engineering
Skill17 starsChanged 3 days agoApache-2.0View on GitHub
Apply engineering judgement to approaches, problems, plans, and existing work, or audit and generate repository agent instructions from project evidence. Use for engineering reviews and requests to inspect or author AGENTS.md, conventions, or delegating CLAUDE.md. Reviews are read-only; instruction generation requires an explicit writing request. Does not implement code.
12 other projects use a version of this.
Adopted by
Most recently taken up by
- iammrduncan/skillsword for word, first seen Nov 16, 2026
- jameswyse/skillstheir own version, first seen Nov 15, 2026
- HsuanYuLee/polaristheir own version, first seen Nov 12, 2026
- wubinKKK/private-ai-workbenchtheir own version, first seen Nov 8, 2026
- wellwelwel/lagunetheir own version, first seen Oct 31, 2026
- PhoenixNest/AGENT-GLOBAL-BASEtheir own version, first seen Oct 20, 2026
Used with
Projects with it also use testing.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
NVIDIASkilltesting
Run Python and Rust tests for physicsnemo-curator using uv + pytest with coverage reporting via pytest-cov, and cargo-nextest for Rust tests. Includes benchmark workflows with pytest-benchmark and criterion.
Also in 3 projects that use this
Other versions12 projects use a version of this.
Word for word · 1
- iammrduncan/skills★ 4100% same
Edited · 0
None found.
Wrote their own version · 11
- wellwelwel/poku★ 1.2ksame name
- wellwelwel/lagune★ 216same name
- HsuanYuLee/polaris★ 5same name
- wubinKKK/private-ai-workbench★ 4same name
- eXodes/skills-workspace★ 3same name
- PhoenixNest/AGENT-GLOBAL-BASE★ 2same name
- Dhi13man/skivolve★ 1same name
- jameswyse/skills★ 1same name
- modeled-information-format/mif-docs-plugin★ 1same name
- Patasse97/claude-skills★ 0same name
1 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository6 other files it gives its agents.
- AGENTS.mdAGENTS.md
- packages/AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- naive-panelSkill
- prep-releaseSkill
- show-meSkill
The file
We split this skill into the rules it gives an agent
1 rule. 0 are common, 0 are distinctive, and 1 is very specific to this project.
0
Common
Most of the 7 edited versions we read keep these.
0
Distinctive
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 16
Review by default: Only generate-agent-instructions writes files, and only on an explicit request to create or update repository …
“Common” is measured: kept by at least 6 in 10 of the 7 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
The sidekick
A colleague you hand something to for an honest read. It reviews engineering work and derives repository instructions from evidence.
The value is in pushing back. A sidekick that agrees with you is worth nothing — but so is one that always finds something. Both are failures of the same kind: an answer decided before the work was looked at.
The rules that govern every command
Review by default. Only generate-agent-instructions writes files, and only on an explicit
request to create or update repository instructions. Every other command is read-only, including
when a correction looks obvious. Generation does not authorize code, CI, dependency, or policy changes.
Preserve scope and owner choices. Discover parent and nested instructions. Keep their scope, explicit decisions, stable rule IDs, and generated blocks intact. Report unresolved conflicts rather than choosing a new policy for the owner. Do not overwrite concurrent changes.
Say when you do not know. GOOD_ENGINEERING_H.md makes this
a principle, and it is the one most often broken by sounding confident instead. An honest "I cannot
tell without X" is worth more than a graded opinion.
Understand before judging. The ladder in reference/ladder.md runs after you understand the problem, never instead of it. A verdict on code you have not traced is noise dressed as judgement.
Setup
Resolve this installed skill's directory from the loaded SKILL.md location. Resolve its references
relative to that directory; keep the working directory at the user's target project. Load the one
command playbook needed. No source checkout or sibling skill is required for instruction work.
Commands
| Command | Purpose | Writes |
gut-check [thing] | Fast read on an approach before it is built | no |
talk-it-out [problem] | Rounds of questions until the problem is actually stated, then stop | no |
plan-it [work] | Engineering judgement over a plan; hands the shape to the objectives skill | no |
grill-it [target] | Adversarial review of something that exists. Our review and audit | no |
audit-agent-instructions [target] | Inventory instructions, verify claims, and report conflicts | no |
generate-agent-instructions [target] | Derive or update repository instructions from evidence | yes, explicit request |
Routing:
- Explicit command — load its reference and follow it.
- A bare instruction file or repository instruction target —
audit-agent-instructions. Offer generation after reporting; do not perform it. - Other bare targets —
gut-check. - Create or update repository agent instructions —
generate-agent-instructionswhen the request explicitly asks to write them. A request to review them remains an audit. - Neither command nor target — ask which work or instruction scope to examine; do not write.
- "Is this right?", "should I build this?" —
gut-check. - "I am stuck", "I cannot explain this" —
talk-it-out. - "Tear this apart", "what is wrong with this?" —
grill-it.
talk-it-out and grill-it share a questioning mechanism and point in opposite directions:
talk-it-out | grill-it | |
| Input | A fuzzy problem | Something that already exists |
| Stance | Collaborative — help you state it | Adversarial — try to break it |
| Ends when | The problem is stated | Findings are ranked and stated |
Composition
Reach for the skill that owns a job rather than restating it:
- an unrecorded decision →
write-adrs suggest - work too big to hold in one change →
write-objectives - a page that is the wrong shape →
write-docs - prose that must meet the standard →
simplified-technical-english
Reference
| File | What it is |
| reference/GOOD_ENGINEERING_H.md | The human-owned principles bundled for installed use; see provenance for maintenance |
| reference/style-sources.md | Source selection and adaptation when building conventions |
| reference/ladder.md | The seven rungs before writing code, and what they do not apply to |
| reference/gut-check.md | Fast read on an approach |
| reference/talk-it-out.md | Rounds, frontier, and when to stop |
| reference/plan-it.md | Judgement over a plan |
| reference/grill-it.md | Adversarial review |
See reference/provenance.md for inspected sources, intentional differences, and the principles-copy maintenance procedure.
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.

