LoopEngineering-CrashCourse
CLAUDE.md16 starsChanged 4 months agoMITView on GitHub
This repo builds the Loop Engineering Crash Course (see loop-plan.md, the single source of truth). Two deliverables: (1) a GitHub-browsable markdown course, (2) a Next.js website in web/ that renders the same docs/ content. These rules are mandatory in every session and for every loop. 1. Single source of truth. Course content is written once in docs/ and consumed by both surfaces (GitHub + website). Never duplicate content elsewhere. 2. Follow the §10 page template for every concept page: hook…
1 other project uses a version of this.
Adopted by
Most recently taken up by
- ayeshakhalid192007-dev/LoopEngineering-CrashCourseword for word, first seen Jul 16, 2026
Other versions1 project uses a version of this.
Word for word · 1
- ayeshakhalid192007-dev/LoopEngineering-CrashCourse★ 15100% same
Edited · 0
None found.
Wrote their own version · 0
None found.
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 repository20 other files it gives its agents.
- .agents/skills/vercel-composition-patterns/AGENTS.mdAGENTS.md
- .agents/skills/vercel-optimize/AGENTS.mdAGENTS.md
- .agents/skills/vercel-react-best-practices/AGENTS.mdAGENTS.md
- .agents/skills/vercel-react-native-skills/AGENTS.mdAGENTS.md
- .agents/skills/vercel-react-view-transitions/AGENTS.mdAGENTS.md
- shared/AGENTS.mdAGENTS.md
- deploy-to-vercelSkill
- frontend-designSkill
- skill-creatorSkill
- vercel-cli-with-tokensSkill
- vercel-composition-patternsSkill
- vercel-optimizeSkill
- vercel-react-best-practicesSkill
- vercel-react-native-skillsSkill
- vercel-react-view-transitionsSkill
- web-design-guidelinesSkill
- writing-guidelinesSkill
- loop-budgetSkill
- loop-constraintsSkill
- loop-triageSkill
The file
We split this CLAUDE.md into the rules it gives an agent
7 rules. About 0 are common, 3 are distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 32
Attribution: All nine primary sources are credited in resources/sources.md.
- Distinctive: line 34
L1 first: Every loop starts in report-only mode (L1).
- Very specific: line 40
One owner per file/folder: Only the designated owner loop edits a path.
- Very specific: line 42
The spine is sacred: Update your state file every beat so an interrupted run resumes instead of restarting.
- Distinctive: line 51
Green ≠ done: Verify output (link check, template check) before declaring a checkpoint met.
- Distinctive: line 56
Read this file (automatic).
- Very specific: line 58
If running a loop: read LOOP.md, then your own loops/day1/<loop>/loop.md and state.md, then shared/goal.md.
≈ “Common” is an estimate: at least 1 in 5 of the 26,980 CLAUDE.md 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
CLAUDE.md — Project Rules (loaded every session)
This repo builds the Loop Engineering Crash Course (see loop-plan.md, the single
source of truth). Two deliverables: (1) a GitHub-browsable markdown course, (2) a Next.js
website in web/ that renders the same docs/ content. These rules are mandatory in
every session and for every loop.
Where things live
| File / folder | Role |
loop-plan.md | Master plan — the single source of truth. Never contradict it. |
days-plans/ | Per-day build plans (day1–day4). |
LOOP.md | Main loop rulebook — rules shared by ALL loops. No single loop owns it. |
loop-constraints.md | Binding constraints, read by the loop-constraints skill at the start of every run. |
loops/day1/<loop>/loop.md | One loop's own definition (prompt, stop, limit, ownership). |
loops/day1/<loop>/state.md | That loop's private spine. Only that loop writes it. |
shared/goal.md | The goal and definition of done the loops work toward. |
shared/AGENTS.md | Rules for any AI agent in this repo. |
shared/loop-budget.md | Token/run budget for the whole fleet. |
shared/loop-run-log.md | Shared run log — every loop appends one line per beat. |
STATE.md | Project-level spine (overall progress across days). |
Mandatory rules (from loop-plan.md)
- Single source of truth. Course content is written once in
docs/and consumed by both surfaces (GitHub + website). Never duplicate content elsewhere. - Follow the §10 page template for every concept page: hook → plain-English explanation → mermaid diagram → dual-tool code tabs (Claude Code ↔ OpenCode) → going-deeper callout → check-yourself quiz → Try With AI exercise → when-it-goes-wrong box → glossary popovers.
- Attribution. All nine primary sources are credited in
resources/sources.md. Adapted MIT material carries attribution. License is MIT. - L1 first. Every loop starts in report-only mode (L1). No auto-fix (L2) until a human has watched one full real run succeed.
- Minimum-safe checklist before starting any loop: provable success condition · run limit · spine (state file) written first · human gate · one log line per beat.
- Separate state files. Each loop has its own
state.md; loops never write another loop's state file. This is how loops avoid overriding each other's work. - One owner per file/folder. Only the designated owner loop edits a path; all other
loops are read-only on it (ownership map is in
LOOP.md). - The spine is sacred. Update your state file every beat so an interrupted run
resumes instead of restarting. Append one line to
shared/loop-run-log.mdevery beat. - Human gates. The human reviews
README.mdand foundations pages before a day's checkpoint is declared met. Never push, merge, or close issues/PRs without approval. - Budget discipline. Respect
shared/loop-budget.md. At 80% of the daily cap, switch to report-only. Ifloop-pause-allis active, exit immediately. - Stop conditions are specs. A loop stops on exactly three things: success condition met, run limit hit, or no progress for 3 consecutive beats. "Feels done" is not a stop.
- Green ≠ done. Verify output (link check, template check) before declaring a checkpoint met. The maker never grades its own work — use a separate checker.
Session startup order
- Read this file (automatic).
- Read
loop-constraints.md(binding). - If running a loop: read
LOOP.md, then your ownloops/day1/<loop>/loop.mdandstate.md, thenshared/goal.md.
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.