agentleFS
pahlevikunpahlevikun/levistack/skills/architecture/super-protobuf/SKILL.md

super-protobuf

Skill0 starsChanged 19 days agoMITView on GitHub

Protocol Buffers (proto3) design, buf.yaml/buf.gen.yaml, Connect and gRPC services, protovalidate, lint and breaking checks, schema evolution, mixed-version compatibility review, and defensive gRPC hardening. Use when writing or reviewing .proto files, setting up buf, generating Connect or gRPC code, adding protovalidate, catching breaking changes, or hardening an exposed gRPC surface — not for REST-only OpenAPI, GraphQL schemas, or offensive testing.

The file

We split this skill 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
  1. Distinctive:

    Do this first; do not announce it.

    line 48
  2. Distinctive:

    Review-compat: when consumers, storage, or mixed versions exist.

    line 70
  3. Very specific:

    Check the Makefile or package.json first—many repos wrap these as make lint / make generate.

    line 72
  4. Distinctive:

    Keep this router lean. Do not paste language samples or extra checklists here.

    line 76
  5. Very specific:

    Language track: add references/tech/<lang>.md (one screen), one row in references/tech/README.md, and one row in the Step 0 table.

    line 78
  6. Very specific:

    Template: add a file under templates/ or assets/ and mention its basename in templates/README.md.

    line 79
  7. Very specific:

    Core topic: add references/core/<topic>.md and one table row here.

    line 80

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Super protobuf

One router for proto3 schemas, the buf toolchain, Connect/gRPC services, and compatibility review. Language-agnostic core; language tracks are optional and short.

Merged scope: formerly protobuf. Index: references/README.md. Research: research-synthesis.md.

Principles

  1. Match the repo first. Copy existing package layout, field order, validation, and comments. If none exist, use this skill's defaults.
  2. Buf is the default toolchain. buf format -w && buf lint, then buf breaking --against '.git#branch=main' when the schema already shipped. EasyP is a buf-compatible alternative, not a second source of truth.
  3. Protovalidate on production fields. The .proto file is the contract for structure and constraints.
  4. Evolve additively. New fields, enum values, and RPCs stay in v1. Breaking changes get a new package version. Never reuse field or enum numbers.
  5. Treat gRPC as an API. Authz on every method, including "internal" services that just became reachable. Do not leak .proto files or descriptors. Do not run or document attacks.

Step 0: detect the job

Do this first; do not announce it. If the request already names a mode, go there.

If the user is doing thisModeLoad
New messages, RPCs, package layout, AIP-style resourcesdesignmodes/design.md, design.md
buf.yaml / buf.gen.yaml, generate stubs, Connect vs gRPCgeneratemodes/generate.md, buf-workflow.md
Review a proto diff, mixed-version deploys, wire vs JSON vs storagereview-compatmodes/review-compat.md, compatibility.md
buf lint / buf breaking, naming, EasyP as alternativelint-breakingmodes/lint-breaking.md, style.md
Auth on gRPC, reflection, grpc-web, leaked schemashardensecurity.md
Go plugins, opaque APIlanguage: Gogo.md
Elixir / protobuf-elixirlanguage: Elixirelixir.md
C# / grpc-dotnetlanguage: C#csharp.md
Connect vs gRPC choicerpc stackconnect-vs-grpc.md

Stack index: references/tech/README.md. Depth: evolution.md, protovalidate.md, review-checklist.md, quick-reference.md, troubleshooting.md, migration.md, protoc-toolchain.md.

Default pipeline

  1. Detect the job (table above).
  2. Design the messages and RPCs. Copy templates/ or assets/proto/example/v1/.
  3. Generate with the project's buf.gen.yaml (templates in assets/).
  4. Verify: buf format -w && buf lint. If the schema shipped, buf breaking --against '.git#branch=main'.
  5. Review-compat when consumers, storage, or mixed versions exist.

Check the Makefile or package.json first—many repos wrap these as make lint / make generate.

Contribute

Keep this router lean. Do not paste language samples or extra checklists here.

  • Language track: add references/tech/<lang>.md (one screen), one row in references/tech/README.md, and one row in the Step 0 table.
  • Template: add a file under templates/ or assets/ and mention its basename in templates/README.md.
  • Core topic: add references/core/<topic>.md and one table row here. Prefer extending an existing file.

Old skill name

OldUse
protobufThis skill (super-protobuf)

Related skills

  • super-elixir: the service is Elixir.
  • super-nodejs: the service is Node and uses Connect or gRPC.
  • super-architecture: decide service boundaries before you design the schema.
  • verify: show buf lint and breaking output before you claim compatibility.
  • create-bruno-collection: add runnable gRPC examples as a Bruno collection.

Done when

  • The matching mode ran; only needed references were opened.
  • buf format -w && buf lint (or the project's wrap) succeeded; breaking check ran when the schema already shipped.
  • Production fields have protovalidate constraints; reserved numbers cover removals.
  • gRPC exposure was treated as an API (authz, no schema leak) when that was in scope.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.