The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external data such as bug reports and stack traces which represents an attack surface for indirect prompt injection.
Ingestion points: SKILL.md instructions for classifying reports and building failing inputs.
Boundary markers: The instructions do not specify the use of delimiters when processing user-provided bug descriptions.
Capability inventory: Includes shell command execution for testing and git operations.
Sanitization: No specific sanitization of external input is described.
[COMMAND_EXECUTION]: The skill guides the agent to execute repository-specific test commands and git operations to automate the debugging process.
Evidence: Instructions in SKILL.md to run test scripts from package.json, Makefile, or go test and instructions in references/bisect.md for running git bisect and management of isolated worktrees.
[DYNAMIC_EXECUTION]: The skill involves the creation and execution of new code at runtime to verify bugs and automate bisection.
Evidence: SKILL.md Step 2 (Build the smallest failing input) involves writing and running new tests, and references/bisect.md Step 1 involves creating a predicate script (predicate.sh) for git bisect.

reproduce