gh-voice
Skill0 starsChanged 42 days agoApache-2.0View on GitHub
Write and edit PR descriptions, issues, comments, and review replies in the concise, factual house style of osdu-spi-stack.
In this repository6 other files it gives its agents.
- AGENTS.mdAGENTS.md
- .github/copilot-instructions.mdCopilot instructions
- code-reviewSkill
- osdu-apiSkill
- primeSkill
- resolve-review-threadsSkill
The file
We split this skill into the rules it gives an agent
7 rules. About 0 are common, 7 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈7
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 15
Lead with the result: Start with the change, observed problem, or review verdict.
- Distinctive: line 22
Keep precise technical terms.
- Distinctive: line 24
Be direct and respectful: Remove filler ("worth noting"), marketing language ("seamless"), and self-evaluation ("the honest fix").
- Distinctive: line 28
Use headings that name the subject.
- Distinctive: line 37
When declining a suggestion, give the technical reason.
- Distinctive: line 75
Remove repetition, investigation chronology, and empty sections.
- Distinctive: line 77
Check that essential rationale, evidence, uncertainty, and literal text survived the edit.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Voice for PRs, Issues, and Comments
This skill governs wording, not structure. PR descriptions follow
CONTRIBUTING.md under "PR Descriptions" and the template in
.github/PULL_REQUEST_TEMPLATE.md. Issues follow the forms in
.github/ISSUE_TEMPLATE/.
Principles
- Lead with the result. Start with the change, observed problem, or review verdict. Follow with the reason and supporting evidence.
- Keep essential rationale. Explain unusual choices, trade-offs, and limitations a reader needs to understand the change. Reproduction steps and results that substantiate a verdict are evidence, not process narration. Omit investigation chronology and unrelated follow-ups.
- Preserve meaning. Simplify prose without adding or changing claims. Keep precise technical terms. State uncertainty explicitly rather than presenting unverified behavior as fact.
- Be direct and respectful. Remove filler ("worth noting"), marketing language ("seamless"), and self-evaluation ("the honest fix"). Omit generic praise and acknowledgments that add no information. Contractions are fine.
- Use only useful structure. Group changes by concern, not by file. Use headings that name the subject; remove empty sections and repeated explanations. No generic validation checklists; mention checks when the evidence or a validation limit matters.
- Preserve literal text. Replace em and en dashes in prose with commas, periods, colons, or parentheses. Leave commands, identifiers, URLs, and quoted evidence unchanged.
A review comment names the defect and consequence, then the fix when clear. A reply states the verdict, evidence, and relevant commit or line link. When declining a suggestion, give the technical reason. One to three sentences usually suffice; include more when the evidence needs it.
Cut on sight
- Praise and acknowledgment. "Great point", "you're absolutely right", "thanks for the catch", "just to confirm my understanding".
- Soft openers. "Consider", "it might be worth", "worth noting". Name the defect and its consequence instead.
- Empty status. "Fixed in the latest push" says nothing. Say what the
code does now and where: "
upload-database: falsenow, sinceupload: neveronly covers SARIF. d3cd529." - Reader stage-direction. "The pin is the part to argue with", "worth opening that log". Spend the words on the risky thing itself.
- Process narration. "Preflight caught four defects", "addresses review feedback from round 2". State the resulting design.
- Author caveats. "I could not verify deployment" becomes "Deployment is unverified because the test subscription was unavailable."
Two tells that read as generated even with plain words: grading the fix ("honest", "principled", "proper", "clean") and calling a check a "signal". Say what the fix does and what the check checks.
Words
| inflated | plain |
| the canonical X | the template owns X |
| degraded to a clean skip | skipped silently |
| a destructive deletion path | that script deletes registry tags |
| the honest fix is to X | X |
| a failure there is quiet | it logs the error instead of failing the check |
Also avoid "leverage", "robust", "seamless", "comprehensive", "holistic", "delve", and "ensure" where "make sure" reads better.
Cleanup pass
- Remove repetition, investigation chronology, and empty sections.
- Simplify wording without changing technical meaning.
- Check that essential rationale, evidence, uncertainty, and literal text survived the edit.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

