The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Python subprocess module to execute git and gh (GitHub CLI) commands for repository analysis. It retrieves information through standard operations such as blame, log, and gh api calls, which are necessary for its documented functionality.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data by fetching GitHub Pull Request bodies, review comments, and commit messages. This data is provided to the agent as evidence, creating a potential surface for indirect prompt injection.
Ingestion points: Untrusted data enters the agent context via gh api calls in scripts/history_context.py and scripts/compact_pr.py.
Boundary markers: SKILL.md provides instructions for assessing evidence confidence but does not utilize specific delimiters to isolate the untrusted external text.
Capability inventory: The skill scripts possess capabilities to execute shell commands and read/write files within the repository scope.
Sanitization: The scripts implement basic formatting and length truncation for retrieved text but do not include specific sanitization aimed at preventing instruction injection.
[EXTERNAL_DOWNLOADS]: The skill downloads repository metadata and history from GitHub via the authenticated gh CLI. These operations are legitimate and scoped to the repository being investigated.

investigating-repository-history