The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes git commit messages and diffs, which are untrusted external inputs. This creates a surface for indirect prompt injection where a malicious actor could include instructions in a commit message to influence the agent's report generation process.
Ingestion points: Git history data is retrieved by scripts/git-developer-performance.cjs and written to work-packets/*.md files.
Boundary markers: Evidence is organized into Markdown tables and headers. While the script escapes Markdown syntax, it does not include explicit instructions for the agent to ignore any natural language instructions found within the commit data.
Capability inventory: The agent is authorized to read files, run specific git commands, and write reports to the local filesystem.
Sanitization: The scripts/git-developer-performance.cjs script includes helper functions like escapeMarkdown and escapeTable to sanitize author names and commit subjects before writing them to the analysis packets.
[COMMAND_EXECUTION]: The script scripts/git-developer-performance.cjs executes the git binary to extract log data. While it uses execFileSync with a list of arguments and performs validation on branch names to mitigate shell injection risks, the ability to query the full repository history is a significant capability.
Evidence: The runGit function in scripts/git-developer-performance.cjs wraps execFileSync('git', ...) calls.

git-developer-performance