The Agent Skills Directory

[Indirect Prompt Injection] (LOW): The skill processes untrusted data from git commits and external documentation files. If these inputs contain malicious instructions, they could influence the agent during the release note generation or transformation process.\n
Ingestion points: Git commits processed by parse-commits.cjs and feature documentation accessed via the --source parameter in README.md.\n
Boundary markers: No explicit boundary markers or instruction-ignore delimiters are defined in the provided workflow.\n
Capability inventory: The skill uses Bash to execute scripts and has Read and Write access to the filesystem (e.g., docs/release-notes/).\n
Sanitization: No evidence of input sanitization or content validation is provided in the skill definitions.\n- [Command Execution] (SAFE): The skill executes local Node.js scripts (.cjs) to perform its functions. These scripts are part of the skill's library and do not represent arbitrary remote code execution.\n- [Credential Safety] (SAFE): The mention of ANTHROPIC_API_KEY in references/release-pipeline.md uses a placeholder value (your-api-key), which is a safe way to document configuration requirements.

release-notes