The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external diagnostic data such as failure reports, stack traces, and logs to establish causal explanations.
Ingestion points: The process involves analyzing reports, stack traces, correlations, and changed artifacts provided as evidence of failure.
Boundary markers: The instructions do not specify explicit delimiters or markers to separate untrusted diagnostic data from the agent's instructions.
Capability inventory: The skill references capabilities such as git bisect for reproduction, akosile for workspace management, and html-artifact for generating visual diagnoses.
Sanitization: There are no explicit instructions for sanitizing or validating the contents of the ingested diagnostic data before analysis.
[SAFE]: The skill outlines a standard engineering workflow for debugging. It explicitly advises on safe reproduction, prefers read-only evidence, and cautions against mutating production environments or performing correction attempts within the diagnostic phase.

root-cause