The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions and workflows for executing various command-line interface tools to perform debugging and version control operations.\n
Evidence: SKILL.md and references/debugging-tools.md detail the use of python -m pdb, node --inspect, git bisect, and dlv debug for step-through debugging and regression hunting.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted data provided by users, creating a surface for potential prompt injection via external inputs.\n
Ingestion points: The skill processes error messages, log entries, and stack traces as defined in the triggers and Core Workflow sections of SKILL.md.\n
Boundary markers: The instructions do not define delimiters or specific isolation protocols to distinguish between user-provided log data and agent instructions.\n
Capability inventory: The skill utilizes shell-based debugging tools and version control systems across SKILL.md and references/debugging-tools.md, which could be targeted by instructions embedded in processed logs.\n
Sanitization: No methods for sanitizing, filtering, or escaping the content of error logs are specified.\n- [METADATA_POISONING]: The skill contains metadata that is inconsistent with the provided author context.\n
Evidence: The author field in SKILL.md points to https://github.com/Jeffallan, which differs from the attributed vendor seltherpython.

debugging-wizard