rules-from-coding-agent-failures

Pass

Audited by Gen Agent Trust Hub on Jul 26, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructions specify the use of shell commands, primarily grep, to analyze metadata within failure logs stored in docs/reflection-log/. It also incorporates a test script evals/run-static-checks.sh which executes a local Python script for skill validation and consistency checking.
  • [PROMPT_INJECTION]: The skill implements a workflow that reads user-generated or agent-generated log entries and translates them into directive content like instructions or executable git hooks, creating an indirect prompt injection surface. Ingestion points: Processes markdown files in docs/reflection-log/[0-9]*.md. Boundary markers: No explicit delimiters are used to wrap or isolate the content of the logs when the agent triages or promotes them. Capability inventory: The skill is authorized to modify core instruction files (README.md, AGENTS.md), create shell-based hooks, and update CI gate configurations. Sanitization: There is no mention of sanitizing or escaping the content retrieved from logs before it is interpolated into rules or hooks.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 26, 2026, 09:31 PM
Security Audit — agent-trust-hub — rules-from-coding-agent-failures