agentsclimarketplace

Spec lite

Skill Aquish-Lee/agent-handrails/skills/spec-lite

Explicit, claim-scaled engineering handrails for AI coding agents.

Install
npx -y skills add Aquish-Lee/agent-handrails --skill spec-lite

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

2 things to look at

  • 25 days oldThe repository was created 25 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
  • 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.

What its author says it does

Copied from the file, not written here

Turn converged intent into a lightweight outcome contract when a user explicitly requests a spec or durable acceptance semantics are needed before consequential implementation. Skip routine tasks and unresolved discovery.

SKILL.md

2.2 KB, 394 tokens by cl100k_base, as published. Nobody here has run it

Spec Lite

Record what must be true without prescribing unnecessary implementation detail. This is an optional outcome contract, not a required document in every task.

<!-- handrails-contract contract: usable-loop/v2 role: outcome-contract-author reference: references/usable-loop-v2.md runtime: none -->

Read the local v2 contract for exact record and authority fields.

When to write a spec

Use this skill for durable behavior/acceptance semantics, cross-session coordination, public interfaces, migrations, or consequential tradeoffs. Return to alignment when core intent remains unresolved. Skip a spec for a clear micro-task or when a plan alone is enough.

An unconflicted current user request is authoritative when the user controls the task. Preserve its exact meaning. Any agent-generated elaboration, inferred constraint, or reinterpretation is provisional until confirmed.

Write the outcome contract

Keep the document proportional and include:

  • complete Outcome Reference;
  • goal and non-goals;
  • desired externally observable behavior;
  • constraints, compatibility, risks, and authorization limits;
  • testable acceptance criteria;
  • verification signals;
  • assumptions and open questions;
  • implementation readiness: ready | conditionally-ready | not-ready.

Separate facts, user decisions, and assumptions. Avoid architecture or file-level prescription unless the outcome depends on it. Use examples for semantic edge cases when prose would remain ambiguous.

Do not make persistence confer authority. A repository spec can be provisional; a conversational request can be authoritative. When the outcome changes, create a new Outcome Revision and supersede the exact old revision rather than editing history conceptually.

Direct result

Lead with readiness and the agreed behavior, then acceptance criteria, constraints, and unresolved blockers. Do not create a plan or invoke another handrail automatically. Expand a Checkpoint only on a continuity trigger.

What ships with it: 2 files

3.1 KB alongside SKILL.md

agents/

references/

Keep looking

Skills are one crate of 328,083. 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.