows must keep private repositories inaccessible to the primary through other means. Budgets apply per run, so maintainers must consider repeated workflow runs.

### Source excerpt

Enclaves are a new security primitive for cross-repository GitHub Agentic Workflows. Enclaves let agents work on different repositories while limiting what they say to each other.

## Universe 2026: What's Next with GitHub Next

DevFeed: [Universe 2026: What's Next with GitHub Next](<https://devfeed.tech/articles/universe-2026-what-s-next-with-github-next-117598.md>)

Original publisher: [Read original article](<https://githubnext.com/talks/whats-next-with-github-next/>)

Published: 2026-10-29T00:00:00Z

Content type: release

Language: en

Sources: [GitHub Next](<https://devfeed.tech/sources/github-next.md>)

Topics: [GitHub](<https://devfeed.tech/topics/github.md>), [GitHub Copilot](<https://devfeed.tech/topics/github-copilot.md>), [Spec Driven Development](<https://devfeed.tech/topics/spec-driven-development.md>), [GitHub Agentic Workflows](<https://devfeed.tech/topics/github-agentic-workflows.md>), [agentic workflows](<https://devfeed.tech/topics/agentic-workflows.md>), [Test-driven development](<https://devfeed.tech/topics/tdd.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>), [GitHub Actions](<https://devfeed.tech/topics/github-actions.md>), [GitHub Issues](<https://devfeed.tech/topics/github-issues.md>), [GitHub Pages](<https://devfeed.tech/topics/github-pages.md>), [roadmaps](<https://devfeed.tech/topics/roadmaps.md>)

Tags: [2026](<https://devfeed.tech/tags/2026.md>), [ace](<https://devfeed.tech/tags/ace.md>), [agentic](<https://devfeed.tech/tags/agentic.md>), [agentic-development](<https://devfeed.tech/tags/agentic-development.md>), [agentic-ops](<https://devfeed.tech/tags/agentic-ops.md>), [agentic-workflows](<https://devfeed.tech/tags/agentic-workflows.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [alignment](<https://devfeed.tech/tags/alignment.md>), [automation](<https://devfeed.tech/tags/automation.md>), [back-end](<https://devfeed.tech/tags/back-end.md>), [backend](<https://devfeed.tech/tags/backend.md>), [canvas](<https://devfeed.tech/tags/canvas.md>), [collaborative-planning](<https://devfeed.tech/tags/collaborative-planning.md>), [github](<https://devfeed.tech/tags/github.md>), [github-next](<https://devfeed.tech/tags/github-next.md>), [multiplayer](<https://devfeed.tech/tags/multiplayer.md>), [planning](<https://devfeed.tech/tags/planning.md>)

### AI overview

GitHub Next presents three experiments for collaborative software work with agents. Ace 2 runs locally and lets invited teammates share sessions, environments, chat context, and agents; because collaborators get shell access to the host machine, the authors say to work only with trusted people and use a VPN rather than expose it publicly. Chopin provides real-time planning with repository-aware agents, shared plans, decision records, diagrams, and interactive prototypes; it can turn plans into pull requests and update code from later specification changes. Its PDD Kit translates natural-language intent into a reviewed plan and Verus formal specification, then checks implementation with proofs, tests for trusted components, and LLM review; it is in limited research preview. Nexthub is a concept for restructuring GitHub around projects, reusable or overlaid code sources, and first-class goals, decisions, tasks, and releases, with multi-repository changes and app deployment. The article describes Nexthub as an exploration, not an established GitHub offering.

### Source excerpt

Idan Gazit and Maggie Appleton present multiplayer agents, collaborative planning, and proof-driven development as new foundations for GitHub.

## Sparkle Pills

DevFeed: [Sparkle Pills](<https://devfeed.tech/articles/sparkle-pills-67135.md>)

Original publisher: [Read original article](<https://githubnext.com/posts/sparkle-pills/>)

Published: 2026-09-18T00:00:00Z

Content type: tutorial

Language: en

Sources: [GitHub Next](<https://devfeed.tech/sources/github-next.md>)

Topics: [AI Prompt Templates](<https://devfeed.tech/topics/ai-prompt-templates.md>), [coding assistant](<https://devfeed.tech/topics/coding-assistant.md>), [Boilerplate](<https://devfeed.tech/topics/boilerplate.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [ai](<https://devfeed.tech/tags/ai.md>), [prompt](<https://devfeed.tech/tags/prompt.md>), [templating](<https://devfeed.tech/tags/templating.md>)

### AI overview

The article describes "sparkle pills," a templating convention that marks slots for an AI model to fill while preserving surrounding text verbatim. It explains using deterministic code to request slot values, constraints and structured output, parallel generation, and write-time, compile-time, and run-time filling. It also shows examples for document templates, website updates, and pseudocode.

### Source excerpt

A flexible templating convention for contextual inference

## Chopin explores collaborative planning with AI agents

DevFeed: [Chopin explores collaborative planning with AI agents](<https://devfeed.tech/articles/chopin-67141.md>)

Original publisher: [Read original article](<https://githubnext.com/projects/chopin/>)

Published: 2026-09-02T00:00:00Z

Content type: article

Language: en

Sources: [GitHub Next](<https://devfeed.tech/sources/github-next.md>)

Topics: [agentic-coding](<https://devfeed.tech/topics/agentic-coding.md>), [Frontend & real-time](<https://devfeed.tech/topics/frontend-real-time.md>), [Developer experience](<https://devfeed.tech/topics/developer-experience.md>)

Tags: [agentic-engineering](<https://devfeed.tech/tags/agentic-engineering.md>), [agents](<https://devfeed.tech/tags/agents.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [audit](<https://devfeed.tech/tags/audit.md>), [building-software](<https://devfeed.tech/tags/building-software.md>), [collaboration](<https://devfeed.tech/tags/collaboration.md>), [multiplayer](<https://devfeed.tech/tags/multiplayer.md>), [real-time](<https://devfeed.tech/tags/real-time.md>)

### AI overview

GitHub Next describes Chopin, an early open source research prototype for collaborative software planning with teammates and AI agents. It combines an editable MDX-based plan, real-time collaboration, comments, research blocks, visual components, and a persistent decision log. The project explores ways to improve upfront planning and review in agent-assisted software development; the authors say it is not yet a product or finished prototype and do not recommend using it today.

### Source excerpt

A real-time multiplayer, agentic planning environment

## How Knowledge Compressor Reduces Documentation Tokens While Testing Its Usefulness

DevFeed: [How Knowledge Compressor Reduces Documentation Tokens While Testing Its Usefulness](<https://devfeed.tech/articles/knowledge-compressor-67130.md>)

Original publisher: [Read original article](<https://githubnext.com/posts/knowledge-compressor/>)

Published: 2026-08-24T12:00:00Z

Content type: article

Language: en

Sources: [GitHub Next](<https://devfeed.tech/sources/github-next.md>)

Topics: [knowledge-management](<https://devfeed.tech/topics/knowledge-management.md>), [context window](<https://devfeed.tech/topics/context-window.md>), [AI Chat](<https://devfeed.tech/topics/ai-chat.md>), [rate-limiting](<https://devfeed.tech/topics/rate-limiting.md>)

Tags: [compression](<https://devfeed.tech/tags/compression.md>), [context-window](<https://devfeed.tech/tags/context-window.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [gpt](<https://devfeed.tech/tags/gpt.md>), [gpt-5-6](<https://devfeed.tech/tags/gpt-5-6.md>)

### AI overview

Knowledge Compressor iteratively shortens technical documentation and tests whether language models can still answer questions from the compressed text. The article reports that one evaluation cut a 996-token document to 480 tokens while preserving enough information for the tests. Compression has an upfront token and monetary cost, so repeated use determines whether the savings offset it.

### Source excerpt

How far can documentation be compressed without losing its meaning?

## WikiKB: a fast, repository-native knowledge base for people and agents

DevFeed: [WikiKB: a fast, repository-native knowledge base for people and agents](<https://devfeed.tech/articles/wikikb-a-fast-repository-native-knowledge-base-for-people-and-agents-67137.md>)

Original publisher: [Read original article](<https://githubnext.com/posts/wikikb/>)

Published: 2026-08-23T00:00:00Z

Content type: article

Language: en

Sources: [GitHub Next](<https://devfeed.tech/sources/github-next.md>)

Topics: [Wikidata](<https://devfeed.tech/topics/wikidata.md>), [context-architecture](<https://devfeed.tech/topics/context-architecture.md>), [Repository](<https://devfeed.tech/topics/repository.md>), [Agent Skills](<https://devfeed.tech/topics/agent-skills.md>), [GitHub Issues](<https://devfeed.tech/topics/github-issues.md>)

Tags: [agents](<https://devfeed.tech/tags/agents.md>), [graphrag](<https://devfeed.tech/tags/graphrag.md>), [knowledge-base](<https://devfeed.tech/tags/knowledge-base.md>), [repository](<https://devfeed.tech/tags/repository.md>)

### AI overview

WikiKB is a curated knowledge base built on GitHub repository wikis for humans and AI agents. It supports ingesting and querying material through command-line tools, GitHub issues, and an agent skill, with Markdown entries, provenance, metadata, and namespaces. Its local CPU-based indexing and retrieval work without an LLM provider; an LLM can optionally be configured for summarization and question answering.

### Source excerpt

WikiKB is an efficient, transparent, and explicitly curated knowledge base layer for humans and agents, built on the GitHub repository wiki.

## Evergreen automates pull request fixes and branch updates

DevFeed: [Evergreen automates pull request fixes and branch updates](<https://devfeed.tech/articles/evergreen-your-pr-gardener-67127.md>)

Original publisher: [Read original article](<https://githubnext.com/posts/evergreen/>)

Published: 2026-07-22T00:00:00Z

Content type: release

Language: en

Sources: [GitHub Next](<https://devfeed.tech/sources/github-next.md>)

Topics: [Pull Request](<https://devfeed.tech/topics/pull-request.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [agentic-workflow](<https://devfeed.tech/tags/agentic-workflow.md>), [ci](<https://devfeed.tech/tags/ci.md>), [guide](<https://devfeed.tech/tags/guide.md>), [pull-requests](<https://devfeed.tech/tags/pull-requests.md>), [software-development](<https://devfeed.tech/tags/software-development.md>), [workflow](<https://devfeed.tech/tags/workflow.md>), [workflows](<https://devfeed.tech/tags/workflows.md>)

### AI overview

Evergreen is an agentic workflow that monitors pull requests for CI failures, linting issues, outdated tests, and merge conflicts. It learns a repository's mergeability rules from its guidance and history, then can resolve issues and keep branches current. Installation copies workflow files into the repository, configures policy, and opens an installation pull request.

### Source excerpt

Evergreen helps keep pull requests green and mergeable by fixing CI failures and merge conflicts, and keeping PR branches up to date with the base branch.

## How Agentic Workflows Automate Code While Keeping Planning and Collaboration Multiplayer

DevFeed: [How Agentic Workflows Automate Code While Keeping Planning and Collaboration Multiplayer](<https://devfeed.tech/articles/realtime-multiplayer-automation-and-you-67188.md>)

Original publisher: [Read original article](<https://githubnext.com/talks/realtime-multiplayer-automation-and-you/>)

Published: 2026-07-02T00:00:00Z

Content type: article

Language: en

Sources: [GitHub Next](<https://devfeed.tech/sources/github-next.md>)

Topics: [GitHub Agentic Workflows](<https://devfeed.tech/topics/github-agentic-workflows.md>), [agentic workflows](<https://devfeed.tech/topics/agentic-workflows.md>), [GitHub](<https://devfeed.tech/topics/github.md>), [GitHub Copilot](<https://devfeed.tech/topics/github-copilot.md>), [agentic-coding](<https://devfeed.tech/topics/agentic-coding.md>)

Tags: [agentic](<https://devfeed.tech/tags/agentic.md>), [agentic-workflows](<https://devfeed.tech/tags/agentic-workflows.md>), [coding-agents](<https://devfeed.tech/tags/coding-agents.md>), [collaboration](<https://devfeed.tech/tags/collaboration.md>), [github](<https://devfeed.tech/tags/github.md>), [github-copilot](<https://devfeed.tech/tags/github-copilot.md>)

### AI overview

The talk demonstrates agentic workflows that automate code while keeping planning and collaboration multiplayer. It features secure GitHub Copilot automations and Ace, GitHub Next's shared workspace for people and coding agents.

### Source excerpt

How agentic workflows can automate code while keeping planning and collaboration multiplayer, with demos of secure GitHub Copilot automations and Ace, GitHub Next's shared workspace for people and coding agents.

## Can agents be proud of their work?

DevFeed: [Can agents be proud of their work?](<https://devfeed.tech/articles/can-agents-be-proud-of-their-work-67109.md>)

Original publisher: [Read original article](<https://githubnext.com/posts/can-agents-be-proud-of-their-work/>)

Published: 2026-07-01T00:00:00Z

Content type: opinion

Language: en

Sources: [GitHub Next](<https://devfeed.tech/sources/github-next.md>)

Topics: [agent orchestration](<https://devfeed.tech/topics/agent-orchestration.md>), [AI-assisted coding](<https://devfeed.tech/topics/ai-assisted-coding.md>)

Tags: [agents](<https://devfeed.tech/tags/agents.md>), [autonomous-agents](<https://devfeed.tech/tags/autonomous-agents.md>), [large-language-models](<https://devfeed.tech/tags/large-language-models.md>)

### AI overview

The document suggests asking autonomous agents whether they are proud of their work and having them keep iterating when they are not. It presents this as a possible way to improve agent output, building on the earlier practice of asking language models to show their work.

### Source excerpt

Can agents be proud of their work?

## Agentic Workflows Can Use Local Inference

DevFeed: [Agentic Workflows Can Use Local Inference](<https://devfeed.tech/articles/agentic-workflows-can-use-local-inference-67107.md>)

Original publisher: [Read original article](<https://githubnext.com/posts/agentic-workflows-local-inference/>)

Published: 2026-06-22T00:00:00Z

Content type: article

Language: en

Sources: [GitHub Next](<https://devfeed.tech/sources/github-next.md>)

Topics: [GitHub Agentic Workflows](<https://devfeed.tech/topics/github-agentic-workflows.md>), [Local AI](<https://devfeed.tech/topics/local-ai.md>), [self-hosted runner](<https://devfeed.tech/topics/self-hosted-runner.md>)

Tags: [agentic](<https://devfeed.tech/tags/agentic.md>), [inference](<https://devfeed.tech/tags/inference.md>), [local](<https://devfeed.tech/tags/local.md>), [workflows](<https://devfeed.tech/tags/workflows.md>)

### AI overview

GitHub Agentic Workflows can run on custom Actions runners and use alternate models or model-routing platforms, including local inference on a Mac. Demonstration workflows are available to help users get started.

### Source excerpt

Agentic Workflows gives you complete control to customize your Actions runtime and choose your AI model. You can deploy Agentic Workflows to a custom runner of your choosing and use any compatible AI host or model-routing platform. You can even use local inference running on your Mac!

## GitHub Goal turns repo issues into agentic workflows with success criteria

DevFeed: [GitHub Goal turns repo issues into agentic workflows with success criteria](<https://devfeed.tech/articles/goooooooal-67128.md>)

Original publisher: [Read original article](<https://githubnext.com/posts/goal/>)

Published: 2026-06-11T00:00:00Z

Content type: article

Language: en

Sources: [GitHub Next](<https://devfeed.tech/sources/github-next.md>)

Topics: [Loop Engineering](<https://devfeed.tech/topics/loop-engineering.md>), [agentic-coding](<https://devfeed.tech/topics/agentic-coding.md>), [GitHub Issues](<https://devfeed.tech/topics/github-issues.md>)

Tags: [agentic](<https://devfeed.tech/tags/agentic.md>), [agentic-workflow](<https://devfeed.tech/tags/agentic-workflow.md>), [coding](<https://devfeed.tech/tags/coding.md>), [github](<https://devfeed.tech/tags/github.md>), [loop-engineering](<https://devfeed.tech/tags/loop-engineering.md>)

### AI overview

Goal is an agentic workflow that starts from a repo issue labeled `goal` and works toward defined success criteria. The article says a good goal should be a coherent objective with an observable stopping condition, larger than a quick prompt but smaller than an open-ended backlog.

### Source excerpt

Turn an issue into an agentic mission

## GitHub Next and the Future of Coding

DevFeed: [GitHub Next and the Future of Coding](<https://devfeed.tech/articles/github-next-and-the-future-of-coding-67184.md>)

Original publisher: [Read original article](<https://githubnext.com/talks/github-next-and-the-future-of-coding/>)

Published: 2026-06-10T00:00:00Z

Content type: opinion

Language: en

Sources: [GitHub Next](<https://devfeed.tech/sources/github-next.md>)

Topics: [future of software](<https://devfeed.tech/topics/future-of-software.md>), [rachels-ramblings](<https://devfeed.tech/topics/rachels-ramblings.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [ai](<https://devfeed.tech/tags/ai.md>), [engineering](<https://devfeed.tech/tags/engineering.md>)

### AI overview

The document argues that AI can make manual code typing obsolete, shifting software engineering teams toward curating agent-generated code and applying architectural judgment.

### Source excerpt

Why AI makes typing code obsolete, and why engineering teams must shift from writing syntax to curating agent-generated code and mastering architectural judgment.

## Crane: verified code migration

DevFeed: [Crane: verified code migration](<https://devfeed.tech/articles/crane-verified-code-migration-67112.md>)

Original publisher: [Read original article](<https://githubnext.com/posts/crane/>)

Published: 2026-06-08T00:00:00Z

Content type: article

Language: en

Sources: [GitHub Next](<https://devfeed.tech/sources/github-next.md>)

Topics: [cloud-migration](<https://devfeed.tech/topics/cloud-migration.md>), [Code](<https://devfeed.tech/topics/code.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>), [Pull Request](<https://devfeed.tech/topics/pull-request.md>), [Refactoring](<https://devfeed.tech/topics/refactoring.md>)

Tags: [agentic](<https://devfeed.tech/tags/agentic.md>), [coding](<https://devfeed.tech/tags/coding.md>), [coding-agent](<https://devfeed.tech/tags/coding-agent.md>), [github](<https://devfeed.tech/tags/github.md>), [migration](<https://devfeed.tech/tags/migration.md>), [migrations](<https://devfeed.tech/tags/migrations.md>)

### AI overview

Crane is a GitHub-based agentic workflow for planning and carrying out code migrations in small, verifiable steps. It uses tests and a configurable JSON migration score to decide whether to accept progress, records migration state in a version-controlled branch, and advances work through a draft pull request for human review.

### Source excerpt

Crane is a migration assistant for GitHub repositories that plans, executes, and verifies code migrations in small agentic steps while keeping humans in control.

## Agentics Beyond Code

DevFeed: [Agentics Beyond Code](<https://devfeed.tech/articles/agentics-beyond-code-67111.md>)

Original publisher: [Read original article](<https://githubnext.com/posts/chrizbo-agentics-beyond-code/>)

Published: 2026-05-29T00:00:00Z

Content type: article

Language: en

Sources: [GitHub Next](<https://devfeed.tech/sources/github-next.md>)

Topics: [GitHub Agentic Workflows](<https://devfeed.tech/topics/github-agentic-workflows.md>), [Loop Engineering](<https://devfeed.tech/topics/loop-engineering.md>), [AI-assisted coding](<https://devfeed.tech/topics/ai-assisted-coding.md>)

Tags: [agentic-workflows](<https://devfeed.tech/tags/agentic-workflows.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [github-agentic-workflows](<https://devfeed.tech/tags/github-agentic-workflows.md>), [open-source](<https://devfeed.tech/tags/open-source.md>), [products](<https://devfeed.tech/tags/products.md>), [workflows](<https://devfeed.tech/tags/workflows.md>)

### AI overview

Agentics Beyond Code is an open-source collection of GitHub Agentic Workflows that applies agents to product, compliance, and leadership operations. The article describes artifact-focused workflows, policy files, constrained outputs, and a split in which deterministic scripts gather data while agents analyze it and generate reports. Teams can start with one recurring artifact, such as launch readiness or weekly status, and adapt its policy to their needs.

### Source excerpt

What happens when you give PMs, compliance teams, and leaders their own agents? A tour of Agentics Beyond Code -- an open-source set of GitHub Agentic Workflows for the non-engineering roles that ship, govern, and operate products.

## Control what your agentic workflows see with integrity filtering

DevFeed: [Control what your agentic workflows see with integrity filtering](<https://devfeed.tech/articles/control-what-your-agentic-workflows-see-with-integrity-filtering-67129.md>)

Original publisher: [Read original article](<https://githubnext.com/posts/integrity-filtering/>)

Published: 2026-05-26T00:00:00Z

Content type: article

Language: en

Sources: [GitHub Next](<https://devfeed.tech/sources/github-next.md>)

Topics: [GitHub Agentic Workflows](<https://devfeed.tech/topics/github-agentic-workflows.md>), [ai security](<https://devfeed.tech/topics/ai-security.md>), [agentic-coding](<https://devfeed.tech/topics/agentic-coding.md>)

Tags: [agent-security](<https://devfeed.tech/tags/agent-security.md>), [github](<https://devfeed.tech/tags/github.md>), [github-agentic-workflows](<https://devfeed.tech/tags/github-agentic-workflows.md>), [integrity](<https://devfeed.tech/tags/integrity.md>), [workflows](<https://devfeed.tech/tags/workflows.md>)

### AI overview

GitHub Agentic Workflows adds integrity filtering to control which repository content an agent can see. The policy uses author relationships, maintainer endorsements, and configurable rules such as trusted or blocked accounts, labels, and reactions. The article explains how the system applies these rules through a decentralized information-flow-control model and notes that it does not yet prevent data leaks across repositories.

### Source excerpt

GitHub Agentic Workflows filter untrusted GitHub content before it reaches the agent. Here's why integrity filtering matters for repository maintainers, and how we built it.

## Agent Functions

DevFeed: [Agent Functions](<https://devfeed.tech/articles/agent-functions-67106.md>)

Original publisher: [Read original article](<https://githubnext.com/posts/agent-functions/>)

Published: 2026-05-22T00:00:00Z

Content type: article

Language: en

Sources: [GitHub Next](<https://devfeed.tech/sources/github-next.md>)

Topics: [Agent Harness](<https://devfeed.tech/topics/agent-harness.md>), [agent orchestration](<https://devfeed.tech/topics/agent-orchestration.md>), [hardened ai agent skills](<https://devfeed.tech/topics/hardened-ai-agent-skills.md>), [Developer experience](<https://devfeed.tech/topics/developer-experience.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [agent-harness](<https://devfeed.tech/tags/agent-harness.md>), [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [code](<https://devfeed.tech/tags/code.md>), [cost](<https://devfeed.tech/tags/cost.md>), [deterministic-workflows](<https://devfeed.tech/tags/deterministic-workflows.md>), [engineering](<https://devfeed.tech/tags/engineering.md>)

### AI overview

Agent Functions proposes structuring agent work as reusable, typed functions stored in a registry and executed in capability-limited microVMs. Planning agents select or create task definitions and implementations; execution agents receive only the tools needed for their assigned steps. The proposal emphasizes reuse, testing, benchmarking, reduced context use, and limiting permissions.

### Source excerpt

Prompts are programs. You wouldn't write a complex program completely from scratch, in a big, soupy loop without subroutines, and then write it again the next time you wanted to run it. Why let your agent work that way?

## Canary: a harm gate for agentic systems

DevFeed: [Canary: a harm gate for agentic systems](<https://devfeed.tech/articles/canary-a-harm-gate-for-agentic-systems-67110.md>)

Original publisher: [Read original article](<https://githubnext.com/posts/canary-harm-gate-agentic-systems/>)

Published: 2026-05-18T00:00:00Z

Content type: article

Language: en

Sources: [GitHub Next](<https://devfeed.tech/sources/github-next.md>)

Topics: [malicious packages](<https://devfeed.tech/topics/malicious-packages.md>), [agentic-coding](<https://devfeed.tech/topics/agentic-coding.md>), [Agentic AI](<https://devfeed.tech/topics/what-is-agentic-ai.md>)

Tags: [agentic](<https://devfeed.tech/tags/agentic.md>), [agentic-systems](<https://devfeed.tech/tags/agentic-systems.md>), [agentic-workflows](<https://devfeed.tech/tags/agentic-workflows.md>), [canary](<https://devfeed.tech/tags/canary.md>), [claude](<https://devfeed.tech/tags/claude.md>), [claude-opus-4-8](<https://devfeed.tech/tags/claude-opus-4-8.md>), [cost](<https://devfeed.tech/tags/cost.md>)

### AI overview

Canary is a constrained classification gate placed before more capable agents that analyze untrusted artifacts. It applies task-specific harm policies, gathers binary votes from detectors, and blocks when their results are uncertain. The article describes the gate's design, benchmark trade-offs, and limits as a way to reduce prompt-injection and malicious-package risks in agentic workflows.

### Source excerpt

Canary puts a small, auditable gate in front of agentic workflows so untrusted artifacts are classified before powerful agents act on them.

## The Impact of Automated Repository Maintenance Assistance

DevFeed: [The Impact of Automated Repository Maintenance Assistance](<https://devfeed.tech/articles/the-impact-of-automated-repository-maintenance-assistance-67116.md>)

Original publisher: [Read original article](<https://githubnext.com/posts/dsyme-impact-automated-repo-maintenance/>)

Published: 2026-05-14T00:00:00Z

Content type: article

Language: en

Sources: [GitHub Next](<https://devfeed.tech/sources/github-next.md>)

Topics: [Application Development](<https://devfeed.tech/topics/application-development.md>), [Software Engineering](<https://devfeed.tech/topics/software-engineering.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [automated](<https://devfeed.tech/tags/automated.md>), [maintainers](<https://devfeed.tech/tags/maintainers.md>), [merge](<https://devfeed.tech/tags/merge.md>), [open-source](<https://devfeed.tech/tags/open-source.md>), [proactive](<https://devfeed.tech/tags/proactive.md>), [repository](<https://devfeed.tech/tags/repository.md>)

### AI overview

A proactive AI repository agent was deployed across 13 open source repositories, where it closed 578 issues and was associated with an eightfold increase in median issue closure velocity and a tenfold increase in pull request merge velocity. The account identifies how quickly human maintainers decide to act as the most important factor.

### Source excerpt

The Impact of Automated Repository Maintenance Assistance

## Agents are power tools

DevFeed: [Agents are power tools](<https://devfeed.tech/articles/agents-are-power-tools-67108.md>)

Original publisher: [Read original article](<https://githubnext.com/posts/agents-are-power-tools/>)

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

Content type: opinion

Language: en

Sources: [GitHub Next](<https://devfeed.tech/sources/github-next.md>)

Topics: [Loop Engineering](<https://devfeed.tech/topics/loop-engineering.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>), [Advanced AI workflows](<https://devfeed.tech/topics/advanced-ai-workflows.md>)

Tags: [agentic-engineering](<https://devfeed.tech/tags/agentic-engineering.md>), [agents](<https://devfeed.tech/tags/agents.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [in-the-loop](<https://devfeed.tech/tags/in-the-loop.md>), [orchestration](<https://devfeed.tech/tags/orchestration.md>), [precision](<https://devfeed.tech/tags/precision.md>), [safety](<https://devfeed.tech/tags/safety.md>), [workflows](<https://devfeed.tech/tags/workflows.md>)

### AI overview

The article proposes treating agents as powerful tools rather than teammates, and recommends designing agent workflows with clear constraints, repeatable steps, supervision, and human review. It argues that process design helps teams manage risks while making useful automation possible.

### Source excerpt

A practical mental model for agents, workflows, and human-machine systems in agentic engineering.

## Understanding Repositories as Human/Agent Knowledge Factories

DevFeed: [Understanding Repositories as Human/Agent Knowledge Factories](<https://devfeed.tech/articles/understanding-repositories-as-human-agent-knowledge-factories-67122.md>)

Original publisher: [Read original article](<https://githubnext.com/posts/dsyme-repositories-knowledge-factories/>)

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

Content type: article

Language: en

Sources: [GitHub Next](<https://devfeed.tech/sources/github-next.md>)

Topics: [knowledge-management](<https://devfeed.tech/topics/knowledge-management.md>), [agentic workflows](<https://devfeed.tech/topics/agentic-workflows.md>), [Loop Engineering](<https://devfeed.tech/topics/loop-engineering.md>)

Tags: [ai-generated-code](<https://devfeed.tech/tags/ai-generated-code.md>), [automation](<https://devfeed.tech/tags/automation.md>), [code](<https://devfeed.tech/tags/code.md>), [repositories](<https://devfeed.tech/tags/repositories.md>), [velocity](<https://devfeed.tech/tags/velocity.md>)

### AI overview

The article discusses using repository automation and agentic processes to clean up AI-generated code after it is produced, as an alternative to adding review hurdles that slow team velocity.

### Source excerpt

Understanding Repositories as Human/Agent Knowledge Factories

## Agency is the New Resilience

DevFeed: [Agency is the New Resilience](<https://devfeed.tech/articles/agency-is-the-new-resilience-67105.md>)

Original publisher: [Read original article](<https://githubnext.com/posts/agency-is-the-new-resilience/>)

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

Content type: article

Language: en

Sources: [GitHub Next](<https://devfeed.tech/sources/github-next.md>)

Topics: [Resilience](<https://devfeed.tech/topics/resilience.md>), [Loop Engineering](<https://devfeed.tech/topics/loop-engineering.md>)

Tags: [agentic-engineering](<https://devfeed.tech/tags/agentic-engineering.md>), [agentic-workflows](<https://devfeed.tech/tags/agentic-workflows.md>), [agents](<https://devfeed.tech/tags/agents.md>), [resilience](<https://devfeed.tech/tags/resilience.md>)

### AI overview

The article argues that the nondeterministic behavior of language model agents can make them resilient as they pursue goals across changing environments. It describes strategies such as adapting API requests after errors, rediscovering moved resources, and using information from error responses, and suggests designing systems to help agents recover.

### Source excerpt

Agents can power robust workflows by intelligently reacting to unexpected conditions, creating a new form of flexible resilience.

[Next page](<https://devfeed.tech/sources/github-next.md?cursor=WyIyMDI2LTA1LTA1VDAwOjAwOjAwKzAwOjAwIiwgIjE5ZDAxOGI5LTI2YjktNDc4OS1iM2RhLTc5OTdmMDViMzc4MCJd>)