self-improve-with-tinyfish

Pass

Audited by Gen Agent Trust Hub on Sep 28, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDYNAMIC_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the live web to generate new instruction sets (SKILL.md). Ingestion points: TinyFish Search and Fetch outputs (SKILL.md, Core Workflow steps 2 and 4). Boundary markers: Absent. Capability inventory: Shell command execution, Python execution, and file system modification (writing new skills). Sanitization: Absent.
  • [COMMAND_EXECUTION]: The skill uses the tinyfish CLI and Python scripts to perform search and fetch operations.
  • [DYNAMIC_EXECUTION]: The skill dynamically generates and installs new instruction files, which effectively modifies the agent's future behavior and procedural memory.
  • [EXTERNAL_DOWNLOADS]: The skill fetches documentation and search results from official vendor-owned API endpoints at api.search.tinyfish.ai and api.fetch.tinyfish.ai.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 28, 2026, 01:04 AM
Security Audit — agent-trust-hub — self-improve-with-tinyfish