agents-md
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill establishes a workflow where the agent analyzes untrusted repository data to define its own operational rules.\n
- Ingestion points: The agent is directed in
SKILL.mdto gather evidence fromgit log --oneline -30,package.json, and various linting/formatting configuration files (e.g.,biome.json,.eslintrc*).\n - Boundary markers: The instructions do not specify the use of delimiters or 'ignore instructions' warnings when processing these external data sources.\n
- Capability inventory: The skill's output is the
AGENTS.mdfile, which contains instructions that serve as project-wide rules influencing the agent's future logic and decision-making.\n - Sanitization: No validation or sanitization of the gathered evidence is required before it is merged into the rulebook.\n- [COMMAND_EXECUTION]: The skill utilizes local command execution to perform repository reconnaissance.\n
- Evidence: The
SKILL.mdworkflow for 'Create mode' instructs the agent to rungit logto determine established commit styles. This is a standard and benign use of development tools.
Audit Metadata