nitro-research
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's core function is to ingest information from untrusted external sources, such as documentation websites and code repositories, which serves as an entry point for indirect prompt injection.\n
- Ingestion points: The research pipeline uses sub-agents to read external URLs and repository paths.\n
- Boundary markers: Prompts in the briefs.md file require sub-agents to provide verbatim quotes and exact locations to ensure transparency and provenance.\n
- Capability inventory: The pipeline involves file system writes to a research folder and the execution of nitro CLI commands for state tracking.\n
- Sanitization: An investigation-size pipeline includes a Verifier role that cross-checks claims against the original sources to detect inaccuracies or manipulation.\n- [COMMAND_EXECUTION]: The skill instructs the Lead agent to manage the research lifecycle using the nitro CLI. This includes creating epics and tasks, and interacting with a local fact-based memory store. These operations are integral to the agent's workflow management.\n- [EXTERNAL_DOWNLOADS]: The skill references the official chillicream.com website for the nitro CLI installation. This is a neutral reference to a vendor-provided tool, and the agent is explicitly instructed not to perform the installation itself.
Audit Metadata