tack-eval
Pass
Audited by Gen Agent Trust Hub on Nov 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the AI agent to execute shell commands extracted from the repository's documentation (
AGENTS.md) and to run local maintenance scripts, specificallyscripts/check-context.sh, to verify the repository's state. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted repository data to drive agent actions, creating an attack surface where instructions embedded in documentation could manipulate agent behavior.
- Ingestion points: Repository files used as context, including
AGENTS.md,GLOSSARY.md, and Architecture Decision Records (ADRs) found in thedocsandreferencesdirectories. - Boundary markers: The instructions lack explicit delimiters or instructions for the agent to ignore potentially malicious embedded commands within the analyzed documentation.
- Capability inventory: The agent is directed to use shell execution for installation, building, testing, and running repository-specific scripts.
- Sanitization: No sanitization or validation of the repository's text content is performed before it is used to influence the agent's task execution.
Audit Metadata