The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes instructions for the agent to proactively offer to execute shell commands for the user.
Evidence: The 'Execution Offer' section in SKILL.md instructs the agent to offer to run git commit --file=<tmpfile> after drafting a message. It specifically directs the agent to write the message to a temporary file to prevent shell metacharacter injection in the command arguments.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user input which is then used to generate content that can be executed via the Git CLI.
Ingestion points: User-provided module name, change intent, and 'WHY' rationale in SKILL.md.
Boundary markers: The instructions lack explicit delimitation or warnings for the agent to ignore instructions embedded within the user's draft rationale.
Capability inventory: The skill utilizes the shell to perform Git commits as described in SKILL.md.
Sanitization: The skill implements a security best practice by using temporary files to avoid shell injection, but does not provide specific sanitization or filtering for the text content itself against prompt injection attempts.
[EXTERNAL_DOWNLOADS]: The skill references documentation hosted by well-known and reputable organizations.
Evidence: references/CONTRIBUTING.md and references/git_guidelines.md link to official GitHub documentation for the Odoo and OCA organizations.

odoo-commit-message-guidelines