The Agent Skills Directory

[SAFE]: The skill provides standard documentation and procedures for instrumenting applications with feature flags using official PostHog resources.
[CREDENTIALS_UNSAFE]: The instructions explicitly guide the user to store API tokens in environment variables (.env files) and retrieve them via authenticated MCP tools, avoiding hardcoding sensitive keys.
Evidence: STEP 6 in SKILL.md instructs the agent to write keys to environment files using framework naming conventions.
[INDIRECT_PROMPT_INJECTION]: The skill analyzes local codebase metadata to determine the environment, representing a low-risk attack surface.
Ingestion points: Dependency files like package.json and requirements.txt analyzed in SKILL.md STEP 1.
Boundary markers: None specified, but usage is limited to platform detection logic.
Capability inventory: The agent possesses file-write capabilities and access to PostHog MCP tools.
Sanitization: Not explicitly implemented, but findings are used only to select appropriate integration templates.
[EXTERNAL_DOWNLOADS]: The skill references official PostHog SDKs and dependencies from established registries and the vendor's verified GitHub repositories.
Evidence: SDK installation commands in platform reference files (e.g., references/react.md, references/python.md) target official packages like posthog-js and posthog.

instrument-feature-flags