story-done
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from repository-resident story files to drive automated tool actions.
- Ingestion points: Markdown story files, requirement registries (tr-registry.yaml), and architectural manifests (control-manifest.md).
- Capability inventory: Utilizes Bash for test execution and git operations, Write and Edit for file updates, and Agent for delegating reviews to sub-agents.
- Boundary markers: The skill does not use delimiters or specific instructions to treat content from these data files as untrusted when constructing prompts or commands.
- Sanitization: Extracted content like test file paths and requirement IDs are passed to shell tools without explicit validation or escaping.
- [COMMAND_EXECUTION]: The skill invokes the Bash tool to perform verification and repository management tasks.
- It automatically runs test scripts using paths provided in the story file's metadata.
- It generates and suggests complex git commit commands based on the context of the completed task.
Audit Metadata