custom-skills-dev
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The skill uses npx to download and run the skills package from the npm registry for discovery and validation purposes.
- [EXTERNAL_DOWNLOADS]: Fetches framework source code and resource files from the vendor's official GitHub repositories and established package registries.
- [COMMAND_EXECUTION]: Employs local CLI tools such as uv for environment synchronization and python for executing repository validation scripts.
- [DATA_EXFILTRATION]: Manages and accesses tool-specific configuration files and agent skill directories located in ~/.config/ai-dev/, ~/.claude/skills/, and ~/.gemini/skills/.
- [INDIRECT_PROMPT_INJECTION]: The ai-dev framework processes external skill definitions and metadata, creating an attack surface for indirect prompt injection.
- Ingestion points: upstream/npx-skills.yaml, upstream/distribution.yaml, and skill markdown files.
- Boundary markers: Not explicitly defined in the provided architectural reference.
- Capability: The tool can perform file system operations, network requests via git and npx, and execute local scripts.
- Sanitization: Not explicitly described in the technical documentation.
Audit Metadata