The Agent Skills Directory

[SAFE]: The skill implements a comprehensive read-only policy for repository diagnostics. Hard rules in SKILL.md and references/checklist.md explicitly forbid the agent from modifying tracked files, the Git index, or refs during its operation.
[COMMAND_EXECUTION]: The skill involves executing local commands and tests to reproduce technical failures. This capability is restricted to the local environment and is governed by strict rules against permanent mutations to the target repository.
[PROMPT_INJECTION]: The skill processes external diagnostic data such as logs and stack traces, which provides a surface for indirect prompt injection. Ingestion points: Processes logs and stack traces within SKILL.md and checklist.md. Boundary markers: No explicit delimiters or ignore-instructions markers are defined for diagnostic data ingestion. Capability inventory: Local command execution for reproduction loops. Sanitization: No explicit sanitization of diagnostic output is mentioned. This surface is evaluated as safe given the restrictive execution boundaries.

diagnose