The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes standard development tools, including git log and the GitHub CLI (gh pr view, gh run list, gh run view), to collect historical data and CI/CD logs related to an incident. These operations are limited to project metadata and are consistent with the skill's primary function.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection because it processes untrusted data from user-supplied incident descriptions and external command outputs.
Ingestion points: User input provided via $ARGUMENTS in SKILL.md, as well as data retrieved from git and gh command outputs.
Boundary markers: The skill uses <incident_context> tags to encapsulate user-provided input, providing basic delimitation.
Capability inventory: The skill can execute shell commands (git, gh), call a branch creation tool (@create-branch), and write files to the docs/debriefs/ directory.
Sanitization: There is no explicit sanitization or instruction-filtering mentioned for the ingested data, but the skill's purpose is restricted to document generation rather than code execution based on that data.

debrief