The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes a shell script, find-polluter.sh, designed to identify which test file in a suite is creating unwanted state or files. It uses find to discover test files and executes npm test on each discovered path. Additionally, the skill's instructions provide examples of diagnostic commands (e.g., git, security, codesign) that an agent may use to investigate environmental failures or configuration issues.
[INDIRECT_PROMPT_INJECTION]: The systematic debugging process requires the agent to ingest and analyze external data such as error messages, stack traces, and log files. While these sources represent an untrusted data surface, the skill treats them as diagnostic evidence for root cause analysis rather than instructions to be followed.
Ingestion points: Technical logs, error messages, and stack traces provided during a debugging session (SKILL.md, root-cause-tracing.md).
Boundary markers: Not explicitly defined within the methodology, as the skill assumes the data is provided by the user within a debugging context.
Capability inventory: Access to shell command execution and file system operations as described in the debugging examples and utility scripts.
Sanitization: The skill focus is on technical forensics, and it does not explicitly describe automated sanitization methods for ingested logs.

systematic-debugging