caveman-compress

Pass

Audited by Gen Agent Trust Hub on Jun 3, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes a local Python module using the command python3 -m scripts <absolute_filepath>. This allows the skill to perform arbitrary processing on files specified by the user.
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests and processes untrusted data.
  • Ingestion points: The skill reads natural language files such as CLAUDE.md, todo lists, and preferences provided via an absolute file path.
  • Boundary markers: Absent. The instructions define how to compress text but do not include delimiters or warnings to ignore potentially malicious instructions embedded within the files being processed.
  • Capability inventory: The skill can execute shell commands (python3), overwrite existing files on the system, and create/modify files in the llmwiki/ directory.
  • Sanitization: None. The file content is passed directly to the LLM for compression and validation without filtering or escaping.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 3, 2026, 04:49 AM
Security Audit — agent-trust-hub — caveman-compress