The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the git and gh (GitHub CLI) commands to fetch repository metadata, commit histories, and release notes. These operations are restricted to read-only information gathering and are consistent with the skill's primary function.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external sources (GitHub release bodies and git commit messages) to generate summaries. While this represents a theoretical attack surface for indirect prompt injection, the impact is confined to the project's documentation files.
Ingestion points: Output from gh release view and git log commands.
Boundary markers: The skill instructs the agent to follow a specific local style and structure, which acts as a natural constraint, though no formal delimiters for the external content are specified.
Capability inventory: The skill writes to docs/CHANGELOG.md but does not possess high-risk capabilities like network exfiltration of local secrets or system-level persistence.
Sanitization: The skill relies on the agent's distilling and writing instructions to filter 'noisy' content, which serves as a basic form of data sanitization.

update-changelog