agentleFS
agentik-osagentik-os/claude-code-skills/blueprint-os/SKILL.md

blueprint-os

Skill1 starsChanged 55 days agoMITView on GitHub

Compile software, AI, platform, service, marketplace, mobile, web, or internal-tool ideas and existing project context into a complete, coherent, traceable Product + Technical Definition Pack before implementation planning or coding. Trigger on /blueprint, Blueprint {OS}, product blueprint, product-definition audit, recovery, revision, extension, delta, or preparation for Stepper {OS}. Preserve project decisions and stable IDs, separate evidence from assumptions, define product/UX/domain/data/API/AI/security/operations/test contracts, run gates, and continue across outputs without declaring partial work complete.

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 7 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈7

    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:

    Never write product code, create atomic DEV steps, or invoke Stepper/Build implicitly.

    line 17
  2. Distinctive:

    If AI is relevant, define responsibilities/forbidden responsibilities, prompts, context, memory, tools, autonomy, confirmation, provenance, …

    line 49
  3. Distinctive:

    Run specialist critics, resolve findings, evaluate all gates, checkpoint, and create a frozen Stepper manifest only when eligible.

    line 52
  4. Distinctive:

    Use stable monotonic prefixes including:

    line 56
  5. Distinctive:

    Never renumber or recycle.

    line 60
  6. Distinctive:

    Ask at most three high-leverage questions when the answer changes product promise, economics, trust, legal/privacy exposure, data …

    line 66
  7. Distinctive:

    Never hide an unanswered question inside a decision.

    line 67
  8. Very specific:

    If output is split, preserve IDs and emit completed/current/next/remaining sections, blockers, ID counters, gate snapshot, …

    line 79

≈ “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

Blueprint {OS}

Operate as a product-definition compiler over one canonical project state.

Boundary

Enforce Idea → Blueprint {OS} → Stepper {OS} → Build {OS} → Ship.

  • Blueprint defines product/system truth and stops at BLUEPRINT COMPLETE — STEPPER READY.
  • Stepper creates the implementation DAG and stops at BUILD READY.
  • Build writes code and changes systems.
  • Never write product code, create atomic DEV steps, or invoke Stepper/Build implicitly.

Required context

Load:

  1. references/system-prompt.md as the full operating contract.
  2. references/blueprint-contract.md for required artifacts and record schemas.
  3. references/orchestration-and-gates.md for role graph, critics, and gates.
  4. references/response-and-continuation.md for output and resume behavior.

Load references/functions-and-state.md for tool/state work, references/omega-os-integration.md for installation/integration, and references/deep-guide.md for explanations/training.

Modes

  • NEW: new product.
  • RECOVER: recover canonical truth from prior sources.
  • EXTEND: add a module while preserving IDs and impact links.
  • REVISE: supersede accepted decisions and propagate change.
  • AUDIT: find gaps, conflicts, orphans, and failed gates.
  • DELTA: compare semantic versions and downstream impact.

Workflow

  1. Frame project, scope, stage, request, constraints, non-goals, and completion criteria.
  2. Recover all authorized context and establish source authority.
  3. Classify statements as FACT, DECISION, ASSUMPTION, PROPOSAL, UNKNOWN, CONFLICT, DEFERRED, or SUPERSEDED.
  4. Define vision, users/JTBD, value events, positioning, business/economic logic, goals, metrics, guardrails, and non-goals.
  5. Define capabilities, atomic requirements, actors, roles, permissions, consent, actions, and end-to-end flows.
  6. Define IA, navigation, screen/surface contracts, all non-happy states, design/content rules, accessibility, and localization.
  7. Define bounded contexts, entities, ownership, state machines, rules, invariants, idempotency, concurrency, expiry, reversals, and reconciliation.
  8. Define architecture, trust boundaries, data lifecycle, APIs, tools, integrations, events, failure domains, observability, migration, and evolution.
  9. If AI is relevant, define responsibilities/forbidden responsibilities, prompts, context, memory, tools, autonomy, confirmation, provenance, abstention, evals, fallback, monitoring, cost, latency, and rollback.
  10. Define security, privacy, abuse, NFRs, operations, acceptance, test architecture, analytics, risks, release boundaries, and validation strategy.
  11. Build bidirectional traceability from sources/decisions through requirements and behavior/system contracts to tests, metrics, and risks.
  12. Run specialist critics, resolve findings, evaluate all gates, checkpoint, and create a frozen Stepper manifest only when eligible.

Stable IDs

Use stable monotonic prefixes including:

SRC FCT DEC ASM PRP UNK CNF DEF GOAL MET PER JTBD CAP REQ ACT FLOW SCR CMP RULE INV ENT CMD QRY API EVT INT ADR AIC MEM EVAL SEC PRIV ABU NFR OPS RSK AC TST EXP REL

Never renumber or recycle. Preserve supersession history.

Question policy

  • Continue with explicit assumptions for reversible, low-impact details.
  • Register material unknowns and deferred choices.
  • Ask at most three high-leverage questions when the answer changes product promise, economics, trust, legal/privacy exposure, data ownership, irreversible architecture, or major scope.
  • Never hide an unanswered question inside a decision.

Completion

Allowed statuses:

  • BLUEPRINT IN PROGRESS
  • BLUEPRINT BLOCKED
  • BLUEPRINT COMPLETE — STEPPER READY

Completion requires all mandatory artifacts present or N/A with rationale, no critical conflict/failing gate, 100% traceability for critical decisions/requirements, at least 95% for normative requirements, explicit acceptance, and a frozen Stepper Input Manifest.

If output is split, preserve IDs and emit completed/current/next/remaining sections, blockers, ID counters, gate snapshot, revision/checksum. On “continue”, resume exactly from the pointer. Never call a partial Blueprint complete.

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.