api-design
Skill0 starsChanged 46 days agoMITView on GitHub
Designing or changing an interface — library API, service endpoint, or module boundary. Contracts, input validation, idempotency, versioning, public surface.
192 other projects use a version of this.
Adopted by
Most recently taken up by
- Badhon-00/MDCPtheir own version, first seen Nov 16, 2026
- AndresArreguy/ecc-pi-porttheir own version, first seen Nov 16, 2026
- youcisla/Agent-Foundrytheir own version, first seen Nov 16, 2026
- samsam007b/claude-code-standards-kittheir own version, first seen Nov 16, 2026
- agenticluke/rest-api-designer-plustheir own version, first seen Nov 15, 2026
- dovuthanhhien1609/claude-skillstheir own version, first seen Nov 15, 2026
Used with
Projects with it also use db-design, everything-claude-code and 10 more.
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.
yuexueyuSkilldb-design
数据库工程师技能 - Schema设计、索引优化、并发安全、数据库审计。当你涉及数据库表/字段/索引/SQL/迁移/MCP数据库操作、建表改表、写查询时必须使用此技能。即使用户只是说"加个字段"或"查下数据",也应触发。
Also in 4 projects that use this
InfopibeSkilleverything-claude-code
Also in 9 projects that use this
mryeastcodeSkilleverything-claude-code-conventions
Development conventions and patterns for everything-claude-code. JavaScript project with conventional commits.
Also in 8 projects that use this
tpavanipradeepSkillnextjs-turbopack
Next.js 16+ and Turbopack — incremental bundling, FS caching, dev speed, and when to use Turbopack vs webpack.
Also in 15 projects that use this
Indium-AI-LabsSkilldata-pipeline-design
Design batch or incremental ETL and ELT pipelines and warehouse models: grain, partitioning, idempotent loads, late-arriving data, data quality tests, PII controls, and cost monitoring, before building them. Use when adding or changing a dbt model, warehouse table, scheduled transformation, or ingestion job. Not for transactional schema design (use the database-design skill) or event-sourced stores (use the event-sourcing-architecture skill).
Also in 3 projects that use this
thejordanleopoldSkillthird-party-integration
Use when integrating with external APIs, payment processors (Stripe), webhooks, or third-party services. Use when implementing API clients with authentication and rate limiting, handling webhooks with signature verification and idempotency, writing contract tests for external services, or isolating third-party dependencies behind adapters. Triggers: \"Stripe\", \"payment integration\", \"webhook\", \"third-party API\", \"external service\", \"API client\", \"integrate with\", \"signature verification\", \"idempotency key\".
Also in 3 projects that use this
Other versions192 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 192
- affaan-m/ECC★ 246ksame name
- RostislavDugin/postgresus★ 9.3ksame name
- databasus/databasus★ 9.3ksame name
- authgear/authgear-server★ 2.1ksame name
- xu-xiang/everything-claude-code-zh★ 1.9ksame name
- Aaronontheweb/dotnet-skills★ 1.2ksame name
- epam/ai-dial-chat★ 501same name
- loulanyue/awesome-claude-notes★ 272same name
- KbWen/agentic-os★ 236same name
- DevelopersGlobal/ai-agent-skills★ 216same name
182 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 repository29 other files it gives its agents.
- hosts/AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- adrSkill
- auditSkill
- code-reviewSkill
- concurrency-performanceSkill
- coverageSkill
- debuggingSkill
- debugSkill
- fast-laneSkill
- fixSkill
- implementSkill
- intakeSkill
- leanSkill
- migration-safetySkill
- nonnaSkill
- observabilitySkill
- planSkill
- refactoringSkill
- releaseSkill
- reviewSkill
- rollbackSkill
- security-reviewSkill
- shipSkill
- supply-chainSkill
- syncSkill
- tddSkill
- tdd-workflowSkill
- testSkill
The file
We split this skill into the rules it gives an agent
3 rules. 0 are common, 2 are distinctive, and 1 is very specific to this project.
0
Common
Most of the 166 edited versions we read keep these.
2
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 18
Make illegal states unrepresentable: Prefer a closed set (enum/union) over a free string.
- Distinctive: line 20
Minimize surface area: Every public symbol is a forever-promise.
- Distinctive: line 80
Write the contract test against the boundary, then implement behind it — the implementation is replaceable.
“Common” is measured: kept by at least 6 in 10 of the 166 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
API Design
An API is a contract and a promise of stability. Design the boundary so callers fall into the pit of success — and so illegal states are unrepresentable rather than merely discouraged.
Contracts: types in, types out, errors in between
The contract is the whole signature: the accepted input types, the returned type, and the errors the caller must handle. Errors are part of the contract, not an afterthought — enumerate them.
- Types carry intent.
Money(currency, minor_units)beats a barefloat;UserIdbeats astr. The type system rejects whole classes of misuse for free. - Make illegal states unrepresentable. Prefer a closed set (
enum/union) over a free string; prefer a type that can't hold a bad combination over runtime validation of one that can. - Minimize surface area. Every public symbol is a forever-promise. Expose the smallest set that serves the use case; keep the rest internal. You can add later; you can't quietly remove.
Validate untrusted input at the edge
Everything from outside — request bodies, query params, file contents, another service — is hostile
until checked. Validate at the boundary, convert to a trusted internal type, and let the core
assume validity. Reject out-of-range; never clamp-and-proceed silently. (.claude/rules/boundaries.md)
# edge: parse-and-validate into a trusted type, or reject
def create_order(raw):
qty = require_int(raw["qty"], min=1, max=10_000) # reject, don't clamp
sku = require_known_sku(raw["sku"])
return place(Order(sku=sku, qty=qty)) # core trusts the typeErrors: typed, meaningful, never swallowed
Return errors that tell the caller what to do: distinguish invalid input (caller's fault, don't
retry) from transient failure (retry with backoff) from conflict (re-read and reconcile). Use
typed errors or a result type — never a bare boolean or a null that erases the reason. Never
swallow an error to keep a signature clean; an unhandled failure that vanishes is the worst outcome.
Idempotency and retry safety for mutations
Networks retry. A mutation that runs twice must not act twice.
- Stable idempotency keys. The caller supplies a key; the server records it and returns the
original result on replay. Same key + same request → one effect. (
.claude/rules/safety.md) - Reads are safe; writes are guarded. Design
GET-shaped operations to be side-effect-free so retries are free. For writes, define what a duplicate means before shipping.
Consistency, least surprise, and bounded collections
Like things look alike: consistent naming, argument order, pagination shape, and error format across
the whole surface. Name for intent — cancel_order, not set_status; expires_at, not flag.
Any endpoint returning a list must bound its output — there is no "return all." Page with a stable
cursor (not a fragile offset), cap page size with a server-side maximum, and document the default.
Versioning — additive is safe, breaking needs a version
Additive changes are safe (new optional field, new endpoint); breaking changes need a new
version (removing/renaming a field, tightening a type, changing semantics). Default new fields so old
clients keep working; deprecate with a window and a migration path — never yank. Full policy — version
schemes, the deprecation sequence, sunset headers, and compatibility gates — in
references/versioning-and-deprecation.md.
Design the contract first, then test it, then implement
# bad: ambiguous in, lossy out, unbounded, untyped error
def transfer(a, b, amt) -> bool # which way? what currency? false = ?
# good: intent in the types, errors enumerated, idempotent, bounded
def transfer(
src: AccountId, dst: AccountId, amount: Money, idempotency_key: str,
) -> Result[Receipt, TransferError] # InsufficientFunds | AccountFrozen | DuplicateWrite the contract test against the boundary, then implement behind it — the implementation is
replaceable; the contract is not. Copy-ready starting points in templates/:
contract.test.ts (a request/response contract test asserting
validate-at-edge, typed errors, and idempotent retry) and
openapi-contract.stub.yaml (an endpoint schema with typed
money, enumerated errors, a required idempotency key, and capped pagination).
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.

