tack-drift
Pass
Audited by Gen Agent Trust Hub on Nov 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a local shell script (scripts/check-context.sh) to identify drift between code and quality context.\n- [DYNAMIC_EXECUTION]: The audit procedure includes a "Command check" which executes command blocks extracted from documentation files like AGENTS.md to verify their correctness.\n- [INDIRECT_PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by executing code snippets found within the project's documentation.\n
- Ingestion points: Documentation files (AGENTS.md, GLOSSARY.md) are read and parsed for executable commands.\n
- Boundary markers: The instructions do not define boundary markers to differentiate between documentation text and potentially malicious embedded instructions.\n
- Capability inventory: The agent can execute shell commands via the local audit script and documentation verification process.\n
- Sanitization: There is no evidence of sanitization or safety filtering for the command blocks before they are executed.
Audit Metadata