How real projects brief Codex, Cursor and every other agent that reads AGENTS.md.
OpenLAIRAGENTS.mdOpenLAIR/dr-claw/server/AGENTS.md
try/catch. - Return proper HTTP status: 400, 403, 404, 500. - Log with `console.error('[ERROR]', ...)`. ## Path security Always validate file paths: ```js const resolved = path.resolve(allowedRoot, userInput); if (!resolved.startsWith(allowedRoot + path.sep)) { return
SentryAGENTS.mdgetsentry/sentry-go/AGENTS.md
conventions - `/code-review` — Review PRs following Sentry practices - `/find-bugs` — Audit local changes for bugs and security issues
lee-toAGENTS.mdlee-to/ai-factory/AGENTS.md
test plan → test cases │ ├── aif-qa-check/ # Execute QA cases in human/automated-agent mode │ ├── aif-security-checklist/ # Security audit │ ├── aif-skill-generator/ # Generate new skills │ └── aif-verify/ # Verify implementation against plan
alibabaAGENTS.mdalibaba/skill-up/AGENTS.md
internal/runner/`, also run `make e2e`. 5. Update `CHANGELOG.md` if the change is user-visible. ## Security considerations - Credentials resolve in this priority order: CLI flag -> environment variable -> `~/.skill-up/credentials.yaml`. Never
marcusAGENTS.mdmarcus/sidecar/AGENTS.md
state until the result message arrives. This matters most on machines running an endpoint security agent (e.g. CrowdStrike), where every file open and process spawn carries a large fixed
StripeAGENTS.mdstripe/sync-engine/AGENTS.md
they create real objects). - The npm-bundled `esbuild` binary is blocked by Santa (endpoint security). Set `ESBUILD_BINARY_PATH` to the Homebrew-installed binary: `source scripts/prefer-system-esbuild.sh
openops-cloudAGENTS.mdopenops-cloud/openops/AGENTS.md
body format --- ## Tech Stack - **Backend:** Fastify, TypeORM 0.3 (PostgreSQL or SQLite), JWT auth with secure cookie sessions - **Frontend:** React 18, Zustand, react-query v5, shadcn, Axios (via `api.ts` wrapper
asklokeshAGENTS.mdasklokesh/loki-mode/references/agents.md
creation and optimization - Kubernetes manifest review - Helm chart development - Infrastructure as Code review - Container security - Multi-stage builds - Resource limits and requests **Task Types:** - `dockerfile`: Create/optimize Dockerfile - `k8s-manifest`: Write
asklokeshAGENTS.mdasklokesh/loki-mode/skills/agents.md
management **The 41 agent types are ROLES defined through prompts.** Create specialized behavior: ```python # Security reviewer = general-purpose + security-focused prompt Task( subagent_type="general-purpose", model="opus", description="Security
SentryAGENTS.mdgetsentry/sentry-cli/AGENTS.md
times - Memory usage matters for CI environments - Network operations should be optimized and retryable ### Security Best Practices - Handle authentication tokens securely - Validate file paths to prevent directory traversal - Consider impact
SentryAGENTS.mdgetsentry/sentry-cli/.github/workflows/AGENTS.md
CI/CD Workflow Guidelines ## Security: Pin Actions by Commit Hash **ALWAYS pin GitHub Actions to specific commit SHAs** instead of version tags for enhanced security. This prevents supply chain attacks where
michaelshimelesAGENTS.mdmichaelshimeles/skills/AGENTS.md
project, append what agents need to execute the beats there: commands & checks, hard invariants (security and architecture rules), an environment quick reference, local test infrastructure (stubs, fixtures), and anything that
melandlabsAGENTS.mdmelandlabs/openloomi/skills/composio/AGENTS.md
Use 1000+ external apps via Composio - either directly through the CLI or by building AI agents and apps with the SDK
stackqlAGENTS.mdstackql/stackql/AGENTS.md
when not applicable). - All required checks (build, lint, robot) must be green before merge. ## Security & Configuration Tips StackQL supports flexible configuration management, including authentication secrets and connection parameters, through environment
johannesjoAGENTS.mdjohannesjo/parallel-code/AGENTS.md
when changing exports or module dependencies. - See `.github/workflows/ci.yml` for the complete CI sequence, including security-rule fixture tests and the real-PTY coordinator test. Ordinary test runs skip
alpacahqAGENTS.mdalpacahq/alpaca-mcp-server/AGENTS.md
defined in `toolsets.py`. A trust-boundary middleware (`security.py`) wraps every tool result in a security envelope to mitigate prompt injection via tool outputs. The test suite has three layers: - `tests/test_integrity.py
Azure-SamplesAGENTS.mdAzure-Samples/AI-Gateway/AGENTS.md
Gateway pattern through Azure API Management. It focuses on managing AI services APIs with security, reliability, performance, and cost controls. Labs use Jupyter notebooks with Python, Bicep templates, and Azure
MicrosoftAGENTS.mdmicrosoft/power-platform-skills/plugins/mobile-apps/AGENTS.md
same Wrap-page + pasted-client-ID flow. - ✅ `/add-native` v0 scope: camera, location, push, biometrics, secure-store (already in template) - ✅ Cross-host Metro diagnostics: user-owned `npm run dev`, port-probe
MicrosoftAGENTS.mdmicrosoft/power-platform-skills/plugins/model-apps/AGENTS.md
privileges on the backing table that ACTIVATION creates, so they are applied in the **`security` phase**, after both the flow and the personas' roles exist — persona lookup there is **case
MicrosoftAGENTS.mdmicrosoft/power-platform-skills/plugins/power-pages/AGENTS.md
directory). Validator changes are not an exception; they must always ship with test coverage. - **Secure process validation** — After changing Power Pages `child_process` usage, run `node scripts/validate-secure-process-execution.js` from the repository
An open format for instructions to coding agents, read by Codex, Cursor and others. Think of it as a README written for agents.
At the repository root, with more specific files in subdirectories. Agents read the one closest to the file they're editing.
Setup and test commands, code style, and the rules a new contributor would need to know.
Claude Code reads CLAUDE.md. A one-line CLAUDE.md that points at AGENTS.md covers both.
