The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted data from external sources to update local documentation.
Ingestion points: External data is ingested from GitHub releases via the gh CLI and from git logs/diffs as described in the Gather evidence section of SKILL.md.
Boundary markers: There are no explicit markers or instructions to treat external data as untrusted or to ignore any instructions embedded within commit messages or release bodies.
Capability inventory: The skill performs filesystem writes to docs/CHANGELOG.md and executes shell commands using git and gh.
Sanitization: The skill does not implement sanitization or validation of the external content; it only instructs the agent to distill the information for brevity.
[COMMAND_EXECUTION]: The skill executes local shell commands to perform repository operations.
Utilizes git log and git diff to extract commit history and code changes.
Utilizes the GitHub CLI (gh) to retrieve release metadata.
[EXTERNAL_DOWNLOADS]: The skill fetches data from GitHub's official services.
Uses the GitHub CLI to view release details from GitHub repositories, which involves network communication with a well-known service.

update-changelog