Installs just this skill. Get the whole plugin for auto-invocation.
⚡ How it fires
How this skill gets triggered: by you, by Claude, or both.
Fires itselfClaude auto-loads it when your prompt matches the work.
You can call itInvoke it directly when you want it.
Slash command/flow-define
👁️ Context preview
The summary Claude sees to decide when to auto-load this skill.
Multi-AI requirements scoping using available external providers (Double Diamond Define phase)
📊 Stats
Stars3,869
Forks362
LanguageShell
LicenseMIT
📦 Ships with octo
</> SKILL.md
flow-define.SKILL.md
---name: flow-define
aliases:
- define
- define-workflow
- grasp
- grasp-workflow
description: Multi-AI requirements scoping using available external providers (Double Diamond Define phase)
PRIORITY TRIGGERS (always invoke): "octo define", "octo scope", "co-define", "co-scope"
DO NOT use for: implementation tasks (use flow-develop), research (use flow-discover),
review/validation (use flow-deliver), or built-in commands.
# Claude Code v2.1.12+ Integration
agent: Plan
context: fork
task_management: true
task_dependencies:
- flow-discover
execution_mode: enforced
pre_execution_contract:
- visual_indicators_displayed
validation_gates:
- orchestrate_sh_executed
- synthesis_file_exists
trigger: |
AUTOMATICALLY ACTIVATE when user requests clarification or scoping:
- "define the requirements for X"
- "clarify the scope of Y"
- "what exactly does X need to do"
- "help me understand the problem with Y"
**Use the ACTUAL results below. PROHIBITED: Showing only "🔵 Claude: Available ✓" without listing all providers.**
If `OCTO_ALLOWED_PROVIDERS` is set, treat it as the source of truth for which providers may participate. Providers filtered out by that allowlist are intentionally reported as unavailable; do not invoke or recommend them in the workflow.
**Display this banner BEFORE orchestrate.sh execution:**
🔵 Claude: Available ✓ - Consensus building and synthesis
💰 Estimated Cost: $0.01-0.05
⏱️ Estimated Time: 2-5 minutes
```
**DO NOT PROCEED TO STEP 2 until banner displayed.** The banner shows users which providers will run and what costs they'll incur — starting API calls without this visibility violates cost transparency.
---
### STEP 2: Read Prior State (MANDATORY - State Management)
**Before executing the workflow, read any prior context:**
- If **claude-mem** is installed, its MCP tools (`search`, `timeline`, `get_observations`) are available — use them to find past decisions on similar topics
**Before executing expensive multi-AI orchestration, capture the user's vision to scope the work effectively.**
**Ask clarifying questions using AskUserQuestion:**
```
Use AskUserQuestion tool to ask:
1. **User Experience**
Question: "How should users interact with this feature?"
Header: "User Flow"
Options:
- label: "API-first (programmatic access)"
description: "Build API endpoints first, UI later"
- label: "UI-first (user-facing interface)"
description: "Build user interface first, API supports it"
- label: "Both simultaneously"
description: "Develop API and UI in parallel"
- label: "Not applicable"
description: "This feature doesn't have a user interaction"
2. **Implementation Approach**
Question: "What technical approach do you prefer?"
Header: "Approach"
Options:
- label: "Fastest to market"
description: "Prioritize speed, use existing libraries"
- label: "Most maintainable"
description: "Focus on clean architecture, may take longer"
- label: "Best performance"
description: "Optimize for speed and efficiency"
- label: "Multi-LLM debate (Claude + available providers)"
description: "Multiple AI models debate the best approach — may use external provider credits or subscriptions"
3. **Scope Boundaries**
Question: "What's explicitly OUT of scope for this phase?"
Header: "Out of Scope"
Options:
- label: "Testing and QA"
description: "Focus on implementation, test later"
- label: "Performance optimization"
description: "Get it working first, optimize later"
- label: "Edge cases"
description: "Handle happy path only initially"
- label: "Nothing excluded"
description: "Everything is in scope"
multiSelect: true
```
**If user selected "Multi-LLM debate (Claude + available providers)" for approach:**
Before proceeding with orchestrate.sh, run a Multi-LLM debate to determine the technical approach:
```
/octo:debate --rounds 2 --debate-style collaborative "What is the best technical approach for [feature]? Consider: speed to market, maintainability, performance, and the existing codebase patterns."
```
Use the debate synthesis to set the approach context for the Define phase.
# Extract user answers from AskUserQuestion results
user_flow="[Answer from question 1]"
approach="[Answer from question 2]"
out_of_scope="[Answer from question 3]"
# Create context file with user vision
create_templated_context \
"define" \
"$(echo "$USER_REQUEST" | head -c 50)..." \
"User wants: $user_flow approach with $approach priority" \
"$approach" \
"Implementation of requested feature" \
"$out_of_scope"
echo "📋 Context captured and saved to .claude-octopus/context/define-context.md"
```
**This context will be used to:**
- Scope the multi-AI research (discover phase)
- Focus the requirements definition (define phase)
- Guide implementation decisions (develop phase)
- Validate against user expectations (deliver phase)
**DO NOT PROCEED TO STEP 4 until context captured.** User vision (UX approach, priorities, out-of-scope items) scopes the multi-AI research — without it, providers research too broadly and the definition misses the user's actual intent.
These spinner verb updates happen automatically - orchestrate.sh calls `update_task_progress()` before each agent execution. Users see exactly which provider is working and what it's doing.
**If NOT running in Claude Code v2.1.16+:** Progress indicators are silently skipped, no errors shown.
**After orchestrate.sh completes, verify it succeeded:**
```bash
# Find the latest synthesis file (created within last 10 minutes)
SYNTHESIS_FILE=$(find ~/.claude-octopus/results -name "grasp-synthesis-*.md" -mmin -10 2>/dev/null | head -n1)
if [[ -z "$SYNTHESIS_FILE" ]]; then
echo "❌ VALIDATION FAILED: No synthesis file found"
echo "orchestrate.sh did not execute properly"
exit 1
fi
echo "✅ VALIDATION PASSED: $SYNTHESIS_FILE"
cat "$SYNTHESIS_FILE"
```
**If validation fails:**
1. Report error to user
2. Show logs from `~/.claude-octopus/logs/`
3. DO NOT proceed with presenting results
4. DO NOT substitute with direct analysis — fallback to single-model analysis defeats the purpose of multi-provider consensus and produces narrower requirements
---
### STEP 6: Update State (MANDATORY - Post-Execution)
**After synthesis is verified, record findings and decisions in state:**
```bash
# Extract key definition from synthesis
key_definition=$(head -50 "$SYNTHESIS_FILE" | grep -A 3 "## Problem Definition\|## Summary" | tail -3 | tr '\n' ' ')
# Record any architectural decisions made
# (You should identify these from the synthesis - e.g., tech stack, approach, patterns)
