The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill analyzes git diffs and commit logs, which are external data sources that could theoretically contain malicious instructions designed to influence the agent's behavior. However, this risk is mitigated by the skill's specific focus on generating commit messages and the lack of high-privilege capabilities beyond the repository context.
Ingestion points: The skill reads untrusted data from the repository using git diff, git diff --staged, and git log commands.
Capability inventory: The skill uses the Bash, Read, Grep, and Glob tools to interact with the filesystem and execute git commands.
Boundary markers: The instructions focus on analyzing intent and conventions but do not define explicit delimiters for separating code content from instructions.
Sanitization: No explicit sanitization or filtering of the incoming diff content is performed before analysis.
[COMMAND_EXECUTION]: The skill is designed to execute specific git commands (git commit, git apply, git status) via the Bash tool to fulfill user requests. These operations are within the expected scope of a git management utility.
Evidence: The reference.md and SKILL.md files contain predefined bash command sequences for status reporting, diffing, and applying patches for multi-concern changes.

writing-commits