finalize-step-pre-submission-self-review
Pass
Audited by Gen Agent Trust Hub on Jul 11, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill processes project source code for review, which introduces an indirect prompt injection surface where malicious instructions could be embedded in the code files to influence the agent's behavior.
- Ingestion points: Source code and configuration files within the project context as specified in the interface contract.
- Boundary markers: No explicit delimiters or instructions are provided in this wrapper to isolate the code being analyzed from the agent's core instructions.
- Capability inventory: The skill has access to
Bash,Read, andTasktools, allowing it to perform file system operations and execute commands based on its analysis as noted in the frontmatter. - Sanitization: There is no evidence of content sanitization or filtering for the data being ingested for review.
- [COMMAND_EXECUTION]: The workflow involves executing external logic via script notation and externalized markdown documents.
- The documentation specifies the execution of a deterministic surfacer resolved via
pm-plugin-development:ext-self-review-plan-marshall:self_review. - The orchestration logic is dispatched to an external workflow file located at
marketplace/bundles/plan-marshall/skills/phase-6-finalize/workflow/pre-submission-self-review.md.
Audit Metadata