agentleFS
System-Error-WorldwideSystem-Error-Worldwide/the-loop/.agents/skills/spec-pack/SKILL.md

spec-pack

Skill0 starsChanged 3 months agoMITView on GitHub

Produce mutually consistent implementation-ready specifications and map every requirement to a build slice and verification gate. Use for a new app, website or major feature that needs PRD, technical design, flow, design brief, backend schema and engineering plan before code.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    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 for a new application, website, major feature, data-model change, multi-subsystem change or new user-facing surface.

    line 19
  2. Very specific:

    run ID, stage spec_pack, track and declared asset.

    line 47
  3. Distinctive:

    Ask whether two competent implementers could build materially different but specification-compliant products.

    line 69

≈ “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

Spec pack

Purpose

Provide the complete bundled fallback for lifecycle.spec_pack. Establish the contract an implementer and independent tester will follow before gated implementation.

Use when

Use for a new application, website, major feature, data-model change, multi-subsystem change or new user-facing surface. Use a smaller recorded specification set only when the strategy proves the full gate does not apply.

Required inputs

  • Validated run ID, declared asset, track, approved strategy and current stage.
  • Problem, users, scope, constraints, risks, success measures and done gate.
  • Known architecture, data, workflow, design and release context.
  • Frozen budgets, current authority, owner and matching lease for any writes.
  • Current evidence, issue state and recorded human decisions.

Reject missing or contradictory inputs rather than choosing an architecture by implication.

Procedure

  1. Determine the required documents from asset type and change scope.
  2. For a new app, website or major feature, produce the six documents: PRD, TDD, app flow, design brief, backend schema and engineering plan.
  3. Give every functional and non-functional requirement a stable identifier and testable outcome.
  4. Specify happy paths, failure paths, denial, interruption, recovery, unsupported and terminal states.
  5. Define data shape, authority, privacy, security, migration and rollback behavior.
  6. Separate structure and operational behavior from surface treatment where a user-facing output exists.
  7. Map every requirement-to-slice relationship and name the exact done gate and evidence method for each slice.
  8. Cross-check terminology, enums, scope, dependencies and deferred work across all documents.
  9. Record unresolved decisions that materially alter the product as human gates; do not replace them with assumptions.

Output contract

Return:

  • run ID, stage spec_pack, track and declared asset;
  • required document inventory and paths;
  • requirements, assumptions and decision ledger;
  • architecture, state, workflow, design and release contracts;
  • implementation slices with dependencies, owners, done gates and rollback;
  • requirement-to-slice and requirement-to-verification mapping;
  • evidence references, issues, gate state and next stage or halt.

Evidence gate

Pass only when the required documents exist, are mutually consistent and implementation-ready; every requirement maps to a slice and verification method; local references resolve; and no unresolved decision can change architecture, data, public behavior, authority or release risk. Evidence must be persisted before Build.

Track requirements

For the code track, specify repository boundaries, executable behavior, security checks, migrations and technical rollback. For the non-code track, specify sources, factuality, format, calculation or render checks, review method and outward publication gate.

Safety and authority

Specification is not authority to implement, publish, deploy or contact an external system. Writes stay inside the declared specification paths and use the active lease. Private data, credential-shaped values and unsupported compatibility claims stay out of public artifacts.

Self-refutation

Ask whether two competent implementers could build materially different but specification-compliant products. If yes, identify and close the ambiguous requirement or halt for the missing decision.

Halt conditions

Halt for a material unresolved decision, missing source of truth, conflicting requirement, unbounded scope, absent verification method, lost ownership, namespace change, expired authority, exhausted budget or detected stop. Name incomplete documents and the exact input needed.

Core execution does not require network access or the source checkout. The links below are optional public documentation.

References

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.