agentsclimarketplace
Sign in

Engineering standard

Skill SylphxAI/skills/skills/engineering-standard

Apply binding engineering constraints when implementing, fixing, debugging, designing, reviewing, refactoring, or migrating durable product code, APIs, handlers, workers, architecture, capabilities, DDD/Clean/Hexagonal boundaries, functional cores, contracts, storage, concurrency, security, tests, observability, or delivery-critical runtime paths. Use when code or runtime behavior is the artifact; use a specialist procedure when a whole observability, security, migration, or product contract is primary.From its SKILL.md

Install
npx -y skills add SylphxAI/skills --skill engineering-standard

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

5.3 KB, 934 tokens by cl100k_base, as published. Nobody here has run it

Engineering Standard

Requirement: apply this standard when the task matches its scope.

Full progressive-disclosure body: references/full-standard.md.
Rule IDs: references/binding-predicates.md.

For code structure or architecture migration, also read:

When creating a project, adding a package, or upgrading a runtime, framework, SDK, generator, plugin, or library, use dependency-version-selection. Query the authoritative live release source; never select a version from model memory, a stale template, or an old example.

When invariants span a broad generated state or input space, read Generative verification.

When exact expected outputs are unavailable but relations or independent implementations can falsify behavior, read Oracle-free verification.

When a test, gate, alert, validator, or recovery control claims to catch a failure, read Control-effectiveness verification.

When diagnosing or fixing a defect, changing observable behavior, or refactoring behavior whose current contract is not yet executable, read Reproduction-driven engineering.

For critical concurrency, distributed coordination, authorization, or irreversible-effect state spaces, read Formal modelling and model checking.

When concurrent or offline replicated state needs an algebraic merge, read CRDT convergence.

For a large shared capability catalog, concept scheme, faceted vocabulary, or cross-domain taxonomy, use semantic-taxonomy-design; do not force identity, hierarchy, and mutable labels into one tree path.

For a new or materially changed trust boundary, privileged action, data flow, integration, or deployment exposure, use security-threat-modeling to produce the security-design contract before or alongside implementation.

Method

  1. Load this package for durable engineering work on product code.
  2. Read references/full-standard.md for the complete standard.
  3. Before broad implementation, create or update the owning repository ADR for a material durable decision about architecture, ownership, public contracts, persistence, security/privacy, delivery semantics, or an enterprise default.
  4. Keep ordinary implementation detail governed by an accepted decision in code, schemas, and tests; do not create ceremonial ADRs.
  5. Apply the strongest relevant subset and prefer executable evidence over copying this body into product repositories.
  6. Procedure Skills should name this standard as a composition handoff when durable product code or runtime behavior changes. Native runtimes select matching Skills from metadata; the catalog has no dependency primitive, so do not invent one or restate these predicates in every procedure.

Boundaries

  • Does not grant deploy, credential, or MCP capabilities.
  • Does not replace product-local technical ADRs.
  • Does not auto-route as a procedure; it is binding policy text with separate conformance evidence.

When not to use

  • Commercial packaging alone → commercial-decision class packages.
  • Product-specific live coordination operations → the matching adapter Skill and that product's current API or tool schema.
  • One-off scripts with no durable product surface and no shared contract impact.

Composition and output

Apply this standard as constraints on the artifact requested by the user or owned by the primary procedure. Do not emit a separate per-standard compliance report merely because this Skill was injected.

Integrate only material obligations, deviations, evidence, and gaps into the primary artifact or final status. When conformance assessment is itself the requested job, produce one standalone domain conformance record from this standard.

Guardrails

  • Fail closed on secret material in source and logs.
  • Never treat health/readiness probes as product capability proof.
  • In development or without live compatibility/state risk, prefer a verified one-step cutover. Use expand-contract only when a demonstrated live data, compatibility, availability, or external-effect boundary requires it.
  • Apply the canonical Capability-first architecture from the first durable product commit; project size and migration cost are not exemptions.

What ships with it: 17 files

176.7 KB alongside SKILL.md

agents/

references/

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.