configure-feature-flags
Pass
Audited by Gen Agent Trust Hub on Apr 21, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses dynamic context injection to run
findandgrepcommands at load time for project environment discovery. These operations are benign and intended for identifying the project language and existing configuration files.\n- [EXTERNAL_DOWNLOADS]: The skill installs feature flag SDKs and providers from official package registries (NPM, PyPI) and official tool repositories (GitHub). These include standard OpenFeature implementations which are safe and expected for this functionality.\n- [PROMPT_INJECTION]: A low-severity surface for indirect prompt injection exists as the skill ingests data from untrusted project metadata files likepackage.json,pyproject.toml, anddocker-compose.yml.\n - Ingestion points: Manifest and configuration files in the project root.\n
- Boundary markers: None present.\n
- Capability inventory: The agent can execute
Bash,Write, andEditcommands, potentially allowing modification of the environment based on instructions hidden in project files.\n - Sanitization: No content validation is performed on the ingested metadata files.
Audit Metadata