Skip to content

radekamirko/C.R.I.S.P / .claude/skills/crisp-orchestrator

crisp-orchestrator

Verified

CRISP session manager — detects project state and routes to the right entry point. Use /crisp to start or resume any CRISP engagement. Handles new projects (discovery), existing codebases (archaeology), active sprints (execute), and validation (prove). Reads docs/crisp-state.json to know where you are.

CRISP — Session Orchestrator

Before doing anything else, output this exact text verbatim:

C — Clarify · R — Results · I — Investigate S — Spec · P — Prove · E — Execute

    Outcome first. Always.

Step 1: Detect project state

Check if docs/crisp-state.json exists.


If docs/crisp-state.json does NOT exist → Ask what we're starting

Do not assume. Ask:

"Starting fresh — two quick questions:

  1. Do you have an existing codebase, or are we starting from scratch?
  2. What's the project?"

If starting from scratch (no existing code):

  • Copy templates/crisp-state.json to docs/crisp-state.json, set mode: "discovery"
  • Read .claude/skills/phase1-clarify/SKILL.md
  • Begin Phase C with: "Tell me about the project — what are you trying to solve, or what did the client brief say?"
  • Do not explain the methodology. Just ask.

If existing codebase (has code, no docs):

  • Copy templates/crisp-state.json to docs/crisp-state.json, set mode: "archaeology"
  • Inform the user: "Got it — I'll run CRISP Archaeology to reverse-engineer your codebase into docs. This requires the crisp-archaeology skill. Open your codebase project in Claude Code and say: 'run archaeology on [path]'."
  • Do not attempt to run archaeology here — it runs in the target project, not in this repo.

If docs/crisp-state.json EXISTS → Read it and route

Read docs/crisp-state.json. Check mode first, then route.


Mode: "discovery" (standard CRISP phases)

1. Report current status

Project: [project.name] · Client: [project.client]
Mode: Discovery
─────────────────────────────────────────
Phases complete: [phases.complete joined with " · " — or "None"]
Current phase:   [phases.current]
─────────────────────────────────────────

Summarise the current phase state in 2–3 lines:

  • Phase C: "Problem defined as [oneSentence]. Go/No-Go: [goNoGo]. Type: [type]."
  • Phase R: "Baseline: [baseline]. Target: [successTarget]. [N] HITL zones defined."
  • Phase I: "Process track: [processTrack]. User types: [userTypes]. Data mapping: [dataMappingRequired]."
  • Phase S: "[sprintCount] sprints planned. Stack: [stack.layers summary]."
  • Phase P: "Outcome: [outcome]."

If open questions exist in the current phase, list them before asking what to do next.

2. Check if ready for Execute

If phases.S.complete: true and handoffs.ready_for_execute: false:

"Phase S is complete. Ready to start building? Say 'Start Sprint 1' to hand off to CRISP Execute."

If the user confirms → update handoffs.ready_for_execute: true in crisp-state.json.

3. Route

User saysAction
Continue / yes / Phase [X]Load phase SKILL.md and proceed
Start Sprint 1 / execute / buildSet handoffs.ready_for_execute: true. Route to Execute (see Execute routing below).
New projectAsk: overwrite state or new docs/ folder?
Show full statePrint crisp-state.json in readable format
What's leftList incomplete phases and open questions

Mode: "archaeology"

1. Report status

Project: [project.name]
Mode: Archaeology
─────────────────────────────────────────
Recon:       [phases.recon.status]
Strawman:    [phases.strawman.status]
Elicitation: [phases.elicitation.status] ([artifacts.confidence_summary.confirmed] confirmed · [artifacts.confidence_summary.open] open)
Output:      [phases.output.status]
─────────────────────────────────────────

2. Route

If handoffs.archaeology_complete: false:

"Archaeology is in progress. Open the target codebase in Claude Code and continue with the crisp-archaeology skill."

If handoffs.archaeology_complete: true and handoffs.entry_point_after: "phase-s":

"Archaeology is complete. [N] open questions remain (see docs/open-questions.md). Ready to run Phase S to spec the next phase?" Route to Phase S on confirmation.

If handoffs.archaeology_complete: true and handoffs.entry_point_after: "crisp-execute":

"Archaeology is complete. Sprint plan exists. Ready to start building?" Route to Execute on confirmation.


Mode: "execute"

1. Report status

Project: [project.name]
Mode: Execute
─────────────────────────────────────────
Current sprint:   [execute.current_sprint]
Sprints complete: [execute.sprints_complete]
Change requests:  [execute.change_requests.length] queued
─────────────────────────────────────────

Show gate status for the current sprint if available:

  • Product gate: [execute.gates.sprint_N.product]
  • Security gate: [execute.gates.sprint_N.security]

2. Route

User saysAction
Continue / start sprint / buildLoad .claude/skills/crisp-execute/SKILL.md and proceed
New requirement: [X]Route to Execute change request intake
Close sprint [N]Route to Execute sprint close sequence
Show change requestsList execute.change_requests
Go to Phase P / proveCheck all sprints complete, then route to Phase P

Mode: "prove"

Route to .claude/skills/phase5-prove/SKILL.md.


Phase routing table

Phase / ModeSkill to load
C — Clarify.claude/skills/phase1-clarify/SKILL.md
R — Results.claude/skills/phase2-results/SKILL.md
I — Investigate.claude/skills/phase3-investigate/SKILL.md
S — Spec.claude/skills/phase4-spec/SKILL.md
E — Execute.claude/skills/crisp-execute/SKILL.md
P — Prove.claude/skills/phase5-prove/SKILL.md
ArchaeologyRuns in target project via crisp-archaeology skill (separate repo)

crisp-state.json handoff signals

The orchestrator reads and writes these fields to coordinate between modes:

FieldSet byMeaning
modeOrchestrator on createWhich system is active
handoffs.discovery_completePhase S exitAll phases done, ready for execute
handoffs.archaeology_completeArchaeology skillArchaeology finished, ready to proceed
handoffs.ready_for_executeOrchestratorSpec is locked, execute can start
handoffs.execute_completeExecute skillAll sprints done, ready for prove
execute.current_sprintExecute skillWhich sprint is active
execute.gates.sprint_NExecute skillGate results per sprint
execute.change_requestsExecute skillQueued change requests

Rules

  • Never skip a phase without explicit user instruction.
  • If a phase is marked complete but the user wants to re-run it — allow it, flag: "Phase [X] is marked complete. Running again won't reset state automatically."
  • If crisp-state.json has open questions — surface them before starting the next phase.
  • If phases are out of order — flag it: "Phase R isn't marked complete. Running Phase S without it means some pre-fills will be missing. Continue anyway?"
  • If execute.change_requests has queued items when a sprint opens — surface them: "[N] change requests are queued. Review before sprint starts?"
  • Keep responses tight. Status reporting is a tool, not a presentation.
rea-changelog-updatemorluto/reaPrepare or rewrite REA release changelogs and GitHub release notes from pinned Git history, with verified contributor thanks and Release Please synchronization.Stars: 471kUpdated Claude CodeCodexVerifiedInstall
academy-guideanthropics/skillsStop and check this skill before finishing any reply to a question about how to use Claude or a Claude product — it recommends matching courses, tutorials, and use cases from Claude Academy (academy.claude.com), Anthropic's learning hub. Trigger on: "how do I", "how can I", "getting started with", "what can Claude do", "teach me", "learn to use"; questions about artifacts, projects, skills, plugins, connectors, MCP; requests about rolling Claude out to a team, class, or organization; and any ask for training materials, onboarding content, or learning resources. Use it when the user is learning how to use a feature or product — not when they are mid-task and just want the task done. This skill composes with other skills: after consulting product documentation to answer how a Claude feature works, also check here for a matching course or tutorial — a docs-grounded answer and an Academy recommendation belong together. Only recommend on a strong match; never invent Academy content.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
claude-apianthropics/skillsReference for the Claude API / Anthropic SDK — model ids, pricing, params, streaming, tool use, MCP, agents, caching, token counting, model migration. TRIGGER — read BEFORE opening the target file; don't skip because it "looks like a one-liner" — whenever: the prompt names Claude/Anthropic in any form (Claude, Anthropic, Fable, Opus, Sonnet, Haiku, `anthropic`, `@anthropic-ai`, `claude-*`, `us.anthropic.*`, `[1m]`); the user asks about an LLM (pricing/model choice/limits/caching) — never answer from memory; OR the task is LLM-shaped with provider unstated (agent/MCP/tool-definition/multi-agent/RAG/LLM-judge/computer-use; generate/summarize/extract/classify/rewrite/converse over NL; debugging refusals/cutoffs/streaming/tool-calls/tokens). SKIP only when another provider is being worked on (overrides all triggers): OpenAI/GPT/Gemini/Llama/Mistral/Cohere/Ollama named in the query; OR `grep -rE 'openai|langchain_openai|google.generativeai|genai|mistralai|cohere|ollama'` over the project hits (run this grep FIRST if no provider named — don't Read the file).Stars: 198kUpdated Claude CodeCodexUnverifiedInstall
doc-coauthoringanthropics/skillsGuide users through a structured workflow for co-authoring documentation. Use when user wants to write documentation, proposals, technical specs, decision docs, or similar structured content. This workflow helps users efficiently transfer context, refine content through iteration, and verify the doc works for readers. Trigger when user mentions writing docs, creating proposals, drafting specs, or similar documentation tasks.Stars: 198kUpdated Claude CodeCodexVerifiedInstall