architecture-principles
Skill78 starsChanged 5 months agoMITView on GitHub
Architecture principles for manifest-dev: how its maintainer wants skills, wrappers, references, state and the workflow structured, and why. Use before planning, designing, building or reviewing a new skill or entry point, a change to where state or a decision lives, how skills call or coordinate with each other, or a wait or loop a workflow can enter.
4 other projects use a version of this.
Adopted by
Most recently taken up by
- bolgovr/skillstheir own version, first seen Oct 31, 2026
- Khuirul-Huda/clean-principlestheir own version, first seen Oct 25, 2026
- vospr/context-engineering-templatetheir own version, first seen Oct 1, 2026
- Mesteriis/Engineering-Bible-AItheir own version, first seen Jun 28, 2026
Other versions4 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 4
- vospr/context-engineering-template★ 9same name
- bolgovr/skills★ 1same name
- Khuirul-Huda/clean-principles★ 0same name
- Mesteriis/Engineering-Bible-AI★ 0same name
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 repository10 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- bump-surfaces-versionSkill
- design-principlesSkill
- harden-task-fileSkill
- learn-from-sessionSkill
- mirror-pairsSkill
- product-principlesSkill
- prompt-principlesSkill
- sync-claude-code-pluginsSkill
- sync-toolsSkill
The file
We split this skill into the rules it gives an agent
1 rule. 0 are common, 1 is distinctive, and 0 are very specific to this project.
0
Common
Most of the 4 edited versions we read keep these.
1
Distinctive
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
Before calling architecture work done, check it against these principles from a fresh start — a new session, or a reviewer who did not do …
“Common” is measured: kept by at least 6 in 10 of the 4 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
Edit rule: each line is a principle with its reason and when it yields. Decisions go to the decision records, rules a check can hold go to the check, and a line is added only on a person's yes, after sharpening an existing line is ruled out.
- A new need lands on a mechanism, concept and state owner that already exist — figure-out, define, do, task files, the logs, the Ticket — before it earns a new skill, setting, file, ledger or surface, and a named entry is at most a thin wrapper that routes and restates no rule — a second loop or record is a second place to drift and disagree, every added concept is one more thing each user learns, and a rule in a wrapper reaches only runs that start there. Yields when no existing owner can carry it without changing what it means, or when people reach for a name: then add only the wrapper.
- Ship skills as the protocol, never the platform: no scheduler, hosting or host-bound runner; start from what the user already runs, use a host capability where it is offered and a plain path where it is not — infrastructure ties the project to one host and adds complexity each user carries. Yields to a coordinating skill, started by hand or by the user's own scheduler, when the alternative leaves users to build the machinery themselves.
- A skill that inspects or reports stays blind to its caller's workflow and returns what it saw with hints that name an action on the thing inspected, never a step of the caller's workflow; each decision lives once, in the skill that holds the context to make it — the caller steers toward the hint, a fixed set alone cannot cover the unexpected, and a rule restated per caller drifts. No yield has been stated beyond the verdict itself, which stays fixed.
- Every wait or loop a workflow can enter has a stated exit — done, a bounded retry, or a named person — reading the prose missed these: they surfaced as a landing that stalled for good and an escalation that fired every tick. No yield has been stated.
Before calling architecture work done, check it against these principles from a fresh start — a new session, or a reviewer who did not do the work. Fix what departs unless the person chose the departure, and check again only after a fix, until a check finds none. Name each departure kept, with the principle and why.
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.

