The Agent Skills Directory

[PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data, such as error logs, stack traces, and CI/CD test outputs, for diagnosis. This establishes a potential surface for indirect prompt injection where malicious instructions could be embedded in the data being analyzed.
Ingestion points: The skill reads problem descriptions, logs, stack traces, and test outputs as defined in SKILL.md.
Boundary markers: There are no explicit delimiters or specific instructions for the agent to disregard instructions found within the data being analyzed.
Capability inventory: The skill is authorized to perform git status, git diff, targeted file searches, and execute 'safe test/lint commands' within the repository.
Sanitization: No specific sanitization or validation logic is provided to filter external content before it is processed by the agent.
[COMMAND_EXECUTION]: The skill uses local shell commands to perform its investigation tasks.
Capabilities: It leverages git for status and diffs, and executes 'safe test/lint commands' to verify hypotheses or validate suggested fixes.
Context: These operations are restricted to local repository analysis and are consistent with the skill's primary purpose of code investigation.

ns-investigator