Skills
skills/johnefemer/skillfish/self-improving-agent-promote/Gen Agent Trust Hub

self-improving-agent-promote

Pass

Audited by Gen Agent Trust Hub on Jul 18, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes shell commands such as grep and sed to programmatically locate the project-specific memory directory within the user's home directory and search for specific patterns.
  • [PROMPT_INJECTION]: The skill provides a mechanism for indirect prompt injection by graduating unvalidated 'learnings' from auto-memory to permanent instruction files. 1. Ingestion points: The skill reads data from MEMORY.md located in the agent's project memory directory. 2. Boundary markers: The workflow lacks explicit boundary markers or instructions to ignore embedded malicious content when distilling notes into rules. 3. Capability inventory: The skill has the capability to execute shell commands (grep, sed) and perform file write operations to critical project files like CLAUDE.md and .claude/rules/. 4. Sanitization: No sanitization or filtering is performed on the content extracted from MEMORY.md before it is appended to project rules.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 18, 2026, 04:18 PM
Security Audit — agent-trust-hub — self-improving-agent-promote