# Nirmata

Unified Governance for Agentic Era

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

## NIST SP 800-218 and Kubernetes: what policy can enforce

DevFeed: [NIST SP 800-218 and Kubernetes: what policy can enforce](<https://devfeed.tech/articles/nist-sp-800-218-and-kubernetes-what-policy-can-enforce-144282.md>)

Original publisher: [Read original article](<https://nirmata.com/2026/11/20/nist-sp-800-218-ssdf-and-kubernetes-which-controls-policy-can-enforce/>)

Author: Ritesh Patel

Published: 2026-11-19T20:41:35.692178Z

Content type: article

Language: en

Sources: [Nirmata](<https://devfeed.tech/sources/nirmata.md>)

Topics: [Kyverno](<https://devfeed.tech/topics/kyverno.md>), [policy-as-code](<https://devfeed.tech/topics/policy-as-code.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>), [federal information processing standards](<https://devfeed.tech/topics/federal-information-processing-standards.md>)

Tags: [3](<https://devfeed.tech/tags/3.md>), [admission](<https://devfeed.tech/tags/admission.md>), [ci](<https://devfeed.tech/tags/ci.md>), [compliance](<https://devfeed.tech/tags/compliance.md>), [container-images](<https://devfeed.tech/tags/container-images.md>), [continuous-compliance](<https://devfeed.tech/tags/continuous-compliance.md>), [image-verification](<https://devfeed.tech/tags/image-verification.md>), [kubernetes](<https://devfeed.tech/tags/kubernetes.md>), [kubernetes-policy](<https://devfeed.tech/tags/kubernetes-policy.md>), [kubernetes-supply-chain-security](<https://devfeed.tech/tags/kubernetes-supply-chain-security.md>), [kyverno](<https://devfeed.tech/tags/kyverno.md>), [nist](<https://devfeed.tech/tags/nist.md>), [pod-security](<https://devfeed.tech/tags/pod-security.md>), [policy](<https://devfeed.tech/tags/policy.md>), [records](<https://devfeed.tech/tags/records.md>), [settings](<https://devfeed.tech/tags/settings.md>), [supply-chain-security](<https://devfeed.tech/tags/supply-chain-security.md>)

### AI overview

The excerpt maps NIST SP 800-218 v1.1's 42 active practices to enforcement: three can be checked at Kubernetes admission, eleven through CI gates, and the remaining 28 require records or human review. Admission checks cover approved image registries, image provenance attestations, and a baseline for deployed pods; they cannot establish that an image's SBOM is complete or that software itself has secure defaults. CI gates can check code and tests, pinned build tools, approved dependencies, and artifact signing and archiving. The mapping is the guide's classification, not NIST guidance. It recommends starting the Kyverno registry policy and Pod Security checks in audit or warning mode, reviewing findings before enforcement; registry examples must be customized, and the policy does not cover custom resources that create pods.

### Source excerpt

NIST SP 800-218 names no cluster settings. Admission policy can enforce 3 of its 42 practices, CI gates can check 11, and 28 need records or human review.

## How to limit MCP tool output size in Claude Code

DevFeed: [How to limit MCP tool output size in Claude Code](<https://devfeed.tech/articles/how-to-limit-mcp-tool-output-size-in-claude-code-144281.md>)

Original publisher: [Read original article](<https://nirmata.com/2026/11/20/how-to-limit-mcp-tool-output-size-in-claude-code/>)

Author: Ritesh Patel

Published: 2026-11-19T20:41:35.692178Z

Content type: tutorial

Language: en

Sources: [Nirmata](<https://devfeed.tech/sources/nirmata.md>)

Topics: [Claude Code](<https://devfeed.tech/topics/claude-code.md>), [MSP MCP](<https://devfeed.tech/topics/msp-mcp.md>), [Claude](<https://devfeed.tech/topics/claude.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [airtable](<https://devfeed.tech/tags/airtable.md>), [audit](<https://devfeed.tech/tags/audit.md>), [call](<https://devfeed.tech/tags/call.md>), [claude](<https://devfeed.tech/tags/claude.md>), [claude-code](<https://devfeed.tech/tags/claude-code.md>), [context](<https://devfeed.tech/tags/context.md>), [control](<https://devfeed.tech/tags/control.md>), [cost](<https://devfeed.tech/tags/cost.md>), [cost-management](<https://devfeed.tech/tags/cost-management.md>), [file](<https://devfeed.tech/tags/file.md>), [how-to](<https://devfeed.tech/tags/how-to.md>), [limit](<https://devfeed.tech/tags/limit.md>), [max](<https://devfeed.tech/tags/max.md>), [mcp](<https://devfeed.tech/tags/mcp.md>), [mcp-tool](<https://devfeed.tech/tags/mcp-tool.md>), [security](<https://devfeed.tech/tags/security.md>), [set](<https://devfeed.tech/tags/set.md>), [size](<https://devfeed.tech/tags/size.md>), [text](<https://devfeed.tech/tags/text.md>), [tokens](<https://devfeed.tech/tags/tokens.md>), [tool](<https://devfeed.tech/tags/tool.md>)

### AI overview

Set `MAX_MCP_OUTPUT_TOKENS` in the shell before starting Claude Code to control when successful MCP results are saved to the session's `tool-results` directory and replaced in the conversation by a file path. Its documented default is 25,000 tokens; a separate 50,000-character threshold for text results remains fixed, and the 10,000-token transcript warning does not stop output. Saving a result to disk does not prevent Claude from reading it later. The source also distinguishes server startup timeout (`MCP_TIMEOUT`, default 30,000 ms) from tool execution timeout (`MCP_TOOL_TIMEOUT`, default 100,000,000 ms), which a server's `.mcp.json` `timeout` can override. The token setting does not filter server output, establish server trust, or change transport limits. Server developers can raise a tool's text threshold with `anthropic/maxResultSizeChars`, up to 500,000 characters.

### Source excerpt

Set MAX_MCP_OUTPUT_TOKENS before starting Claude Code. The default is 25,000 tokens, and text over 50,000 characters is saved to a file whatever the setting.

## Kubernetes admission webhook failure policy: Fail or Ignore

DevFeed: [Kubernetes admission webhook failure policy: Fail or Ignore](<https://devfeed.tech/articles/kubernetes-admission-webhook-failure-policy-fail-or-ignore-141135.md>)

Original publisher: [Read original article](<https://nirmata.com/2026/11/19/kubernetes-admission-webhook-failure-policy-fail-or-ignore/>)

Author: Ritesh Patel

Published: 2026-11-18T20:32:39.731906Z

Content type: article

Language: en

Sources: [Nirmata](<https://devfeed.tech/sources/nirmata.md>)

Topics: [Kyverno](<https://devfeed.tech/topics/kyverno.md>), [policy-as-code](<https://devfeed.tech/topics/policy-as-code.md>), [Open Policy Agent](<https://devfeed.tech/topics/open-policy-agent.md>)

Tags: [admission](<https://devfeed.tech/tags/admission.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [fail](<https://devfeed.tech/tags/fail.md>), [failure](<https://devfeed.tech/tags/failure.md>), [ignore](<https://devfeed.tech/tags/ignore.md>), [kubernetes](<https://devfeed.tech/tags/kubernetes.md>), [kubernetes-security](<https://devfeed.tech/tags/kubernetes-security.md>), [policy](<https://devfeed.tech/tags/policy.md>), [policy-as-code](<https://devfeed.tech/tags/policy-as-code.md>), [policy-engine](<https://devfeed.tech/tags/policy-engine.md>), [validating-admission-policy](<https://devfeed.tech/tags/validating-admission-policy.md>), [webhook](<https://devfeed.tech/tags/webhook.md>)

### AI overview

Kubernetes `failurePolicy` governs what happens when an admission webhook cannot provide an answer: `Fail` rejects the API request (the default), while `Ignore` lets it continue without that check. Neither overrides an explicit webhook denial. Timeouts, network and response errors can trigger the policy; the default timeout is 10 seconds, configurable from 1 to 30. For critical controls on agent pods, the text recommends `Fail`, scoped to agent namespaces with a short timeout, while warning that outages can block writes and dependency loops can prevent the webhook from recovering. It suggests inventorying registered webhooks, testing in a namespace, monitoring error-caused rejections, and checking the generated configuration before relying on Kyverno's failure behavior: its docs specify timeout settings but not the policy applied.

### Source excerpt

failurePolicy sets what happens when an admission webhook cannot answer. Fail rejects the request, Ignore admits it unchecked. How to choose for policy engines.

## How to use .cursorignore to keep secrets out of Cursor

DevFeed: [How to use .cursorignore to keep secrets out of Cursor](<https://devfeed.tech/articles/how-to-use-cursorignore-to-keep-secrets-out-of-cursor-141134.md>)

Original publisher: [Read original article](<https://nirmata.com/2026/11/19/how-to-use-cursorignore-to-keep-secrets-out-of-cursor/>)

Author: Ritesh Patel

Published: 2026-11-18T20:32:39.731906Z

Content type: tutorial

Language: en

Sources: [Nirmata](<https://devfeed.tech/sources/nirmata.md>)

Topics: [cursor](<https://devfeed.tech/topics/cursor.md>), [Secrets Management](<https://devfeed.tech/topics/secrets-management.md>), [Cybersecurity](<https://devfeed.tech/topics/cybersecurity.md>), [AI-assisted coding](<https://devfeed.tech/topics/ai-assisted-coding.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-coding-agents](<https://devfeed.tech/tags/ai-coding-agents.md>), [ai-governance](<https://devfeed.tech/tags/ai-governance.md>), [claude](<https://devfeed.tech/tags/claude.md>), [claude-code](<https://devfeed.tech/tags/claude-code.md>), [coding-agents](<https://devfeed.tech/tags/coding-agents.md>), [credentials](<https://devfeed.tech/tags/credentials.md>), [cursor](<https://devfeed.tech/tags/cursor.md>), [developer](<https://devfeed.tech/tags/developer.md>), [filter](<https://devfeed.tech/tags/filter.md>), [git](<https://devfeed.tech/tags/git.md>), [gitignore](<https://devfeed.tech/tags/gitignore.md>), [global](<https://devfeed.tech/tags/global.md>), [governance](<https://devfeed.tech/tags/governance.md>), [ignore](<https://devfeed.tech/tags/ignore.md>), [indexing](<https://devfeed.tech/tags/indexing.md>), [protection](<https://devfeed.tech/tags/protection.md>), [secret](<https://devfeed.tech/tags/secret.md>), [secrets](<https://devfeed.tech/tags/secrets.md>), [secrets-management](<https://devfeed.tech/tags/secrets-management.md>), [security](<https://devfeed.tech/tags/security.md>), [tab](<https://devfeed.tech/tags/tab.md>)

### AI overview

A root-level `.cursorignore` uses `.gitignore`-style patterns to keep matched files from Cursor Agent and Tab and from @ mention references, but Cursor's documentation says this is not complete protection: terminal and MCP tools may still read ignored files. Defaults cover `.env*` and `.gitignore` matches, but not example names such as `*.key`, `*.pem`, `credentials.json`, or `secrets.json`; add patterns for those. A global ignore list applies across projects, and Hierarchical Cursor Ignore can search parent directories for ignore files. Negation cannot restore a file beneath an excluded directory unless the nested directory is re-included first. Test with a dummy file in Cursor itself, since `git check-ignore` checks only `.gitignore`; record the Cursor version because the cited reference does not specify an app version. Claude Code uses a separate `permissions.deny` format, so rules must be written and tested for each tool.

### Source excerpt

Use a .cursorignore file to keep secret files away from Cursor's Agent and Tab. The Cursor docs say this blocks access but is not complete protection.

## How to pin MCP server versions for Claude Code and Cursor

DevFeed: [How to pin MCP server versions for Claude Code and Cursor](<https://devfeed.tech/articles/how-to-pin-mcp-server-versions-for-claude-code-and-cursor-139212.md>)

Original publisher: [Read original article](<https://nirmata.com/2026/11/18/how-to-pin-mcp-server-versions-for-claude-code-and-cursor/>)

Author: Ritesh Patel

Published: 2026-11-18T01:14:41.413057Z

Content type: tutorial

Language: en

Sources: [Nirmata](<https://devfeed.tech/sources/nirmata.md>)

Topics: [versioning](<https://devfeed.tech/topics/versioning.md>), [MSP MCP](<https://devfeed.tech/topics/msp-mcp.md>), [Claude Code](<https://devfeed.tech/topics/claude-code.md>), [Documentation](<https://devfeed.tech/topics/documentation.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [cache](<https://devfeed.tech/tags/cache.md>), [checklist](<https://devfeed.tech/tags/checklist.md>), [claude-code](<https://devfeed.tech/tags/claude-code.md>), [client](<https://devfeed.tech/tags/client.md>), [command](<https://devfeed.tech/tags/command.md>), [config](<https://devfeed.tech/tags/config.md>), [credentials](<https://devfeed.tech/tags/credentials.md>), [cursor](<https://devfeed.tech/tags/cursor.md>), [dependabot](<https://devfeed.tech/tags/dependabot.md>), [dependencies](<https://devfeed.tech/tags/dependencies.md>), [docs](<https://devfeed.tech/tags/docs.md>), [documentation](<https://devfeed.tech/tags/documentation.md>), [environment-variables](<https://devfeed.tech/tags/environment-variables.md>), [launch](<https://devfeed.tech/tags/launch.md>), [mcp](<https://devfeed.tech/tags/mcp.md>), [mcp-server](<https://devfeed.tech/tags/mcp-server.md>), [pin](<https://devfeed.tech/tags/pin.md>), [security](<https://devfeed.tech/tags/security.md>), [server](<https://devfeed.tech/tags/server.md>), [supply-chain-security](<https://devfeed.tech/tags/supply-chain-security.md>), [updates](<https://devfeed.tech/tags/updates.md>), [version](<https://devfeed.tech/tags/version.md>)

### AI overview

Neither Claude Code nor Cursor documents a server-version field, range, or lockfile in the MCP configuration described. For npm servers, the article says an exact version can be named in the `npx` package specifier; that is npm behavior, and the examples were not tested in either client. It recommends recording the reviewed version alongside the config, while noting this is for reviewers and is not read by the tools. Cursor's FAQ says to remove an npm server, clear the npm cache, and re-add it to get the latest version; custom servers are updated by changing local files and restarting Cursor. Claude Code separately refreshes tool, prompt, and resource lists after `list_changed`; its `MCP_SDK_GENERATION` setting selects the client runtime, not a server version. The excerpt also advises using environment-variable references for credentials and checking tool lists after changes.

### Source excerpt

Neither Claude Code nor Cursor documents a setting to pin an MCP server version. Here is what their docs do say about launch commands, env vars and updates.

## How AT&T Turns Signed Evidence into Deployment Decisions with Kyverno

DevFeed: [How AT&T Turns Signed Evidence into Deployment Decisions with Kyverno](<https://devfeed.tech/articles/how-at-t-turns-signed-evidence-into-deployment-decisions-with-kyverno-139209.md>)

Original publisher: [Read original article](<https://nirmata.com/2026/11/18/how-att-turns-signed-evidence-into-deployment-decisions-with-kyverno/>)

Author: Sagar Kundral

Published: 2026-11-18T01:14:41.413057Z

Content type: article

Language: en

Sources: [Nirmata](<https://devfeed.tech/sources/nirmata.md>)

Topics: [Kyverno](<https://devfeed.tech/topics/kyverno.md>), [software artifact signing](<https://devfeed.tech/topics/software-artifact-signing.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>), [Deployment Strategies](<https://devfeed.tech/topics/deployment-strategies.md>), [audit](<https://devfeed.tech/topics/audit.md>)

Tags: [admission](<https://devfeed.tech/tags/admission.md>), [at-t](<https://devfeed.tech/tags/at-t.md>), [attestations](<https://devfeed.tech/tags/attestations.md>), [cel](<https://devfeed.tech/tags/cel.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [image-verification](<https://devfeed.tech/tags/image-verification.md>), [kubecon](<https://devfeed.tech/tags/kubecon.md>), [kubernetes](<https://devfeed.tech/tags/kubernetes.md>), [kyverno](<https://devfeed.tech/tags/kyverno.md>), [policy-as-code](<https://devfeed.tech/tags/policy-as-code.md>), [provenance](<https://devfeed.tech/tags/provenance.md>), [rules](<https://devfeed.tech/tags/rules.md>), [sbom](<https://devfeed.tech/tags/sbom.md>), [sigstore](<https://devfeed.tech/tags/sigstore.md>), [slsa](<https://devfeed.tech/tags/slsa.md>), [supply-chain-security](<https://devfeed.tech/tags/supply-chain-security.md>)

### AI overview

AT&T uses Kyverno ImageValidatingPolicy rules in CEL to turn verified image attestations into environment-specific admission decisions: a signed image can pass in development and fail in production if it lacks required evidence. The policy verifies signatures before reading claims, checks who built the image, then applies destination requirements; promotion checks do not replace admission because requests, policies and time-sensitive evidence can differ. AT&T uses one engine for internal and approved upstream or vendor images, with explicit trust roots; support for GitHub's private Sigstore root arrived in Kyverno 1.19, and the talk recommends 1.19.2 or later for Enforce mode. A single-laptop benchmark showed fewer registry calls and lower latency in newer versions, but these are not production figures and should be measured with local policies and registries. The talk also recommends planning failure behavior and narrow, expiring exceptions; retaining per-request decision evidence; and rolling out through audit, remediation, tiered enforcement and gradual expansion.

### Source excerpt

AT&T verifies signed provenance and SBOM attestations with a Kyverno ImageValidatingPolicy in CEL, then applies each environment's rules at admission.

## NIST SP 800-190 controls for Kubernetes: what policy can enforce

DevFeed: [NIST SP 800-190 controls for Kubernetes: what policy can enforce](<https://devfeed.tech/articles/nist-sp-800-190-controls-for-kubernetes-what-policy-can-enforce-139214.md>)

Original publisher: [Read original article](<https://nirmata.com/2026/11/18/nist-sp-800-190-controls-for-kubernetes-what-policy-can-enforce/>)

Author: Ritesh Patel

Published: 2026-11-18T01:14:41.413057Z

Content type: article

Language: en

Sources: [Nirmata](<https://devfeed.tech/sources/nirmata.md>)

Topics: [Kyverno](<https://devfeed.tech/topics/kyverno.md>), [policy-as-code](<https://devfeed.tech/topics/policy-as-code.md>)

Tags: [administrator](<https://devfeed.tech/tags/administrator.md>), [admission](<https://devfeed.tech/tags/admission.md>), [admission-controller](<https://devfeed.tech/tags/admission-controller.md>), [annotation](<https://devfeed.tech/tags/annotation.md>), [annotations](<https://devfeed.tech/tags/annotations.md>), [anomaly-detection](<https://devfeed.tech/tags/anomaly-detection.md>), [attestation](<https://devfeed.tech/tags/attestation.md>), [audit](<https://devfeed.tech/tags/audit.md>), [audit-log](<https://devfeed.tech/tags/audit-log.md>), [audits](<https://devfeed.tech/tags/audits.md>), [authentication](<https://devfeed.tech/tags/authentication.md>), [compliance](<https://devfeed.tech/tags/compliance.md>), [container-images](<https://devfeed.tech/tags/container-images.md>), [container-security](<https://devfeed.tech/tags/container-security.md>), [continuous-compliance](<https://devfeed.tech/tags/continuous-compliance.md>), [controls](<https://devfeed.tech/tags/controls.md>), [kubernetes](<https://devfeed.tech/tags/kubernetes.md>), [kubernetes-security](<https://devfeed.tech/tags/kubernetes-security.md>), [kyverno](<https://devfeed.tech/tags/kyverno.md>), [kyverno-sharp-kubernetes](<https://devfeed.tech/tags/kyverno-sharp-kubernetes.md>), [nist](<https://devfeed.tech/tags/nist.md>), [pod-security](<https://devfeed.tech/tags/pod-security.md>), [policy](<https://devfeed.tech/tags/policy.md>), [root](<https://devfeed.tech/tags/root.md>), [runtime](<https://devfeed.tech/tags/runtime.md>), [scanning](<https://devfeed.tech/tags/scanning.md>), [settings](<https://devfeed.tech/tags/settings.md>), [tools](<https://devfeed.tech/tags/tools.md>)

### AI overview

SP 800-190 mentions Kubernetes only in passing and does not define Kubernetes controls; the article maps its recommendations to Kubernetes settings. Pod Security Admission's restricted profile can enforce pod configuration such as non-root execution, disallowed privileged containers and host access, and restricted capabilities and seccomp. The article recommends starting with audit or warn, reviewing results, fixing failures, then enabling enforcement; enforcement applies to resulting pods, not workload templates, and exemptions must be explicitly configured. Kyverno can add policy checks, but its shown capability policy runs in Audit mode and reports violations without blocking pods. Image vulnerability scanning, runtime anomaly detection, secret handling, and administrator identity controls need other tools or processes; signed attestations with an image-verification rule are described as an option for enforcing a pipeline result.

### Source excerpt

NIST SP 800-190 names Kubernetes only in passing. Policy can enforce its non-root, privileged and host settings. Scanning and runtime checks need other tools.

## How to stop AI coding agents from committing secrets to git

DevFeed: [How to stop AI coding agents from committing secrets to git](<https://devfeed.tech/articles/how-to-stop-ai-coding-agents-from-committing-secrets-to-git-139213.md>)

Original publisher: [Read original article](<https://nirmata.com/2026/11/18/how-to-stop-ai-coding-agents-from-committing-secrets-to-git/>)

Author: Ritesh Patel

Published: 2026-11-18T01:14:41.413057Z

Content type: tutorial

Language: en

Sources: [Nirmata](<https://devfeed.tech/sources/nirmata.md>)

Topics: [Claude Code](<https://devfeed.tech/topics/claude-code.md>), [Git](<https://devfeed.tech/topics/git.md>), [hooks](<https://devfeed.tech/topics/hooks.md>), [Secret Scanning](<https://devfeed.tech/topics/secret-scanning.md>)

Tags: [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-coding](<https://devfeed.tech/tags/ai-coding.md>), [ai-coding-agents](<https://devfeed.tech/tags/ai-coding-agents.md>), [ai-governance](<https://devfeed.tech/tags/ai-governance.md>), [coding](<https://devfeed.tech/tags/coding.md>), [commit](<https://devfeed.tech/tags/commit.md>), [git](<https://devfeed.tech/tags/git.md>), [git-hooks](<https://devfeed.tech/tags/git-hooks.md>), [github](<https://devfeed.tech/tags/github.md>), [hooks](<https://devfeed.tech/tags/hooks.md>), [how-to](<https://devfeed.tech/tags/how-to.md>), [pretooluse-hook](<https://devfeed.tech/tags/pretooluse-hook.md>), [protection](<https://devfeed.tech/tags/protection.md>), [push-protection](<https://devfeed.tech/tags/push-protection.md>), [rules](<https://devfeed.tech/tags/rules.md>), [secrets](<https://devfeed.tech/tags/secrets.md>), [secrets-management](<https://devfeed.tech/tags/secrets-management.md>), [security](<https://devfeed.tech/tags/security.md>)

### AI overview

No single layer fully prevents an AI coding agent from committing or pushing secrets. A Claude Code deny rule blocks common `git push` command forms but can miss variants; a PreToolUse hook can inspect the full command and refuse bypass flags such as `--no-verify`. A committed, executable Git pre-commit hook can inspect staged changes, though Git allows it to be skipped with that flag. GitHub push protection checks pushes server-side, including command-line pushes, but requires GitHub Secret Protection for repositories, is disabled by default, supports specific secret patterns, and permits write-access users to bypass by default. Together, these controls cover different gaps; the excerpt also notes limits involving other tools, wrapped commands, and secrets already in local history.

### Source excerpt

Block git push and commit from an AI coding agent with deny rules, a PreToolUse hook, git hooks and GitHub push protection, and see where each layer fails.

## How to set budget alerts for a runaway AI agent

DevFeed: [How to set budget alerts for a runaway AI agent](<https://devfeed.tech/articles/how-to-set-budget-alerts-for-a-runaway-ai-agent-136066.md>)

Original publisher: [Read original article](<https://nirmata.com/2026/11/19/how-to-set-budget-alerts-for-a-runaway-ai-agent/>)

Author: Ritesh Patel

Published: 2026-11-16T19:20:44.557099Z

Content type: tutorial

Language: en

Sources: [Nirmata](<https://devfeed.tech/sources/nirmata.md>)

Topics: [Claude Code](<https://devfeed.tech/topics/claude-code.md>), [AI Agent](<https://devfeed.tech/topics/ai-agent.md>), [agent observability](<https://devfeed.tech/topics/agent-observability.md>), [OpenTelemetry](<https://devfeed.tech/topics/opentelemetry.md>), [Claude](<https://devfeed.tech/topics/claude.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-governance](<https://devfeed.tech/tags/ai-governance.md>), [alerts](<https://devfeed.tech/tags/alerts.md>), [budget](<https://devfeed.tech/tags/budget.md>), [business](<https://devfeed.tech/tags/business.md>), [claude-code](<https://devfeed.tech/tags/claude-code.md>), [cost-management](<https://devfeed.tech/tags/cost-management.md>), [monitoring](<https://devfeed.tech/tags/monitoring.md>), [opentelemetry](<https://devfeed.tech/tags/opentelemetry.md>), [stop](<https://devfeed.tech/tags/stop.md>)

### AI overview

Claude Code documents no built-in budget alert or alert-triggered kill switch. To monitor spend, export the `claude_code.cost.usage` OpenTelemetry counter to your metrics backend and create per-developer alerts; use `user.account_uuid` for signed-in users, since `user.id` can split one developer across installations. Telemetry requires Claude Code v2.1.323 or later, and the default export interval is 60 seconds. For scripted print-mode runs, `--max-budget-usd` provides a cap, but spend may exceed it, resumed totals are excluded, and enforcement requires v2.1.217 or later. A self-hosted Claude apps gateway can block requests at admin-set caps; its cost estimates should be reconciled with provider usage. These controls are distinct, and none automatically stops a session in response to an alert. For immediate course correction, the excerpt says to press Escape.

### Source excerpt

Claude Code documents no budget alert or automatic stop. Contain a runaway agent with spend limits, gateway caps, --max-budget-usd and OpenTelemetry alerts.

## How to restrict egress from AI agent pods with NetworkPolicy

DevFeed: [How to restrict egress from AI agent pods with NetworkPolicy](<https://devfeed.tech/articles/how-to-restrict-egress-from-ai-agent-pods-with-networkpolicy-136065.md>)

Original publisher: [Read original article](<https://nirmata.com/2026/11/19/how-to-restrict-egress-from-ai-agent-pods-with-networkpolicy/>)

Author: Ritesh Patel

Published: 2026-11-16T19:20:44.557099Z

Content type: tutorial

Language: en

Sources: [Nirmata](<https://devfeed.tech/sources/nirmata.md>)

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

Tags: [admission](<https://devfeed.tech/tags/admission.md>), [admission-controller](<https://devfeed.tech/tags/admission-controller.md>), [agent](<https://devfeed.tech/tags/agent.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>), [api](<https://devfeed.tech/tags/api.md>), [busybox](<https://devfeed.tech/tags/busybox.md>), [cilium](<https://devfeed.tech/tags/cilium.md>), [cluster](<https://devfeed.tech/tags/cluster.md>), [cni](<https://devfeed.tech/tags/cni.md>), [container](<https://devfeed.tech/tags/container.md>), [container-networking](<https://devfeed.tech/tags/container-networking.md>), [containers](<https://devfeed.tech/tags/containers.md>), [dns](<https://devfeed.tech/tags/dns.md>), [egress](<https://devfeed.tech/tags/egress.md>), [kubernetes](<https://devfeed.tech/tags/kubernetes.md>), [kubernetes-security](<https://devfeed.tech/tags/kubernetes-security.md>), [networkpolicy](<https://devfeed.tech/tags/networkpolicy.md>), [pod](<https://devfeed.tech/tags/pod.md>), [pods](<https://devfeed.tech/tags/pods.md>)

### AI overview

To restrict AI agent pod egress, first confirm the cluster's network plugin enforces NetworkPolicy; otherwise policies have no effect. Apply a default-deny egress policy, allow DNS explicitly, then add allow rules for each needed destination and port. Rules from applicable policies combine additively, and pod-to-pod traffic must also pass the destination's ingress policy if it is isolated. NetworkPolicy cannot block traffic to the pod's node, directly target hostnames or Services by name, log blocked connections, or express explicit deny rules. Only TCP, UDP and SCTP blocking is guaranteed; host-network behavior and changes to existing connections depend on implementation. Test allowed and blocked destinations from a matching pod, and account for policy timing that can briefly leave new pods unprotected or without connectivity.

### Source excerpt

A pod stays open for egress until a NetworkPolicy selects it and lists Egress. Start with default deny, allow DNS, then add one rule for each destination.

## AI coding agent security controls that don't slow developers

DevFeed: [AI coding agent security controls that don't slow developers](<https://devfeed.tech/articles/ai-coding-agent-security-controls-that-don-t-slow-developers-135245.md>)

Original publisher: [Read original article](<https://nirmata.com/2026/11/17/guardrails-that-dont-slow-developers/>)

Author: Ritesh Patel

Published: 2026-11-16T12:00:48.928398Z

Content type: tutorial

Language: en

Sources: [Nirmata](<https://devfeed.tech/sources/nirmata.md>)

Topics: [Claude Code](<https://devfeed.tech/topics/claude-code.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-coding-agent-security](<https://devfeed.tech/tags/ai-coding-agent-security.md>), [ai-governance](<https://devfeed.tech/tags/ai-governance.md>), [claude-code](<https://devfeed.tech/tags/claude-code.md>), [coding-agent](<https://devfeed.tech/tags/coding-agent.md>), [controls](<https://devfeed.tech/tags/controls.md>), [lock](<https://devfeed.tech/tags/lock.md>), [managed](<https://devfeed.tech/tags/managed.md>), [network](<https://devfeed.tech/tags/network.md>), [policy](<https://devfeed.tech/tags/policy.md>), [prompt](<https://devfeed.tech/tags/prompt.md>), [risk](<https://devfeed.tech/tags/risk.md>), [secret](<https://devfeed.tech/tags/secret.md>), [security](<https://devfeed.tech/tags/security.md>), [security-controls](<https://devfeed.tech/tags/security-controls.md>), [settings](<https://devfeed.tech/tags/settings.md>)

### AI overview

The excerpt recommends a short Claude Code deny list for destructive commands, secret-file reads, and unapproved network tools, while letting routine reads and build work proceed without prompts. Add project rules in `.claude/settings.json`, use `ask` only for infrequent high-risk actions, and use PreToolUse hooks for command forms patterns miss; hooks can add restrictions but cannot override deny rules. Managed settings can lock policy, but `allowManagedPermissionRulesOnly` ignores user and project rules, so move the complete policy there first. The excerpt also notes project allow rules require folder trust, and suggests tracking local saved approvals and user-answered `tool_decision` events to measure friction. Kyverno validates Kubernetes objects at admission, while AIControls is described as a Claude Code gateway with budgets and PII approval, redaction, or blocking capabilities.

### Source excerpt

Deny a short list of destructive, secret and network actions in Claude Code, allow the rest, and prompt only for high-risk calls. Managed settings lock it.

## Chargeback and showback for AI coding agents

DevFeed: [Chargeback and showback for AI coding agents](<https://devfeed.tech/articles/chargeback-and-showback-for-ai-coding-agents-135244.md>)

Original publisher: [Read original article](<https://nirmata.com/2026/11/17/chargeback-and-showback-for-ai-coding-agents/>)

Author: Ritesh Patel

Published: 2026-11-16T12:00:48.928398Z

Content type: article

Language: en

Sources: [Nirmata](<https://devfeed.tech/sources/nirmata.md>)

Topics: [Finance](<https://devfeed.tech/topics/finance.md>), [Claude Code](<https://devfeed.tech/topics/claude-code.md>), [GitHub Copilot](<https://devfeed.tech/topics/github-copilot.md>), [OpenTelemetry](<https://devfeed.tech/topics/opentelemetry.md>)

Tags: [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-coding](<https://devfeed.tech/tags/ai-coding.md>), [ai-coding-agents](<https://devfeed.tech/tags/ai-coding-agents.md>), [ai-governance](<https://devfeed.tech/tags/ai-governance.md>), [amazon-bedrock](<https://devfeed.tech/tags/amazon-bedrock.md>), [budget](<https://devfeed.tech/tags/budget.md>), [business](<https://devfeed.tech/tags/business.md>), [chargeback](<https://devfeed.tech/tags/chargeback.md>), [claude-code](<https://devfeed.tech/tags/claude-code.md>), [coding-agents](<https://devfeed.tech/tags/coding-agents.md>), [copilot](<https://devfeed.tech/tags/copilot.md>), [cost](<https://devfeed.tech/tags/cost.md>), [cost-management](<https://devfeed.tech/tags/cost-management.md>), [costs](<https://devfeed.tech/tags/costs.md>), [finops](<https://devfeed.tech/tags/finops.md>), [team](<https://devfeed.tech/tags/team.md>)

### AI overview

Reliable team allocation starts by mapping each paid request to one team, then reporting spend as showback before charging it to team budgets. Each provider exposes different allocation keys: Claude Code's Analytics API reports estimated daily cost by actor, while Claude Console's API groups by workspace, API key, model and actor; for Claude Code, use the actor rather than its shared workspace. Cloud bills need per-developer assumed-role sessions or tagged, team-specific Bedrock inference profiles to identify teams. GitHub Copilot supports cost centers, with license costs assigned by users and usage-based costs by repositories or organizations. Keep a versioned roster linking identities to exactly one team, map API keys separately, and retain unmatched spend in an unallocated bucket. Reports have limits: seat allowance usage is not dollar-metered, Claude Code estimates use list pricing unless managed settings apply contracted rates, and monthly values may change for about seven days after month end. Team caps also differ: workspace limits are shared across Claude Code users, gateway group caps are per member, and Copilot budgets block usage only when the stop option is enabled.

### Source excerpt

Showback reports AI coding agent spend by team, and chargeback bills it to the team's budget. Map each tool's spend to a team key before you allocate it.

## AI agent security guide: controls for coding agents, MCP and Kubernetes agents

DevFeed: [AI agent security guide: controls for coding agents, MCP and Kubernetes agents](<https://devfeed.tech/articles/ai-agent-security-guide-controls-for-coding-agents-mcp-and-kubernetes-agents-135243.md>)

Original publisher: [Read original article](<https://nirmata.com/2026/11/17/ai-agent-security-guide/>)

Author: Ritesh Patel

Published: 2026-11-16T12:00:48.928398Z

Content type: article

Language: en

Sources: [Nirmata](<https://devfeed.tech/sources/nirmata.md>)

Topics: [AI Agent](<https://devfeed.tech/topics/ai-agent.md>), [Claude Code](<https://devfeed.tech/topics/claude-code.md>), [rbac](<https://devfeed.tech/topics/rbac.md>), [Kyverno](<https://devfeed.tech/topics/kyverno.md>), [Access Control](<https://devfeed.tech/topics/access-control.md>), [Threat monitoring](<https://devfeed.tech/topics/threat-monitoring.md>)

Tags: [admission](<https://devfeed.tech/tags/admission.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agent-security](<https://devfeed.tech/tags/agent-security.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-agent-security](<https://devfeed.tech/tags/ai-agent-security.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-governance](<https://devfeed.tech/tags/ai-governance.md>), [coding](<https://devfeed.tech/tags/coding.md>), [coding-agents](<https://devfeed.tech/tags/coding-agents.md>), [control](<https://devfeed.tech/tags/control.md>), [controls](<https://devfeed.tech/tags/controls.md>), [kubernetes](<https://devfeed.tech/tags/kubernetes.md>), [layer](<https://devfeed.tech/tags/layer.md>), [mcp](<https://devfeed.tech/tags/mcp.md>), [mcp-servers](<https://devfeed.tech/tags/mcp-servers.md>), [rbac](<https://devfeed.tech/tags/rbac.md>), [security](<https://devfeed.tech/tags/security.md>), [servers](<https://devfeed.tech/tags/servers.md>), [settings](<https://devfeed.tech/tags/settings.md>), [threat](<https://devfeed.tech/tags/threat.md>)

### AI overview

AI agent security relies on controls at three enforcement points: developer-machine coding agent settings, MCP servers and gateways, and Kubernetes admission and RBAC. On developer machines, combine permission rules, shell sandboxing, hooks, managed settings, and per-developer credentials; each covers different risks. Claude Code checks deny, then ask, then allow; a hook exit code 2 can block before those rules, but cannot remove a deny. Bash rules may miss alternate command forms, requiring hooks or sandboxing. For MCP, approve servers and tools individually, classify tools, and apply server-side input validation, access controls, rate limits, output sanitization, and logging; clients should prompt for sensitive operations and log usage. Give each agent a revocable identity with least privilege; Kubernetes RBAC has no deny rules, so begin with an empty role and grant explicitly. Admission policies such as Kyverno can reject noncompliant pods before persistence, while runtime tool calls need separate controls. Audit agent actions and policy decisions across the machine, MCP, and cluster. AIControls capabilities stated here include gateway routing for Claude Code, per-user tokens,;

### Source excerpt

Map each AI agent threat to the control that stops it and the layer that enforces it: coding agent settings, MCP servers, Kubernetes admission and RBAC.

## Pod Security Admission levels for AI agent workloads

DevFeed: [Pod Security Admission levels for AI agent workloads](<https://devfeed.tech/articles/pod-security-admission-levels-for-ai-agent-workloads-134660.md>)

Original publisher: [Read original article](<https://nirmata.com/2026/11/16/pod-security-admission-levels-for-ai-agent-workloads/>)

Author: Ritesh Patel

Published: 2026-11-16T05:03:28.562711Z

Content type: tutorial

Language: en

Sources: [Nirmata](<https://devfeed.tech/sources/nirmata.md>)

Topics: [Security](<https://devfeed.tech/topics/security.md>), [AI Agent](<https://devfeed.tech/topics/ai-agent.md>), [Kyverno](<https://devfeed.tech/topics/kyverno.md>)

Tags: [admission](<https://devfeed.tech/tags/admission.md>), [agent](<https://devfeed.tech/tags/agent.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>), [blocks](<https://devfeed.tech/tags/blocks.md>), [kubernetes](<https://devfeed.tech/tags/kubernetes.md>), [kubernetes-pod-security](<https://devfeed.tech/tags/kubernetes-pod-security.md>), [kubernetes-security](<https://devfeed.tech/tags/kubernetes-security.md>), [pod](<https://devfeed.tech/tags/pod.md>), [pod-security](<https://devfeed.tech/tags/pod-security.md>), [security](<https://devfeed.tech/tags/security.md>)

### AI overview

Pod Security Admission applies privileged, baseline, or restricted standards at the namespace level. For agent workloads, baseline blocks known privilege escalations; restricted adds pod hardening and is the stated target for agents that execute model-selected commands. Start by setting warn and audit to restricted, inspect violations, and server-side dry-run workload manifests, since namespace dry-runs only check existing pods. Fix violations, then enforce restricted and pin the tested Kubernetes version. Restricted requires settings such as non-root execution, disabled privilege escalation, a seccomp profile, and dropped capabilities; every container and init container must comply. Enforcement rejects violating pods, not their Deployment objects. These standards do not govern RBAC, service-account permissions, Secret access, or network policy, which require separate review.

### Source excerpt

Pod Security Admission sets one of three levels per namespace. Baseline blocks known escalations and restricted adds container hardening. Choose per agent pod.

## Mapping agent controls to the OWASP Top 10 for LLM apps

DevFeed: [Mapping agent controls to the OWASP Top 10 for LLM apps](<https://devfeed.tech/articles/mapping-agent-controls-to-the-owasp-top-10-for-llm-apps-134659.md>)

Original publisher: [Read original article](<https://nirmata.com/2026/11/16/mapping-agent-controls-to-owasp-llm-top-10/>)

Author: Ritesh Patel

Published: 2026-11-16T05:03:28.562711Z

Content type: article

Language: en

Sources: [Nirmata](<https://devfeed.tech/sources/nirmata.md>)

Topics: [LLM security](<https://devfeed.tech/topics/llm-security.md>), [human in the loop agent](<https://devfeed.tech/topics/human-in-the-loop-agent.md>), [ai observability](<https://devfeed.tech/topics/ai-observability.md>), [LLM observability](<https://devfeed.tech/topics/llm-observability.md>)

Tags: [2026](<https://devfeed.tech/tags/2026.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-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-governance](<https://devfeed.tech/tags/ai-governance.md>), [apps](<https://devfeed.tech/tags/apps.md>), [controls](<https://devfeed.tech/tags/controls.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [limit](<https://devfeed.tech/tags/limit.md>), [llm](<https://devfeed.tech/tags/llm.md>), [logs](<https://devfeed.tech/tags/logs.md>), [map](<https://devfeed.tech/tags/map.md>), [mapping](<https://devfeed.tech/tags/mapping.md>), [model](<https://devfeed.tech/tags/model.md>), [owasp](<https://devfeed.tech/tags/owasp.md>), [owasp-llm-top-10](<https://devfeed.tech/tags/owasp-llm-top-10.md>), [owasp-top-10](<https://devfeed.tech/tags/owasp-top-10.md>), [policy](<https://devfeed.tech/tags/policy.md>), [risks](<https://devfeed.tech/tags/risks.md>), [scope](<https://devfeed.tech/tags/scope.md>), [security](<https://devfeed.tech/tags/security.md>), [tool](<https://devfeed.tech/tags/tool.md>)

### AI overview

The post maps the OWASP LLM Top 10 2026 to controls for limiting agent risk, while excluding OWASP's separate agentic risk list. It highlights tool scoping and least privilege, authorization checks outside the model, approval for high impact or irreversible actions, hard spend and execution limits, and contextual logs. Its review checklist calls for checking tool permissions, allow decisions, approval needs, limits that halt runs, package and description changes, and adaptive attack testing. The post stresses that prompt injection has no reliable prevention mechanism, so controls should constrain what a compromised model can do; pinning tools alone does not prevent malicious payloads or poisoned descriptions. Logs can themselves expose sensitive data and should be restricted and scrubbed. It also describes four AIControls product capabilities as examples, while presenting the mapping as vendor neutral and a summary of OWASP mitigations.

### Source excerpt

Map the OWASP LLM Top 10 2026 risks to the agent controls that limit them: tool scope, policy checks outside the model, approval, spend caps and logs.

## How to detect shadow AI: unapproved model APIs and local LLMs

DevFeed: [How to detect shadow AI: unapproved model APIs and local LLMs](<https://devfeed.tech/articles/how-to-detect-shadow-ai-unapproved-model-apis-and-local-llms-134658.md>)

Original publisher: [Read original article](<https://nirmata.com/2026/11/16/how-to-detect-shadow-ai/>)

Author: Ritesh Patel

Published: 2026-11-16T05:03:28.562711Z

Content type: tutorial

Language: en

Sources: [Nirmata](<https://devfeed.tech/sources/nirmata.md>)

Topics: [shadow AI](<https://devfeed.tech/topics/shadow-ai.md>), [Local AI](<https://devfeed.tech/topics/local-ai.md>), [Ollama](<https://devfeed.tech/topics/ollama.md>), [API Governance](<https://devfeed.tech/topics/api-governance.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-governance](<https://devfeed.tech/tags/ai-governance.md>), [api](<https://devfeed.tech/tags/api.md>), [apis](<https://devfeed.tech/tags/apis.md>), [cluster](<https://devfeed.tech/tags/cluster.md>), [cluster-security](<https://devfeed.tech/tags/cluster-security.md>), [container](<https://devfeed.tech/tags/container.md>), [egress](<https://devfeed.tech/tags/egress.md>), [flag](<https://devfeed.tech/tags/flag.md>), [images](<https://devfeed.tech/tags/images.md>), [kubernetes](<https://devfeed.tech/tags/kubernetes.md>), [llms](<https://devfeed.tech/tags/llms.md>), [local](<https://devfeed.tech/tags/local.md>), [local-llms](<https://devfeed.tech/tags/local-llms.md>), [logs](<https://devfeed.tech/tags/logs.md>), [model](<https://devfeed.tech/tags/model.md>), [ollama](<https://devfeed.tech/tags/ollama.md>), [pod](<https://devfeed.tech/tags/pod.md>), [ps](<https://devfeed.tech/tags/ps.md>), [security](<https://devfeed.tech/tags/security.md>), [servers](<https://devfeed.tech/tags/servers.md>), [shadow-ai](<https://devfeed.tech/tags/shadow-ai.md>), [tags](<https://devfeed.tech/tags/tags.md>)

### AI overview

Shadow AI detection needs three complementary checks because each sees different activity: inventory Kubernetes pod and init-container images, probe reachable Ollama servers, and review controlled egress logs. The image queries identify candidate model-server images, but tags or digests must be compared with the approved list; the standard container query misses init containers. Ollama's documented GET endpoints distinguish API availability (/api/version), locally available models (/api/tags), and models loaded in memory (/api/ps), so an empty process list does not rule out an installed idle model. Egress logs can reveal hosted API calls that inventory and probes miss; findings should be grouped by source, and laptops require endpoint telemetry or network logs. Pulls can add models after image approval. A hit needs owner review: the checks show activity, not whether it was approved. The excerpt says AIControls supports specified Claude paths and per-user tokens, but does not claim it detects unapproved endpoints or local models.

### Source excerpt

Detect shadow AI in Kubernetes: scan pod images for model servers, probe Ollama's /api/tags and /api/ps, and flag unapproved model API calls in egress logs.

## Deploying an AI gateway in front of or behind your existing API gateway

DevFeed: [Deploying an AI gateway in front of or behind your existing API gateway](<https://devfeed.tech/articles/deploying-an-ai-gateway-in-front-of-or-behind-your-existing-api-gateway-134655.md>)

Original publisher: [Read original article](<https://nirmata.com/2026/11/16/deploying-an-ai-gateway-in-front-of-or-behind-your-existing-api-gateway/>)

Author: Ritesh Patel

Published: 2026-11-16T05:03:28.562711Z

Content type: tutorial

Language: en

Sources: [Nirmata](<https://devfeed.tech/sources/nirmata.md>)

Topics: [gateway](<https://devfeed.tech/topics/gateway.md>), [API Governance](<https://devfeed.tech/topics/api-governance.md>), [API Logging](<https://devfeed.tech/topics/api-logging.md>)

Tags: [2](<https://devfeed.tech/tags/2.md>), [2026](<https://devfeed.tech/tags/2026.md>), [3](<https://devfeed.tech/tags/3.md>), [403](<https://devfeed.tech/tags/403.md>), [admission](<https://devfeed.tech/tags/admission.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agent-identity](<https://devfeed.tech/tags/agent-identity.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-gateway](<https://devfeed.tech/tags/ai-gateway.md>), [ai-governance](<https://devfeed.tech/tags/ai-governance.md>), [api](<https://devfeed.tech/tags/api.md>), [api-gateway](<https://devfeed.tech/tags/api-gateway.md>), [block](<https://devfeed.tech/tags/block.md>), [egress](<https://devfeed.tech/tags/egress.md>), [either](<https://devfeed.tech/tags/either.md>), [existing](<https://devfeed.tech/tags/existing.md>), [first](<https://devfeed.tech/tags/first.md>), [gateway](<https://devfeed.tech/tags/gateway.md>), [kubernetes](<https://devfeed.tech/tags/kubernetes.md>), [kubernetes-security](<https://devfeed.tech/tags/kubernetes-security.md>), [networking](<https://devfeed.tech/tags/networking.md>), [point](<https://devfeed.tech/tags/point.md>)

### AI overview

Choose placement by deciding which hop must inspect prompts and tool arguments first: an AI gateway in front applies its checks before API gateway policies, while one behind preserves the API gateway as the entry point and adds model rules afterward. In either design, carry verifiable per-agent identity across hops and block direct agent egress to providers; otherwise agents can bypass the gateway. Kubernetes egress NetworkPolicy works only when enforced by the cluster's network plugin, and default-deny egress also needs a DNS allowance. The excerpt recommends Gateway API for new routing and calls out cross-namespace allowedRoutes, explicit body forwarding for Envoy external processing, and monitoring fail-open counters. The example configuration is not tested on a running cluster.

### Source excerpt

An AI gateway in front of your API gateway checks agent calls first. Behind it, the API gateway stays the single entry point. Either way, block direct egress.

## DORA ICT risk requirements for Kubernetes platforms

DevFeed: [DORA ICT risk requirements for Kubernetes platforms](<https://devfeed.tech/articles/dora-ict-risk-requirements-for-kubernetes-platforms-134656.md>)

Original publisher: [Read original article](<https://nirmata.com/2026/11/16/dora-ict-risk-requirements-for-kubernetes-platforms/>)

Author: Ritesh Patel

Published: 2026-11-16T05:03:28.562711Z

Content type: tutorial

Language: en

Sources: [Nirmata](<https://devfeed.tech/sources/nirmata.md>)

Topics: [Business Security](<https://devfeed.tech/topics/business-security.md>), [Kyverno](<https://devfeed.tech/topics/kyverno.md>), [Access Control](<https://devfeed.tech/topics/access-control.md>)

Tags: [17](<https://devfeed.tech/tags/17.md>), [2](<https://devfeed.tech/tags/2.md>), [2022](<https://devfeed.tech/tags/2022.md>), [2025](<https://devfeed.tech/tags/2025.md>), [access-control](<https://devfeed.tech/tags/access-control.md>), [can](<https://devfeed.tech/tags/can.md>), [change](<https://devfeed.tech/tags/change.md>), [compliance](<https://devfeed.tech/tags/compliance.md>), [continuous-compliance](<https://devfeed.tech/tags/continuous-compliance.md>), [control](<https://devfeed.tech/tags/control.md>), [dora](<https://devfeed.tech/tags/dora.md>), [encryption](<https://devfeed.tech/tags/encryption.md>), [kubernetes](<https://devfeed.tech/tags/kubernetes.md>), [kubernetes-policy](<https://devfeed.tech/tags/kubernetes-policy.md>), [kubernetes-security](<https://devfeed.tech/tags/kubernetes-security.md>), [kyverno](<https://devfeed.tech/tags/kyverno.md>), [kyverno-sharp-kubernetes](<https://devfeed.tech/tags/kyverno-sharp-kubernetes.md>), [network](<https://devfeed.tech/tags/network.md>), [network-segmentation](<https://devfeed.tech/tags/network-segmentation.md>), [platforms](<https://devfeed.tech/tags/platforms.md>), [policy](<https://devfeed.tech/tags/policy.md>), [requirements](<https://devfeed.tech/tags/requirements.md>), [results](<https://devfeed.tech/tags/results.md>), [risk](<https://devfeed.tech/tags/risk.md>), [rules](<https://devfeed.tech/tags/rules.md>)

### AI overview

DORA requires financial entities to maintain a documented, proportionate ICT risk framework; Kubernetes can enforce and report on technical controls such as access limits, network segmentation, encryption and change control, but does not satisfy governance, testing or service-register duties. The entity remains responsible, including when it uses an ICT provider. The text gives practical steps: stage Pod Security and Kyverno policies in Audit before Enforce; verify that the network plugin implements NetworkPolicy and test whether segmentation can be used in an incident; configure encryption at rest and rewrite existing Secrets; and review RBAC bindings. Kyverno reports show current cluster state, not history, so retain dated exports alongside audit logs and change records. DORA does not prescribe Kubernetes settings, and requirements vary by entity category and risk profile.

### Source excerpt

DORA's ICT risk rules cover access control, network segmentation, encryption and change control. Kubernetes policy can enforce several and record the results.

## How to control Claude Code plugins and marketplaces

DevFeed: [How to control Claude Code plugins and marketplaces](<https://devfeed.tech/articles/how-to-control-claude-code-plugins-and-marketplaces-134657.md>)

Original publisher: [Read original article](<https://nirmata.com/2026/11/16/how-to-control-claude-code-plugins-and-marketplaces-in-an-enterprise/>)

Author: Ritesh Patel

Published: 2026-11-16T05:03:28.562711Z

Content type: tutorial

Language: en

Sources: [Nirmata](<https://devfeed.tech/sources/nirmata.md>)

Topics: [Claude Code](<https://devfeed.tech/topics/claude-code.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-governance](<https://devfeed.tech/tags/ai-governance.md>), [claude](<https://devfeed.tech/tags/claude.md>), [claude-code](<https://devfeed.tech/tags/claude-code.md>), [control](<https://devfeed.tech/tags/control.md>), [governance](<https://devfeed.tech/tags/governance.md>), [how-to](<https://devfeed.tech/tags/how-to.md>), [marketplaces](<https://devfeed.tech/tags/marketplaces.md>), [plugins](<https://devfeed.tech/tags/plugins.md>), [policy](<https://devfeed.tech/tags/policy.md>), [security](<https://devfeed.tech/tags/security.md>)

### AI overview

Enterprise administrators can pin approved Claude Code plugins and restrict their sources through managed settings. `extraKnownMarketplaces` registers approved marketplaces, while managed `enabledPlugins` forces selected plugins on or blocks them with `false`; users cannot override these managed keys. `strictKnownMarketplaces` allowlists marketplace sources, `blockedMarketplaces` denies sources first, and `disableSideloadFlags` blocks local-folder and related sideload flags. Policies can arrive through server-managed settings, MDM, or a managed file; only server-managed settings reach Anthropic-hosted cloud sessions. By default, the first source supplying a policy key takes precedence, unless `managedSourcesBehavior` is set to `merge`. Repository settings require workspace trust, and contributors must install project-scoped plugins themselves. The keys do not target users or groups, and plugin auto-updates can introduce versions without review. OpenTelemetry records install and load events, but plugin names are redacted or omitted unless `OTEL_LOG_TOOL_DETAILS=1`; Enterprise Analytics also provides per-plugin counts.

### Source excerpt

Pin Claude Code plugins with enabledPlugins, allowlist marketplace sources with strictKnownMarketplaces, and block sideloaded plugins with managed settings.

## MCP security checklist for enterprises

DevFeed: [MCP security checklist for enterprises](<https://devfeed.tech/articles/mcp-security-checklist-for-enterprises-133518.md>)

Original publisher: [Read original article](<https://nirmata.com/2026/11/16/mcp-security-checklist-for-enterprises/>)

Author: Ritesh Patel

Published: 2026-11-14T03:15:26.237377Z

Content type: tutorial

Language: en

Sources: [Nirmata](<https://devfeed.tech/sources/nirmata.md>)

Topics: [Security](<https://devfeed.tech/topics/security.md>), [Access Control](<https://devfeed.tech/topics/access-control.md>), [API Logging](<https://devfeed.tech/topics/api-logging.md>), [pkce](<https://devfeed.tech/topics/pkce.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-governance](<https://devfeed.tech/tags/ai-governance.md>), [command](<https://devfeed.tech/tags/command.md>), [enterprises](<https://devfeed.tech/tags/enterprises.md>), [in-the-loop](<https://devfeed.tech/tags/in-the-loop.md>), [log](<https://devfeed.tech/tags/log.md>), [mcp](<https://devfeed.tech/tags/mcp.md>), [mcp-security](<https://devfeed.tech/tags/mcp-security.md>), [oauth](<https://devfeed.tech/tags/oauth.md>), [pin](<https://devfeed.tech/tags/pin.md>), [rules](<https://devfeed.tech/tags/rules.md>), [scopes](<https://devfeed.tech/tags/scopes.md>), [security](<https://devfeed.tech/tags/security.md>), [servers](<https://devfeed.tech/tags/servers.md>), [tool](<https://devfeed.tech/tags/tool.md>), [url](<https://devfeed.tech/tags/url.md>)

### AI overview

The checklist assigns MCP security controls across client policy and server enforcement. Inventory servers from all scopes, connectors, and plugins; approve exact URLs or commands rather than names, and verify workspace trust before relying on project approvals. For remote OAuth, require PKCE with S256 and HTTPS, pin approved scopes, and ensure servers accept only tokens issued for themselves. In Claude Code, deny rules override allow rules, so keep blocks narrow; servers can mark consent tools to prompt on every call. Keep a person able to deny invocations, treat tool annotations as untrusted, and review tool-list changes. Servers should validate inputs, enforce per-tool access, rate-limit, and sanitize outputs; clients should use timeouts, validate results, and log calls including denials. The excerpt notes gaps: allowlist behavior around upgrades, incomplete inventories from reviewing one file, and unspecified access-control design.

### Source excerpt

An enterprise MCP security checklist: approve servers by URL or command, pin OAuth scopes, set tool rules, keep a person in the loop and log every call.

## How to classify MCP tools by risk (read, edit, delete)

DevFeed: [How to classify MCP tools by risk (read, edit, delete)](<https://devfeed.tech/articles/how-to-classify-mcp-tools-by-risk-read-edit-delete-133517.md>)

Original publisher: [Read original article](<https://nirmata.com/2026/11/16/classifying-mcp-tools-by-risk/>)

Author: Ritesh Patel

Published: 2026-11-14T03:15:26.237377Z

Content type: article

Language: en

Sources: [Nirmata](<https://devfeed.tech/sources/nirmata.md>)

Topics: [MSP MCP](<https://devfeed.tech/topics/msp-mcp.md>), [CRUD](<https://devfeed.tech/topics/crud.md>), [Agent Harness](<https://devfeed.tech/topics/agent-harness.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-governance](<https://devfeed.tech/tags/ai-governance.md>), [class](<https://devfeed.tech/tags/class.md>), [delete](<https://devfeed.tech/tags/delete.md>), [how-to](<https://devfeed.tech/tags/how-to.md>), [mcp](<https://devfeed.tech/tags/mcp.md>), [mcp-tools](<https://devfeed.tech/tags/mcp-tools.md>), [risk](<https://devfeed.tech/tags/risk.md>), [security](<https://devfeed.tech/tags/security.md>), [tools](<https://devfeed.tech/tags/tools.md>)

### AI overview

Classify MCP tools by reviewed behavior: read tools change nothing, edit tools add or update state, and delete tools may destroy or overwrite it. `readOnlyHint` and `destructiveHint` help map tools to these classes, but annotations are optional, may be inaccurate, and must not drive client decisions on their own. With no hints, treat a tool as potentially destructive until checked. List the server's tools, record its hints, inspect documentation or code, and assign each tool the worst-case class. Track `openWorldHint` separately because a read tool may reach external entities. The article maps read tools to allow rules, edits to prompts, and deletes to deny by default, while emphasizing that client rules do not replace server-side access controls.

### Source excerpt

Sort MCP tools into read, edit and delete classes with the spec's readOnlyHint and destructiveHint. Set each class yourself, since hints are not guarantees.

## KubeCon + CloudNativeCon North America 2026 Recap

DevFeed: [KubeCon + CloudNativeCon North America 2026 Recap](<https://devfeed.tech/articles/kubecon-cloudnativecon-north-america-2026-recap-133515.md>)

Original publisher: [Read original article](<https://nirmata.com/2026/11/13/kubecon-cloudnativecon-north-america-2026-recap/>)

Author: Jim Bugwadia

Published: 2026-11-13T17:00:00Z

Content type: news

Language: en

Sources: [Nirmata](<https://devfeed.tech/sources/nirmata.md>)

Topics: [Kyverno](<https://devfeed.tech/topics/kyverno.md>), [policy-as-code](<https://devfeed.tech/topics/policy-as-code.md>), [ai-governance](<https://devfeed.tech/topics/ai-governance.md>), [CI/CD](<https://devfeed.tech/topics/cicd.md>), [AI-assisted coding](<https://devfeed.tech/topics/ai-assisted-coding.md>)

Tags: [2026](<https://devfeed.tech/tags/2026.md>), [agent](<https://devfeed.tech/tags/agent.md>), [agents](<https://devfeed.tech/tags/agents.md>), [ai](<https://devfeed.tech/tags/ai.md>), [ai-agent-governance](<https://devfeed.tech/tags/ai-agent-governance.md>), [ai-agents](<https://devfeed.tech/tags/ai-agents.md>), [ai-gateway](<https://devfeed.tech/tags/ai-gateway.md>), [ai-governance](<https://devfeed.tech/tags/ai-governance.md>), [at-t](<https://devfeed.tech/tags/at-t.md>), [attestation](<https://devfeed.tech/tags/attestation.md>), [attestations](<https://devfeed.tech/tags/attestations.md>), [audit](<https://devfeed.tech/tags/audit.md>), [authorization](<https://devfeed.tech/tags/authorization.md>), [cel](<https://devfeed.tech/tags/cel.md>), [cloudnativecon](<https://devfeed.tech/tags/cloudnativecon.md>), [events](<https://devfeed.tech/tags/events.md>), [kubecon](<https://devfeed.tech/tags/kubecon.md>), [kubecon-plus-cloudnativecon-north-america](<https://devfeed.tech/tags/kubecon-plus-cloudnativecon-north-america.md>), [kyverno](<https://devfeed.tech/tags/kyverno.md>), [north-america](<https://devfeed.tech/tags/north-america.md>), [policy](<https://devfeed.tech/tags/policy.md>), [policy-as-code](<https://devfeed.tech/tags/policy-as-code.md>), [production](<https://devfeed.tech/tags/production.md>)

### AI overview

AT&T uses Kyverno CEL policies to verify image signatures and attestations, then apply environment-specific admission rules: the same image can pass in development and fail in production for a missing signed SBOM or too few code reviews. A single-laptop benchmark reported registry calls per admission falling from 14 to 3; results may vary by cluster and cache state. Kinexys by J.P. Morgan uses Kyverno for namespace defaults, certificate-triggered restarts, risky RBAC restrictions, and approved Argo CD change windows, shipping policies through Git in Audit mode first. The excerpt says CEL policy types have been stable since v1.17 and Kyverno 1.20 no longer processes legacy ClusterPolicy types, making migration necessary for upgrades. Version 1.20 adds decision explanations, OCI policy packaging, and expiring exceptions. In a demo, Kyverno Authz checked each agent tool request against live Kubernetes permissions: removing a RoleBinding reduced available tools and blocked a subsequent call before it reached the backend; restoring it restored access with the same token and session.

### Source excerpt

Our KubeCon + CloudNativeCon North America 2026 recap: AT&T and Kinexys by J.P. Morgan on Kyverno in production, the move to CEL and policy for AI agents.

## How to pin an AI agent to its approved plan

DevFeed: [How to pin an AI agent to its approved plan](<https://devfeed.tech/articles/how-to-pin-an-ai-agent-to-its-approved-plan-131127.md>)

Original publisher: [Read original article](<https://nirmata.com/2026/11/13/pinning-an-agent-to-its-approved-plan-task-allowlisting/>)

Author: Ritesh Patel

Published: 2026-11-13T03:00:21.126505Z

Content type: tutorial

Language: en

Sources: [Nirmata](<https://devfeed.tech/sources/nirmata.md>)

Topics: [AI Agent](<https://devfeed.tech/topics/ai-agent.md>), [Claude Code](<https://devfeed.tech/topics/claude-code.md>)

Tags: [agent](<https://devfeed.tech/tags/agent.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-coding](<https://devfeed.tech/tags/ai-coding.md>), [ai-coding-agent](<https://devfeed.tech/tags/ai-coding-agent.md>), [ai-governance](<https://devfeed.tech/tags/ai-governance.md>), [allowed](<https://devfeed.tech/tags/allowed.md>), [allows](<https://devfeed.tech/tags/allows.md>), [argument](<https://devfeed.tech/tags/argument.md>), [arguments](<https://devfeed.tech/tags/arguments.md>), [audit](<https://devfeed.tech/tags/audit.md>), [claude](<https://devfeed.tech/tags/claude.md>), [claude-code](<https://devfeed.tech/tags/claude-code.md>), [code](<https://devfeed.tech/tags/code.md>), [how-to](<https://devfeed.tech/tags/how-to.md>), [pin](<https://devfeed.tech/tags/pin.md>), [plan](<https://devfeed.tech/tags/plan.md>), [pretooluse-hook](<https://devfeed.tech/tags/pretooluse-hook.md>), [security](<https://devfeed.tech/tags/security.md>), [tool](<https://devfeed.tech/tags/tool.md>)

### AI overview

Pinning an approved plan in Claude Code requires translating its permitted actions into exact allow rules, explicitly denying excluded calls, running execution in `dontAsk` mode, and using a `PreToolUse` hook to compare call arguments with the plan. The excerpt reports a Claude Code 2.1.321 test where a deny rule blocked `git push` despite the hook allowing it, and the hook blocked an allowed command missing from the plan. Deny rules take precedence over hook approvals; hooks can add blocks but should not be the sole hard limit because timeouts do not block calls by themselves. An allowlist bounds executable tool calls but cannot detect reasoning drift, so the article recommends retaining an audit trail. Project rules require accepting workspace trust, and `dontAsk` can deny MCP or connector tools requiring user interaction even when allowed.

### Source excerpt

Pin an AI agent to its approved plan by allowlisting exact tool calls, denying the rest and checking arguments in a PreToolUse hook. Claude Code steps included.

## SEBI CSCRF and Kubernetes: what policy can enforce

DevFeed: [SEBI CSCRF and Kubernetes: what policy can enforce](<https://devfeed.tech/articles/sebi-cscrf-and-kubernetes-what-policy-can-enforce-131128.md>)

Original publisher: [Read original article](<https://nirmata.com/2026/11/13/sebi-cscrf-and-kubernetes-what-policy-can-enforce/>)

Author: Ritesh Patel

Published: 2026-11-13T03:00:21.126505Z

Content type: article

Language: en

Sources: [Nirmata](<https://devfeed.tech/sources/nirmata.md>)

Topics: [Kyverno](<https://devfeed.tech/topics/kyverno.md>), [policy-as-code](<https://devfeed.tech/topics/policy-as-code.md>)

Tags: [admission](<https://devfeed.tech/tags/admission.md>), [admission-controller](<https://devfeed.tech/tags/admission-controller.md>), [asset-inventory](<https://devfeed.tech/tags/asset-inventory.md>), [audit-log](<https://devfeed.tech/tags/audit-log.md>), [cluster](<https://devfeed.tech/tags/cluster.md>), [compliance](<https://devfeed.tech/tags/compliance.md>), [configuration](<https://devfeed.tech/tags/configuration.md>), [container](<https://devfeed.tech/tags/container.md>), [container-security](<https://devfeed.tech/tags/container-security.md>), [containers](<https://devfeed.tech/tags/containers.md>), [continuous-compliance](<https://devfeed.tech/tags/continuous-compliance.md>), [continuous-monitoring](<https://devfeed.tech/tags/continuous-monitoring.md>), [control](<https://devfeed.tech/tags/control.md>), [controls](<https://devfeed.tech/tags/controls.md>), [cybersecurity](<https://devfeed.tech/tags/cybersecurity.md>), [hardening](<https://devfeed.tech/tags/hardening.md>), [kubernetes](<https://devfeed.tech/tags/kubernetes.md>), [kubernetes-security](<https://devfeed.tech/tags/kubernetes-security.md>), [kyverno](<https://devfeed.tech/tags/kyverno.md>), [kyverno-sharp-kubernetes](<https://devfeed.tech/tags/kyverno-sharp-kubernetes.md>), [least-privilege](<https://devfeed.tech/tags/least-privilege.md>), [logging](<https://devfeed.tech/tags/logging.md>), [mfa](<https://devfeed.tech/tags/mfa.md>), [monitoring](<https://devfeed.tech/tags/monitoring.md>), [policy](<https://devfeed.tech/tags/policy.md>)

### AI overview

The CSCRF does not name Kubernetes, but admission policies can enforce resource-level requirements such as hardened workloads, least privilege and baseline configuration. They cannot by themselves provide evidence of audit logging, MFA, access reviews, change approval or 24x7 monitoring; those require cluster configuration, identity-provider records or operating processes. The article's proposed mapping is its own, not SEBI guidance, and applicability varies across five SEBI entity categories. It highlights that Pod Security Admission `enforce` blocks violations while `audit` and `warn` do not, NetworkPolicy permits traffic by default until policies select pods, and etcd data is unencrypted unless API-server encryption is configured. A Kyverno example checks `runAsNonRoot`; CLI tests passed or rejected five test pods as expected, but no live cluster was tested. The suggested workflow is to map applicable standards and critical network segments, review policy results before enforcement, add default-deny rules where needed, and enable audit logging with a named reviewer. The 2025 clarification expands critical scope to systems sharing a network segment; the excerpt says data-locality

### Source excerpt

SEBI's CSCRF never names Kubernetes. Policy can enforce its hardening, least privilege and baseline clauses. Logging, MFA and monitoring need other controls.

[Next page](<https://devfeed.tech/sources/nirmata.md?cursor=WyIyMDI2LTExLTEzVDAzOjAwOjIxLjEyNjUwNSswMDowMCIsICIzMGY5Njg3MS04MmQ2LTQwNWYtYmFkMC00ZDhmMzZmOWQxYTkiXQ%3D%3D>)