mattpocock/skills: One Author Owns 94% of the Trail
As of August 22, 2026, one contributor accounts for 94% of commits in mattpocock/skills. The project is current, MIT-licensed, and has five releases, but a team adopting its prompts should pin a reviewed state and treat future editorial continuity as unverified.
The 94% is the case
There are four contributors. One authored 94% of the commits, as measured on August 22, 2026. The other three divide the remaining 6%.
For a prompt and skill collection, that is editorial concentration. One person’s judgment has shaped nearly the entire recorded body of work. The number does not reveal who reviews changes, who understands particular files, or whether a succession plan exists. That trail goes cold.
If the dominant contributor stops and the observed authorship pattern does not change, 94% of the project’s recorded commit output disappears. The data cannot identify a successor. A company should therefore own its adopted copy and its review process.
The trail is fresh
The repository was created 199 days before the August 22, 2026 check. Its last push was the same day as the check. The API returned 100 recent commits, and the weekly record includes bursts of 93, 77, and 53 commits.
So inactivity has no alibi here. The repository is being changed. That is useful if you want current material, but movement creates a second obligation: decide exactly which state your team reviewed.
Five releases exist. The latest is v1.2.3 as of August 22, 2026. Use a reviewed release or an exact commit as your internal reference. Do not make an unreviewed moving target part of a production workflow.
The costly mistake is not choosing the wrong release label. It is failing to preserve the exact material your team evaluated, then discovering later that the upstream collection has moved.
No tests is not the indictment
This is a prompt, skill, and configuration collection. It is not an importable software library. The absence of a test directory therefore means something narrower: the file tree provides no test directory that can serve as evidence of automated prompt validation.
CI configuration is present. That proves configuration exists, not that any pipeline was green on August 22, 2026. The audit did not check build status.
The repository contains 162 files. Its recorded language split is 71% Shell and 29% JavaScript. Those figures describe its shape. They do not establish whether the prompts are accurate, effective, safe, or suitable for your work. The data cannot answer that question.
Evaluate the specific prompts and skills against your own tasks before shipping them. Repository activity and popularity cannot substitute for that work.
MIT covers the repository, not every unknown
The repository identifies its license as the MIT License. It is permissive and non-copyleft. Teams reusing the material should retain the license’s copyright and permission notice.
That is not a complete legal clearance. The repository license alone does not establish the status of dependencies, bundled assets, trademarks, or patents. This audit has no evidence resolving those questions.
Popularity is equally limited evidence. The repository had 229,440 stars and 19,607 forks as of August 22, 2026. Those counts do not identify active users, downstream projects, or teams depending on it.
Also trending on August 22, 2026: mahlernim/google-timeline-visualizer, harry0703/MoneyPrinterTurbo, AprilNEA/OpenLogi, PostHog/posthog, and microsoft/TypeScript.
Questions people ask
Is mattpocock/skills actively maintained?
It was pushed on August 22, 2026, the day of the check, and the API returned 100 recent commits. That establishes current activity, not future maintenance.
Who contributes to mattpocock/skills?
Four contributors were counted as of August 22, 2026. The top contributor authored 94% of commits.
Can a company reuse mattpocock/skills?
The repository is MIT-licensed and non-copyleft. Its license does not by itself resolve the status of every dependency, asset, trademark, or patent.
Does mattpocock/skills have tests and releases?
No test directory was found. Five releases were recorded, with v1.2.3 latest as of August 22, 2026; for this kind of collection, direct evaluation of the adopted material matters more than library-style test expectations.
Verdict: adopt a snapshot, not a dependency
Use this repository if your team is prepared to review the material itself, preserve the exact state it approved, retain the MIT notice, and maintain its own copy if upstream direction changes.
Do not use it as an unquestioned external authority or assume another contributor will take over. The 94% commit share does not prove poor quality. It proves concentrated authorship. For a collection built from judgment, that distinction is the whole file.
Comments
Post a Comment