Author Contribution Audit
Trace an author’s files across a branch, including contributions carried through renames.
The skill primarily uses read-oriented git log, diff-tree, and diff operations, with no malware, credential theft, covert exfiltration, or destructive default evident. It does not specify user confirmation, repository boundaries, sensitive commit-data handling, rollback, or data-flow disclosure, so substantial points are deducted.
The procedure is logically ordered and covers exact author matching, branch-wide rename tracing, merge-diff filtering, and statistics. However, there is no execution evidence; the example script does not fully match the requested +/- output and lacks robust handling for merge commits, unusual paths, command failures, and empty results, so the score remains within the static ceiling.
The audience, branch-comparison task, authorship-audit scenarios, rename handling, and table format are clearly stated. Boundaries for unsupported cases, upstream naming differences, missing local repositories, and complex Git histories are under-specified, and there is no Chinese-language or mainland-China network adaptation guidance.
The documentation has clear metadata, staged procedures, notes, and two examples. Repository-level MIT licensing, official organization provenance, and general maintenance context are available, but the skill lacks its own versioning, changelog, owner, installation/dependency notes, and troubleshooting guidance, so points are deducted.
It covers direct contributions, rename-derived contributions, files surviving into the merge diff, and line statistics, with a directly usable intended table. The example does not actually emit complete +/- statistics, complex-history correctness is unverified, and manual review remains necessary, so the score stays below the static maximum.
The documented commands and algorithm provide limited auditability and traceability. There is no skill-specific test suite, fixed fixture, CI coverage, or third-party execution evidence demonstrating reproduction of the key paths, so only limited credit is justified.
- The skill delegates many git operations to a subagent but does not define permission confirmation, timeouts, retries, or repository boundaries; confirm that commands target only the intended repository and refs.
- Rename parsing, merge commits, unusual filenames, and two-dot versus three-dot diff semantics may change results; manually review especially line statistics and VIA_RENAME classifications.
- The documentation mentions GitHub MCP get_me without specifying availability, data-access scope, or a fallback workflow when it is unavailable.
What it does & when to use it
This skill compares a branch with an upstream branch to identify files contributed by a named author that remain in the merge diff. It resolves the author’s exact Git identity, collects files touched by that author’s commits, and builds a rename map across every commit on the branch. Contributions preserved through renames are reported as VIA_RENAME rather than being lost under the new path. The skill is intended to run as a subagent and return a concise table with line statistics.
Reads Git history between the upstream and target branches; uses git log to resolve and match the author and collect the author’s commits; uses git diff-tree to extract touched files and renames from all branch commits; uses git diff to obtain the files in the final merge diff and their statistics; walks rename chains transitively; and produces a Markdown table containing DIRECT or VIA_RENAME status, file path, +/- lines, and a total summary. The instructions also recommend using Python for the heavier Git processing and mention GitHub MCP get_me as an optional way to resolve identity.
- A maintainer is preparing to merge a feature branch into main and needs an audit of one author’s files that will actually land.
- A code owner needs to distinguish files edited directly by an author from files that preserve the author’s contribution after renames.
- A team member wants to review an author’s commit scope and added or removed lines between an upstream branch and a working branch.
- A reviewer needs to exclude files that the author touched but that were later deleted and are absent from the merge result.
Pros & cons
- Covers both direct edits and contributions carried through multiple file renames.
- Reports only files still present in the upstream-to-branch merge diff, avoiding fully deleted changes.
- Provides status, path, and line additions or deletions in a format suitable for human review and downstream model processing.
- Requires exact Git identity matching, reducing the risk of confusing short usernames or similar names.
- Requires accessible Git history, a target branch, and an upstream branch; the supplied material does not describe support for non-Git repositories.
- Performs many sequential Git operations, so it is not directly suitable for a model-only API environment without shell and filesystem access.
- Reliable matching depends on the exact name and email recorded in Git commits; the identity must not be guessed.
- The supplied material documents no skill-specific test suite or platform validation results.
How to install
The supplied sources do not document a standalone installation process or installation command. The skill is located at .github/skills/author-contributions/SKILL.md in the repository, whose source code is licensed under MIT. To use it, load that directory into a compatible Agent Skills client and provide access to the Git repository being analyzed.
How to use
With the skill loaded, use a request such as: "Find every file that author \"Full Name\" contributed to on branch feature-x compared to main. Trace contributions through file renames and return the table with status, file path, and lines (+/-)." Provide the author name, target branch, and upstream branch. The skill instructs the client to run the task as a subagent; it then verifies the exact Git identity and analyzes commits, renames, and the final branch diff.