detect
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data by scanning files within the user's repository to identify the project stack and existing integrations.
- Ingestion points: The agent is instructed to read various manifest files (e.g., package.json, requirements.txt, go.mod), configuration files (e.g., wrangler.toml, vercel.json), and application source code to detect entrypoints and existing LaunchDarkly usage (SKILL.md, sections 1, 2, 4, and 6).
- Boundary markers: No explicit boundary markers or instructions to ignore embedded commands are specified for the external data being read into the context.
- Capability inventory: The skill performs file search and read operations across the repository. It has the capability to write to the file system (creating a demo app in a 'launchdarkly-demo/' subfolder) and invokes a structured question tool based on its findings.
- Sanitization: The instructions do not prescribe sanitization or validation of the data read from repository files before it influences the agent's decision tree or output.
Audit Metadata