The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external code, pull requests, and diffs. This creates an attack surface for indirect prompt injection if the data being analyzed contains malicious instructions. However, the skill includes explicit instructions for the AI to define terms, use fabricated data for examples, and perform self-checks before responding, which helps mitigate accidental obedience to embedded instructions.
Ingestion points: Processes user-provided PR URLs, code snippets, and branch names (referenced in SKILL.md).
Boundary markers: The skill instructs the agent to focus on 'Brief' or 'Full' depths and provides specific formatting rules, but does not explicitly define 'ignore' delimiters for the ingested code.
Capability inventory: No tool calls, file writes, or network operations are present in the provided files.
Sanitization: Instructions in references/full-explanation.md explicitly mandate the use of fabricated values and forbid real personal data or secrets.
[NO_CODE]: The skill consists entirely of Markdown instructions and YAML configuration. There are no scripts (Python, JavaScript, etc.) or binary executables included in the skill package.

explaining-code