make-it-fast
Pass
Audited by Gen Agent Trust Hub on Oct 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and analyze data from sources that could potentially be influenced by an attacker, creating a surface for instructions embedded in data to influence agent behavior.
- Ingestion points: The agent reads from application logs (ActiveSupport::Notifications), SQL EXPLAIN outputs, browser network panels, the performance API, and test suite profiling rows.
- Boundary markers: The instructions lack explicit delimiters or warnings to ignore instructions that might be present in the ingested data.
- Capability inventory: The agent is empowered to modify source code ("fixes what it can measure") and execute shell commands (running the app, console runner, and specs).
- Sanitization: There are no instructions to sanitize or validate the content of logs or traces before they are processed by the agent.
- [COMMAND_EXECUTION]: The skill grants the agent permission to execute various application-level commands which could be abused if the agent is compromised by an injection.
- Evidence: SKILL.md explicitly authorizes the agent to "run specs, the app, and the app's console runner."
- Evidence: The skill dynamically references and executes logic based on a sibling directory's content ("borrow that skill's machinery" from
../make-it-right).
Audit Metadata