hellblazer/strategic-planning

Strategic Planning Skill

Delegates to the strategic-planner agent.

Model Selection

Default: sonnet. Escalate via model parameter on the Agent tool:

| Task Shape | Model | When | |-----------|-------|------| | ≤10 beads, single-phase, clear scope | sonnet (default) | Most planning work | | >15 beads, multi-RDR, or ambiguous requirements | opus | Complex decomposition |

When This Skill Activates

• Before implementing a feature, component, or module
• After brainstorming-gate approves a design — decompose into executable tasks
• When creating beads for a multi-phase effort
• When breaking vague requirements into concrete, sequenced steps
• Before architectural changes affecting multiple files or modules

Pre-Dispatch: Seed Link Context

Before dispatching the strategic-planner agent, seed T1 scratch with link targets so the auto-linker can create catalog links when the agent stores planning artifacts. See /nx:catalog skill for full reference.

1. If the task references an RDR (pattern RDR-\d+), resolve it: mcp__plugin_nx_nexus-catalog__search(query="RDR-NNN")
2. Check T1 scratch for rdr-planning-context (set by rdr-accept for RDR-driven plans)
3. Seed: mcp__plugin_nx_nexus__scratch(action="put", content='{"targets": [{"tumbler": "<tumbler>", "link_type": "relates"}], "source_agent": "strategic-planner"}', tags="link-context")
4. If no RDR/document reference found, skip seeding (auto-linker handles empty context gracefully)

Agent Invocation

Use the Agent tool to invoke strategic-planner:

## Relay: strategic-planner

**Task**: [what needs to be done]
**Bead**: [ID] or 'none'

### Input Artifacts
- Files: [relevant files]

### Deliverable
Implementation plan with tasks and dependencies

### Quality Criteria
- [ ] Execution plan created
- [ ] Dependency graph established
- [ ] Success criteria defined per task

For full relay structure and optional fields, see RELAY_TEMPLATE.md.

Planning Methodology

The agent uses mcp__plugin_nx_sequential-thinking__sequentialthinking:

1. Analyze codebase to understand context and constraints
2. Form hypothesis about optimal architecture approach
3. Validate approach against existing patterns
4. Break down work into logical tasks
5. Identify dependencies between tasks
6. Create beads for trackable work items
7. Define success criteria and test strategy

REQUIRED STEP: Call mcp__plugin_nx_nexus__nx_plan_audit after creating any plan (RDR-080 — no agent spawn needed).

Success Criteria

• [ ] Epic/feature breakdown documented
• [ ] Phased execution plan created
• [ ] Dependency graph established
• [ ] Beads created for all tasks
• [ ] Success criteria defined per task
• [ ] Risks identified and mitigated
• [ ] Plan validated via mcp__plugin_nx_nexus__nx_plan_audit (RDR-080)

Agent-Specific PRODUCE

Outputs generated by the strategic-planner agent:

• T3 knowledge: Validated architectural decisions via store_put tool: content="# Decision: {topic}\n{rationale}", collection="knowledge", title="decision-planner-{topic}", tags="decision,planning"
• T2 memory: Continuation state via memory_put tool: content="content", project="{project}", title="continuation-state.md", ttl="30d"
• T1 scratch: Analysis notes via scratch tool: action="put", content="Planning note: {consideration}", tags="planning,analysis" (flagged for T2 at plan completion)
• Beads: Epic and task beads with context links, success criteria, and file paths

Session Scratch (T1): Agent uses scratch tool for ephemeral working notes during the session. Flagged items auto-promote to T2 at session end.