The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses bash to execute git and gh CLI commands. These are used strictly for repository metadata retrieval and commit history analysis as part of the intended functionality.
[EXTERNAL_DOWNLOADS]: The skill interacts with GitHub repositories via standard CLI tools (git, gh). This network activity is documented and matches the primary purpose of researching upstream codebases.
[PROMPT_INJECTION]: The skill processes external files such as CONTRIBUTING.md, README.md, and issue/PR content. This represents an indirect prompt injection surface where the agent could ingest untrusted instructions from external repository data.
Ingestion points: Reads local copies of README.md, CONTRIBUTING.md, and gh command outputs (SKILL.md Phase 0, Phase 3, Phase 5).
Boundary markers: None explicitly defined in instructions to delimit external content.
Capability inventory: Uses Bash for CLI operations and Write for generating reports.
Sanitization: None observed; the skill relies on standard agent processing of text files.

pr-research