The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data originating from git log and git diff outputs to generate documentation entries. An attacker who can contribute to the repository (e.g., via a Pull Request) could embed malicious instructions within commit messages or code comments. These instructions might attempt to influence the agent's behavior during the changelog generation or translation phases.
Ingestion points: git log and git diff outputs (SKILL.md).
Boundary markers: None explicitly defined to separate untrusted git content from the agent's instructions.
Capability inventory: File system writes to multiple CHANGELOG.md and documentation files; execution of a local Node.js script (SKILL.md).
Sanitization: No explicit sanitization or filtering of the content retrieved from git history is mentioned.
[COMMAND_EXECUTION]: The workflow requires the execution of several shell commands, including git operations and a local build/sync script.
Evidence: git log main..HEAD --oneline, git diff main..HEAD --stat, and node docs/scripts/sync-changelog.mjs (SKILL.md). While these are project-local operations, they represent the execution of logic defined within the repository's file structure.

gen-changelog