caveman-help
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill functions as a static help reference. It does not contain executable code, scripts, or instructions that modify system state or bypass security guardrails.
- [EXTERNAL_DOWNLOADS]: The skill provides a link to an external GitHub repository (
github.com/JuliusBrussee/caveman) for additional documentation. No automated downloads or script executions are performed from this source. - [INDIRECT_PROMPT_INJECTION]: The skill describes a command
caveman:compress <file>which suggests the ingestion of untrusted local files. While this creates a potential surface for indirect prompt injection where malicious instructions embedded in a file could attempt to influence the agent's behavior during compression, the risk is mitigated as this file only provides the documentation for such a command rather than an implementation with elevated privileges. - Ingestion points: The
<file>parameter in thecaveman:compresscommand (SKILL.md). - Boundary markers: None mentioned in the documentation.
- Capability inventory: The skill describes formatting and display capabilities; no file-write, network, or subprocess capabilities are present in this documentation file.
- Sanitization: None mentioned in the documentation.
Audit Metadata