odoo_commit_message_guidelines
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a shell command (
git commit) to finalize the commit process. - Evidence: The 'Response Behavior' section directs the agent to run
git commit --file=<tmpfile>after user confirmation. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input (change intent, rationale, and module names) which is then incorporated into a commit message and used in a system command execution.
- Ingestion points: User-provided inputs for 'module name(s)', 'change intent', and 'WHY rationale' defined in the 'Required Inputs' section of SKILL.md.
- Boundary markers: The skill defines a strict output structure for the commit message but does not specify delimiters for wrapping raw user input during the ingestion phase.
- Capability inventory: Shell execution of the
git commitcommand described in 'Response Behavior'. - Sanitization: The skill contains explicit instructions to avoid shell metacharacter injection by writing the message to a temporary file rather than using string interpolation with the
-mflag. - [DYNAMIC_EXECUTION]: The skill involves generating a text file (the commit message) at runtime and then executing a system command that operates on that generated content.
- Evidence: Instruction in 'Response Behavior' to "write the full commit message to a temporary file and run git commit --file= (then delete the temp file)".
Audit Metadata