The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses git log and git show to read repository history. This is expected behavior for generating release notes.
[PROMPT_INJECTION]: The skill processes untrusted commit messages (git log output) which could contain malicious instructions designed to influence the agent's output (Indirect Prompt Injection).
- Ingestion points: The skill ingests commit history via git log as described in SKILL.md and reference/notes.md.
- Boundary markers: The instructions lack explicit boundary markers or delimiters to separate the raw commit data from the agent's operational instructions.
- Capability inventory: The skill's capabilities are limited to reading history and generating text output; it does not request network access or file system write permissions, which significantly limits the risk impact.
- Sanitization: While the skill is instructed to produce concise bullets and drop 'noise' commits, there is no explicit instruction to sanitize or escape the content of commit bodies before interpolation into the final output.

release