skills/adryanmoldokkr32-pixel/bionicbot-giant-sovereign-skills/recursive-self-improvement/Gen Agent Trust Hub
recursive-self-improvement
Warn
Audited by Gen Agent Trust Hub on Apr 19, 2026
Risk Level: MEDIUMPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill establishes a mechanism for indirect prompt injection. The agent is directed to ingest external logic from GitHub and incorporate it into its instruction sets. 1. Ingestion points: Content from bionicbot-* repositories. 2. Boundary markers: Absent; no delimiting or instructions to disregard malicious prompts within repositories are provided. 3. Capability inventory: The agent can draft and commit changes to skill files. 4. Sanitization: No validation or filtering is applied to the logic retrieved from external sources.
- [EXTERNAL_DOWNLOADS]: The agent periodically fetches content from GitHub repositories. This behavior is documented as part of the skill's auditing function.
- [COMMAND_EXECUTION]: The skill performs git commit operations to update repository content with new instruction versions. This represents a write-capability on the agent's own configuration.
Audit Metadata