The Agent Skills Directory

[PROMPT_INJECTION]: The skill ingests untrusted data from the repository's git history and code diffs to generate human-readable changelogs. While this presents a surface for indirect prompt injection, it is the primary intended function of the skill.
Ingestion points: Commit messages (git log) and source changes (git diff) are read in Steps 2 and 3 of SKILL.md.
Boundary markers: No specific delimiters are used to wrap commit data or instructions to the agent to ignore embedded commands.
Capability inventory: The skill has permissions to write manifest files and MDX documentation, as well as execute shell commands like pnpm and node.
Sanitization: No automated sanitization of commit messages is implemented prior to processing.
[COMMAND_EXECUTION]: The skill executes standard shell commands for version control and build management, including git log, git diff, git tag, and project-local scripts found in packages/cli/scripts/.
[EXTERNAL_DOWNLOADS]: The skill interacts with the official npm registry using npm view to verify package versions and uses npx to initialize test environments with vendor packages (@mindfoldhq/trellis). These operations target well-known services and the author's own official tools.

create-manifest