agentleFS
openclosed-orgopenclosed-org/axum-harness/.agents/skills/backend-engineering/SKILL.md

backend-engineering

Skill48 starsChanged 6 months agoApache-2.0View on GitHub

Always-on backend engineering quality kernel for Rust/Axum harness work. Use when changing or reviewing backend services, servers, workers, contracts, repo-tools, platform-facing backend metadata, or backend documentation. Guides invariant-first design, boundary discipline, evidence language, and template-quality code judgment without overriding ownership skills.

11 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use software-architecture.

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.
  • nek1987Skill

    software-architecture

    Guide for quality focused software architecture. This skill should be used when users want to write code, design architecture, analyze code, in any case that relates to software development.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

10 rules. 0 are common, 9 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 5 edited versions we read keep these.

  • 9

    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
  1. Distinctive:

    Use this skill as the default quality lens for backend harness work.

    line 13
  2. Distinctive:

    Keep the template small, explicit, typed, testable, and production-minded without pretending it is production-proven.

    line 19
  3. Distinctive:

    Prefer the smallest design that preserves:

    line 21
  4. Distinctive:

    Before changing backend behavior, answer:

    line 32
  5. Distinctive:

    Keep expected failures typed and matchable.

    line 53
  6. Distinctive:

    Do not add traits, managers, adapters, compatibility layers, or generic repositories unless they protect a real boundary or remove proven …

    line 57
  7. Distinctive:

    Prefer reference-module reuse over new structure, but do not copy business semantics blindly.

    line 58
  8. Distinctive:

    Use evidence terms precisely:

    line 63
  9. Very specific:

    Never upgrade a claim from declared to checked, tested, or proven without executable evidence.

    line 70
  10. Distinctive:

    Use focused workflow skills only when risk justifies deeper analysis:

    line 74

“Common” is measured: kept by at least 6 in 10 of the 5 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.

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

Backend Engineering

Use this skill as the default quality lens for backend harness work.

Workflow skills guide judgment. Ownership skills still decide writable boundaries.

Objective

Keep the template small, explicit, typed, testable, and production-minded without pretending it is production-proven.

Prefer the smallest design that preserves:

  1. backend correctness
  2. hard service, protocol, server, worker, and platform boundaries
  3. topology-late growth
  4. executable evidence
  5. template-user simplicity
  6. agent resistance to drift

Decision Kernel

Before changing backend behavior, answer:

  1. What invariant must not be violated?
  2. Which boundary owns the violated or changed behavior?
  3. What is synchronous, asynchronous, durable, idempotent, replayable, or eventually consistent?
  4. What is the cheapest correct design that restores the invariant?
  5. What evidence can honestly be claimed after this change?

Boundary Rules

  1. services/** are business capability libraries by default, not service processes.
  2. servers/** adapt protocols and compose services; handlers do not own domain logic.
  3. workers/** advance async state and must make retry, idempotency, checkpoint/replay, dedupe, and recovery explicit when touched.
  4. packages/contracts/** owns shared protocol shapes before external API or event changes land elsewhere.
  5. platform/model/** describes platform-level metadata and global defaults; service-local semantics stay in services/<name>/model.yaml.
  6. Generated artifacts are read-only; change sources and regenerate.
  7. Optional app shells and UI packages must not become backend-core prerequisites.

Quality Rules

  1. Represent important invariants in types, constructors, constraints, or final transaction boundaries.
  2. Keep expected failures typed and matchable; do not leak internal dependency details to clients.
  3. Retries for side-effecting work require durable idempotency or an explicit reason they are safe.
  4. Cross-boundary events require a clear atomic state mutation and publication story, usually outbox-driven.
  5. Async work must have bounded concurrency, cancellation/shutdown behavior, timeout or backpressure expectations, and observable failure paths when relevant.
  6. Do not add traits, managers, adapters, compatibility layers, or generic repositories unless they protect a real boundary or remove proven duplication.
  7. Prefer reference-module reuse over new structure, but do not copy business semantics blindly.
  8. Fix the smallest causal closed loop; avoid broad rewrites without an invariant-driven reason.

Evidence Rules

Use evidence terms precisely:

  1. declared: docs, metadata, manifests, or model files state intent.
  2. checked: schema, static validation, typecheck, drift check, or boundary check inspected structure.
  3. tested: automated tests exercised behavior.
  4. proven: a relevant executed gate or runtime/operational evidence supports the claimed invariant.

Never upgrade a claim from declared to checked, tested, or proven without executable evidence.

When To Deepen Review

Use focused workflow skills only when risk justifies deeper analysis:

  1. architecture review, release readiness, or template admission work
  2. distributed semantics, replay, idempotency, outbox, or worker recovery changes
  3. security, secrets, authorization, or tenant boundary changes
  4. broad backend refactors or cross-boundary protocol changes
  5. user explicitly asks for top-level backend review or code-quality audit

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.