agentleFS
youmissedaspotyoumissedaspot/ymas-agent-workflows/skills/spec-driven-development/SKILL.md

spec-driven-development

Skill1 starsChanged 46 days agoMITView on GitHub

Formalize accepted product or architecture decisions into SPEC and/or SPECARC and maintain systems.md from the complete current specification set, or develop consequential requested work with unresolved definitions. Resume existing projects at the appropriate stage; routine, already specified implementation does not require the full path.

103 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use using-agent-skills, idea-refine and 10 more.

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.
  • LLl0k0laDSkill

    using-agent-skills

    Use when starting a session or when you need to discover which skill applies to the current task. This is the meta-skill that governs how all other skills are discovered and invoked.

    Also in 44 projects that use this

  • LLl0k0laDSkill

    idea-refine

    Refine ideas through structured divergent and convergent thinking. Use "idea-refine" or "ideate" to trigger.

    Also in 54 projects that use this

  • MrCipherSmithSkill

    constraint-driven-development

    Use when a project needs a persistent, measurable quality bar — starting a project without documented standards, 'set up constraints', or when coverage/perf/a11y thresholds get re-debated per-PR. Produces CONSTRAINTS.md with enforceable numbers, ratchets, and a guard against silent weakening. NOT for throwaway prototypes or a one-off CI setup.

    Also in 16 projects that use this

  • HK-hubSkill

    observability-and-instrumentation

    Instruments code so production behavior is visible and diagnosable. Use when adding logging, metrics, tracing, or alerting. Use when shipping any feature that runs in production and you need evidence it works. Use when production issues are reported but you can't tell what happened from the available data.

    Also in 36 projects that use this

  • seikaikyoSkill

    doubt-driven-development

    對非平凡決策啟動新鮮 context 對抗審查(找碴不背書),在修正還便宜的時候抓出錯誤方向。適用:高風險改動(production、資安敏感邏輯、不可逆操作)、不熟的程式碼、要宣稱「這樣是安全的 / 可行的」之前。機械性操作與一行修改不適用。

    Also in 44 projects that use this

  • +7 more from LLl0k0laD

The file

We split this skill into the rules it gives an agent

3 rules. 0 are common, 3 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 96 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Before substantial specification, architecture or implementation, demonstrate a concise Understanding Check after relevant inspection and …

    line 10
  2. Distinctive:

    Read the adopting repository's instructions, authority hierarchy, and current context.

    line 12
  3. Distinctive:

    When a downstream layer exposes an upstream error, correct the smallest authoritative unit, propagate its effects forward, and repeat …

    line 20

“Common” is measured: kept by at least 6 in 10 of the 96 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

Specification-driven development

Use this workflow to formalize a True Specification conversation or carry a consequential product or system change from intent to accepted implementation, within the requested scope. Follow Discuss → True Spec Worksheet → Decide → SPEC and/or SPECARC → systems.md → Plan → Implement → Verify → Human Acceptance where applicable, using only the stages needed. A specification-only request ends with the requested definitions and relevant map review/update. For a narrow defect with established expected behavior, use audit-repair; routine maintenance, cosmetic edits, and implementation already specified at the needed level do not require the full specification path.

Before substantial specification, architecture or implementation, demonstrate a concise Understanding Check after relevant inspection and before proposing implementation. Keep trivial explicitly authorized work proportional; this is not a new approval gate.

  1. Read the adopting repository's instructions, authority hierarchy, and current context. Use its existing authoritative documents and names; this package supplies a development method, not product truth.
  2. Inspect relevant existing specifications, accepted decisions, operational context, and an existing systems.md for architecture, dependencies, and impacts. Resume from the appropriate stage. Use true-spec-worksheet when consequential product or architecture decisions remain unresolved; it prepares a provisional decision aid, not final definitions. Avoid reopening settled choices without cause. Then formalize explicitly accepted decisions into SPEC and/or SPECARC, sibling types with no fixed ordering. SPEC defines product/system requirements; SPECARC defines accepted architecture for a durable concern and may precede surrounding SPEC documents or a complete map. Respect the project's approval and precedence. Do not let an existing map override newly accepted higher product decisions or invent requirements to preserve old architecture.
  3. After substantive SPEC or SPECARC creation or revision, review the complete current specification set. Include unaffected documents and both sibling types. Create or update systems.md when there is enough specification material for useful integration; record an early deferral rather than invent a meaningless map. Map governing specifications, required systems, state ownership, reads/writes, responsibilities, communication, interfaces, dependencies, data/control/event flows, lifecycles, and enforcing cross-system invariants. Record IDs/paths, versions or unversioned status, acceptance status, overlaps, conflicts, gaps, and whole-system coherence. Update the map when architecture, ownership, boundaries, dependencies, interfaces, flows, or cross-system behavior change. Preserve its established path or use the project's directory (default docs/systems/systems.md). Follow the handoff procedure. A SPECARC concern does not replace the complete-set map. Return missing decisions to their owners and refresh affected definitions and integration. Report a review that needs no meaningful map change without mechanically bumping its version.
  4. Reuse existing owners and apply the project's standard and document versioning to specifications and systems.md. Sequence IDs remain identities: use dotted metadata such as Version: v3.1, keep the same ID and live path, and archive the complete predecessor with filename form _v3dot1.md when that revision is superseded. Retain SYS as project-defined system-level documentation and ADR as decision context/rationale; neither is automatically interchangeable with SPECARC. Load only the needed aid: True Specification template, Canonical Specification template, architecture specification template, systems map checklist, systems.md template, detailed system template, or increment checklist. Do not create every document for every task.
  5. When implementation is in scope and the relevant layers are sufficiently settled, inspect implementation, tests, schemas, interfaces, and explanatory Git history. Use an adopted Feature Map for capability-level clauses, entrypoints, dependencies and verification navigation; preserve complete-set systems review and inspect original sources. Build an implementation plan from authority and repository reality, then use the plan and adversarial review rubric when complexity warrants review. Return missing product decisions to the owning specification layer.
  6. For authorized implementation, use bounded units through the agent execution loop and workbranch rule. Use maintained verification tools when repeated nontrivial procedures warrant them. Check affected map/tool freshness after relevant changes. Verify actual behavior, repair narrowly, and keep the three-failure escalation boundary. Do not reinterpret product truth to simplify code. Preserve the specification/prototype boundary; apply the architecture discrepancy trigger immediately, independently of the three-failure ceiling.
  7. Prepare human acceptance guidance where the project reserves final acceptance to a human owner. Report agent verification and human acceptance separately.

When a downstream layer exposes an upstream error, correct the smallest authoritative unit, propagate its effects forward, and repeat affected review. Keep project-specific decisions in the adopting repository.

For implementation and verification, apply the no-code-comments rule; keep necessary explanations and reasoning in the owning specifications and review changed code and examples before completion.

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.