find-the-loop
Pass
Audited by Gen Agent Trust Hub on Nov 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and process potentially untrusted external data, such as build logs, performance telemetry, and CI history, to establish measurement baselines. \n
- Ingestion points:
SKILL.md(Step 3, Step 4) anddecision-tree.md(UI Performance, CI Speed, Prompt Behavior) identify telemetry dashboards, E2E captures, and raw-source queries as data sources.\n - Boundary markers: The skill does not explicitly define delimiters for external data, though it emphasizes pre-registering metrics to prevent goal-post shifting.\n
- Capability inventory: The agent is encouraged to run local scripts, tests, and CLI tools identified in
package.json,Makefile, andscripts/to act as "instruments".\n - Sanitization: No specific sanitization or validation of the input data stream is prescribed, relying on the agent's internal logic to interpret results.\n- [NO_CODE]: The skill consists entirely of Markdown instructions and decision trees without providing any executable scripts or binaries of its own.
Audit Metadata