mreq-apply
Pass
Audited by Gen Agent Trust Hub on Nov 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted task descriptions from documents in the .mreq/ directory, creating a surface for indirect prompt injection attacks where malicious instructions could be embedded in project files.
- Ingestion points: Step 2 of SKILL.md instructs the agent to read .mreq/ documents in full, including goal, tasks, and handoff sections which may be controlled by an attacker.
- Boundary markers: No explicit markers are used to separate untrusted document content from the agent's internal instructions.
- Capability inventory: The agent can execute arbitrary shell commands for testing, perform file writes, and invoke a local Python script.
- Sanitization: Content from the .mreq/ files is used directly to determine the agent's next actions without validation or filtering.
- [COMMAND_EXECUTION]: The skill instructions authorize the agent to execute shell commands to verify work, which can be exploited if the verification steps are influenced by malicious input.
- Evidence: Step 4.3 ("Prove it works by running something...") explicitly directs the agent to execute commands based on the task context.
- [COMMAND_EXECUTION]: The skill invokes a local script (mreq.py) using shell tools like uv run, python, or python3.
- Evidence: The "The script" section defines the execution of a Python script located at a path relative to the skill directory.
Audit Metadata