@monoes/monomindcli 2.16.3 → 2.16.5

package/.claude/agents/github/repo-architect.md

@@ -296,7 +296,7 @@ const integrationPattern = {
     "@monomind/memory": {
       role: "persistence_layer",
       dependencies: [],
-      provides: ["LanceDB", "HNSW_search", "sessions"],
+      provides: ["SQLite", "HNSW_search", "sessions"],
     },
   },
   communication: "MCP_protocol",

package/.claude/agents/specialists/integration-architect.md

@@ -20,7 +20,7 @@ Keep the 5 Monomind packages working as a coherent system. When a package change
 |---------|------|-------------------|
 | `@monomind/cli` | Orchestration layer | MCP server, CLI commands, init generator |
 | `@monoes/hooks` | Intelligence engine | Hook events, background workers, pattern learning |
-| `@monomind/memory` | Persistence layer | LanceDB, HNSW search, session state |
+| `@monomind/memory` | Persistence layer | SQLite (sql.js fallback), HNSW search above 5,000 entries, session state |
 | `@monomind/security` | Input validation | CVE remediation, safe executor, path validator |
 | `@monoes/monograph` | Knowledge graph | Dependency analysis, community detection, impact |
 
@@ -46,14 +46,10 @@ All MCP tools exposed via `@monomind/cli/src/mcp-tools/` must:
 
 ### Memory Access Pattern
 
-```typescript
-// Standard pattern for cross-package memory access
-import { LanceDB } from '@monomind/memory';
-
-const db = LanceDB.getInstance();
-await db.store({ key, value, namespace: 'package-name' });
-const result = await db.search({ query, namespace: 'package-name' });
-```
+Cross-package memory access goes through the CLI's memory bridge
+(`packages/@monomind/cli/src/memory/memory-bridge.ts`), which backs `memory store/search`
+and the MCP memory tools on top of `@monoes/memory`'s SQLite backend. Use a
+package-specific `namespace` for everything a package stores.
 
 ## Integration Checklist
 
@@ -63,7 +59,7 @@ When a new feature spans multiple packages:
 - [ ] Hook events documented in `@monoes/hooks/src/types.ts`
 - [ ] MCP tool registered in `@monomind/cli/src/mcp-tools/index.ts`
 - [ ] Security validation added at system boundary
-- [ ] Memory schema migration written if LanceDB schema changes
+- [ ] Memory schema migration written if the SQLite schema (`sql-schema.ts`) changes
 - [ ] `pnpm run sync:claude-trees` run after any `.claude/` changes
 - [ ] Cross-package integration test added
 
@@ -80,8 +76,8 @@ packages/@monomind/hooks/src/
   workers/             — Background worker definitions
 
 packages/@monomind/memory/src/
-  lancedb/             — LanceDB core
-  hnsw/                — Vector search
+  sql-backend.ts       — SQLite backend; search() switches to HNSW above the threshold
+  hnsw-index.ts        — HNSW ANN index
 
 packages/@monomind/security/src/
   validators/          — Input validators
@@ -90,7 +86,7 @@ packages/@monomind/security/src/
 
 ## Coordination with Other Specialists
 
-- **Memory Specialist** — LanceDB schema changes, HNSW configuration
+- **Memory Specialist** — SQLite schema changes, HNSW configuration
 - **Performance Engineer** — Benchmarking cross-package call overhead
 - **Security Architect** — Validating integration boundary security
 - **Queen Coordinator** — Orchestrating multi-package feature rollouts

package/.claude/commands/agents/agent-capabilities.md

@@ -1,5 +1,6 @@
 ---
 name: agents:agent-capabilities
+description: Matrix of agent capabilities and their specializations.
 ---
 
 # agent-capabilities

package/.claude/commands/agents/agent-coordination.md

@@ -1,5 +1,6 @@
 ---
 name: agents:agent-coordination
+description: Coordination patterns for multi-agent collaboration.
 ---
 
 # agent-coordination

package/.claude/commands/agents/agent-spawning.md

@@ -1,5 +1,6 @@
 ---
 name: agents:agent-spawning
+description: Guide to spawning agents with Claude Code's Task tool.
 ---
 
 # agent-spawning

package/.claude/commands/agents/agent-types.md

@@ -1,5 +1,6 @@
 ---
 name: agents:agent-types
+description: Complete guide to all 60+ available agent types in Monomind.
 ---
 
 # agent-types

package/.claude/commands/analysis/bottleneck-detect.md

@@ -1,5 +1,6 @@
 ---
 name: analysis:bottleneck-detect
+description: Detect performance bottlenecks in system components and swarm operations.
 ---
 
 # bottleneck-detect

package/.claude/commands/analysis/performance-bottlenecks.md

@@ -1,5 +1,6 @@
 ---
 name: analysis:performance-bottlenecks
+description: Identify and resolve performance bottlenecks in your development workflow.
 ---
 
 # Performance Bottleneck Analysis

package/.claude/commands/analysis/performance-report.md

@@ -1,5 +1,6 @@
 ---
 name: analysis:performance-report
+description: Generate performance metrics reports for swarm operations and system components.
 ---
 
 # performance-report

package/.claude/commands/analysis/token-usage.md

@@ -1,5 +1,6 @@
 ---
 name: analysis:token-usage
+description: Analyze token consumption and cost across sessions.
 ---
 
 # token-usage

package/.claude/commands/automation/README.md

@@ -10,7 +10,7 @@ Commands and guidance for agent automation, self-healing, and session management
 
 - [auto-agent](./auto-agent.md) — spawn agents by type and task (`monomind agent spawn`)
 - [smart-spawn](./smart-spawn.md) — auto-spawn agents via hooks pre-task (`monomind hooks pre-task --auto-spawn`)
-- [workflow-select](./workflow-select.md) — run predefined workflow templates (`monomind workflow run`)
+- [workflow-select](./workflow-select.md) — run predefined workflow templates (`monomind org create --template`)
 
 ## Guidance Docs (contextual reference)
 

package/.claude/commands/automation/auto-agent.md

@@ -1,5 +1,6 @@
 ---
 name: automation:auto-agent
+description: Automatically spawn agents based on task requirements.
 ---
 
 # auto-agent

package/.claude/commands/automation/self-healing.md

@@ -1,5 +1,6 @@
 ---
 name: automation:self-healing
+description: Automatically detect and recover from errors without interrupting your flow.
 ---
 
 # Self-Healing Workflows

package/.claude/commands/automation/session-memory.md

@@ -1,5 +1,6 @@
 ---
 name: automation:session-memory
+description: Maintain context and learnings across Claude Code sessions for continuous improvement.
 ---
 
 # Cross-Session Memory

package/.claude/commands/automation/smart-agents.md

@@ -1,5 +1,6 @@
 ---
 name: automation:smart-agents
+description: Automatically spawn the right agents at the right time without manual intervention.
 ---
 
 # Smart Agent Auto-Spawning

package/.claude/commands/automation/smart-spawn.md

@@ -1,5 +1,6 @@
 ---
 name: automation:smart-spawn
+description: Automatically spawn agents at the start of a task using the hooks pre-task system.
 ---
 
 # smart-spawn

package/.claude/commands/automation/workflow-select.md

@@ -1,60 +1,55 @@
 ---
 name: automation:workflow-select
+description: Pick a predefined multi-agent workflow for a common task.
 ---
 
 # workflow-select
 
-Run a predefined workflow template for common tasks.
+Pick a predefined multi-agent workflow for a common task. The CLI has no `workflow` command; predefined workflows are org starter templates, run with `monomind org`.
 
-## Usage
+## Templates
 
-```bash
-npx monomind workflow run [options]
-```
-
-## Options
-
-- `--template, -t <name>` - Workflow template name
-- `--task <description>` - Task description (for template selection)
-- `--parallel` - Enable parallel agent execution
-- `--max-agents <n>` - Max agents to spawn (default: 4)
-- `--dry-run` - Preview workflow without executing
+| Template | Use for |
+|---|---|
+| `dev-team` | Feature development: plan, implement, test, review |
+| `research-pod` | Research and analysis |
+| `content-team` | Content production |
+| `kg-extraction` | Knowledge-graph extraction from documents |
+| `advisor-orchestrator` | An orchestrator that consults advisor roles |
 
 ## Examples
 
-### List available templates
+### Create an org from a template
 
 ```bash
-npx monomind workflow template list
-```
-
-### Run a template
-
-```bash
-npx monomind workflow run --template feature-development --task "Add OAuth login"
+npx monomind org create oauth --template dev-team --goal "Add OAuth login"
 ```
 
 ### Preview without executing
 
 ```bash
-npx monomind workflow run --template deploy --dry-run
+npx monomind org validate oauth
+npx monomind org run oauth --dry-run
 ```
 
-### Run in parallel
+### Run it
 
 ```bash
-npx monomind workflow run --template code-review --parallel --max-agents 6
+npx monomind org run oauth --task "Add OAuth login" --budget-usd 5
 ```
 
 ## Workflow Status
 
 ```bash
-npx monomind workflow list
-npx monomind workflow status --watch
+npx monomind org list
+npx monomind org status oauth
+npx monomind org logs oauth
 ```
 
+For a one-off workflow inside the conversation, skip the org and spawn the stage's agents with the Task tool instead (see `/workflows:workflow-execute`).
+
 ## See Also
 
 - `auto-agent` — spawn agents without a template
 - `smart-spawn` — auto-select agents from task description
-- `swarm init` — manual swarm initialization
+- `monoswarm init` — manual monoswarm initialization

package/.claude/commands/coordination/README.md

@@ -8,6 +8,6 @@ Commands and guidance for swarm coordination and task orchestration in Monomind.
 
 ## Commands (invoke as slash commands)
 
-- [swarm-init](./swarm-init.md) — initialize a swarm with topology and strategy (`monomind swarm init`)
+- [swarm-init](./swarm-init.md) — initialize a swarm with topology and strategy (`monomind monoswarm init`)
 - [agent-spawn](./agent-spawn.md) — spawn a new agent in the swarm (`monomind agent spawn`)
-- [task-orchestrate](./task-orchestrate.md) — coordinate tasks across swarm agents (`monomind swarm init`)
+- [task-orchestrate](./task-orchestrate.md) — coordinate tasks across swarm agents (`monomind monoswarm init`)

package/.claude/commands/coordination/agent-spawn.md

@@ -1,5 +1,6 @@
 ---
 name: coordination:agent-spawn
+description: Spawn a new agent in the current swarm.
 ---
 
 # agent-spawn

package/.claude/commands/coordination/monoswarm-init.md

@@ -1,5 +1,6 @@
 ---
 name: coordination:monoswarm-init
+description: Initialize a Monomind swarm with specified topology and configuration.
 ---
 
 # monoswarm-init
@@ -9,7 +10,7 @@ Initialize a Monomind swarm with specified topology and configuration.
 ## Usage
 
 ```bash
-npx monomind swarm init [options]
+npx monomind monoswarm init [options]
 ```
 
 ## Options
@@ -25,31 +26,31 @@ npx monomind swarm init [options]
 ### Default hierarchical swarm
 
 ```bash
-npx monomind swarm init
+npx monomind monoswarm init
 ```
 
 ### Mesh topology for research tasks
 
 ```bash
-npx monomind swarm init --topology mesh --max-agents 5 --strategy balanced
+npx monomind monoswarm init --topology mesh --max-agents 5 --strategy balanced
 ```
 
 ### Specialized development swarm
 
 ```bash
-npx monomind swarm init --topology hierarchical --max-agents 8 --strategy specialized
+npx monomind monoswarm init --topology hierarchical --max-agents 8 --strategy specialized
 ```
 
 ### Auto-scaling adaptive swarm
 
 ```bash
-npx monomind swarm init --topology adaptive --auto-scale
+npx monomind monoswarm init --topology adaptive --auto-scale
 ```
 
 ### v1 high-performance mode (15 agents)
 
 ```bash
-npx monomind swarm init --v1-mode
+npx monomind monoswarm init --v1-mode
 ```
 
 ## Topologies
@@ -77,6 +78,6 @@ mcp__monomind__swarm_init({
 ## See Also
 
 - `agent-spawn` — create agents within the swarm
-- `swarm status` — check swarm state (`monomind swarm status`)
-- `swarm start` — start execution (`monomind swarm start --objective "..."`)
+- `swarm status` — check swarm state (`monomind monoswarm status`)
+- `swarm start` — start execution (`monomind monoswarm start --objective "..."`)
 - `task-orchestrate` — coordinate tasks across agents

package/.claude/commands/coordination/task-orchestrate.md

@@ -1,5 +1,6 @@
 ---
 name: coordination:task-orchestrate
+description: Coordinate tasks across a swarm of agents.
 ---
 
 # task-orchestrate
@@ -9,7 +10,7 @@ Coordinate tasks across a swarm of agents.
 ## Usage
 
 ```bash
-npx monomind swarm init [options]
+npx monomind monoswarm init [options]
 ```
 
 ## Options
@@ -24,19 +25,19 @@ npx monomind swarm init [options]
 ### Coordinate in parallel (default)
 
 ```bash
-npx monomind swarm init --task "Run all tests" --strategy parallel
+npx monomind monoswarm init --task "Run all tests" --strategy parallel
 ```
 
 ### Sequential pipeline
 
 ```bash
-npx monomind swarm init --task "Build, test, and deploy" --strategy pipeline
+npx monomind monoswarm init --task "Build, test, and deploy" --strategy pipeline
 ```
 
 ### Broadcast to all agents
 
 ```bash
-npx monomind swarm init --task "Update shared context" --strategy broadcast
+npx monomind monoswarm init --task "Update shared context" --strategy broadcast
 ```
 
 ## Task Lifecycle (separate from orchestration)
@@ -76,4 +77,4 @@ mcp__monomind__coordination_orchestrate({
 
 - `swarm-init` — initialize the swarm before orchestrating
 - `agent-spawn` — create agents to orchestrate tasks across
-- `swarm status` — check current swarm state (`monomind swarm status`)
+- `swarm status` — check current swarm state (`monomind monoswarm status`)

package/.claude/commands/github/README.md

@@ -4,7 +4,7 @@ name: github:README
 
 # GitHub Commands
 
-Commands and guidance for GitHub workflow automation in Monomind. All GitHub operations use the `gh` CLI and real monomind MCP tools — there is no `monomind github` CLI command group.
+Commands and guidance for GitHub workflow automation in Monomind. All GitHub operations use the `gh` CLI and real monomind MCP tools — the `monomind` CLI has no `github` command group.
 
 ## Commands (invoke as slash commands)
 

package/.claude/commands/github/github-modes.md

@@ -1,5 +1,6 @@
 ---
 name: github:github-modes
+description: Reference of the GitHub integration modes available in Monomind.
 ---
 
 # GitHub Integration Modes

package/.claude/commands/github/issue-tracker.md

@@ -1,5 +1,6 @@
 ---
 name: github:issue-tracker
+description: Intelligent issue management and project coordination with monomind swarm integration for automated tracking, progress monitoring, and team coordination.
 ---
 
 # GitHub Issue Tracker

package/.claude/commands/github/pr-manager.md

@@ -1,5 +1,6 @@
 ---
 name: github:pr-manager
+description: Comprehensive pull request management with monomind swarm coordination for automated reviews, testing, and merge workflows.
 ---
 
 # GitHub PR Manager

package/.claude/commands/github/release-manager.md

@@ -1,5 +1,6 @@
 ---
 name: github:release-manager
+description: Automated release coordination and deployment with Monomind swarm orchestration for seamless version management, testing, and deployment across multiple packages.
 ---
 
 # GitHub Release Manager
@@ -322,8 +323,6 @@ jobs:
           node-version: "20"
       - name: Install and Test
         run: npm install && npm test
-      - name: Validate Release
-        run: npx monomind release validate
 ```
 
 ## Monitoring and Metrics

package/.claude/commands/github/repo-architect.md

@@ -1,5 +1,6 @@
 ---
 name: github:repo-architect
+description: Repository structure optimization and multi-repo management with Monomind swarm coordination for scalable project architecture and development workflows.
 ---
 
 # GitHub Repository Architect

package/.claude/commands/github/sync-coordinator.md

@@ -1,5 +1,6 @@
 ---
 name: github:sync-coordinator
+description: Multi-package synchronization and version alignment with monomind swarm coordination for seamless integration across packages.
 ---
 
 # GitHub Sync Coordinator

package/.claude/commands/hooks/overview.md

@@ -1,5 +1,6 @@
 ---
 name: hooks:overview
+description: Self-learning hooks for intelligent workflow automation.
 ---
 
 # Hooks System Overview

package/.claude/commands/hooks/post-edit.md

@@ -1,5 +1,6 @@
 ---
 name: hooks:post-edit
+description: Record editing outcome for neural pattern learning.
 ---
 
 # hooks post-edit

package/.claude/commands/hooks/post-task.md

@@ -1,5 +1,6 @@
 ---
 name: hooks:post-task
+description: Record task completion for neural pattern learning.
 ---
 
 # hooks post-task

package/.claude/commands/hooks/pre-edit.md

@@ -1,5 +1,6 @@
 ---
 name: hooks:pre-edit
+description: Get context and agent suggestions before editing a file.
 ---
 
 # hooks pre-edit

package/.claude/commands/hooks/pre-task.md

@@ -1,5 +1,6 @@
 ---
 name: hooks:pre-task
+description: Register task start and get agent suggestions.
 ---
 
 # hooks pre-task

package/.claude/commands/hooks/session-end.md

@@ -1,5 +1,6 @@
 ---
 name: hooks:session-end
+description: End the current session and persist state for later restoration.
 ---
 
 # hooks session-end

package/.claude/commands/hooks/setup.md

@@ -1,5 +1,6 @@
 ---
 name: hooks:setup
+description: Set up Monomind hooks — generate the .claude/settings.json hook configuration and verify it.
 ---
 
 # Setting Up Monomind Hooks

package/.claude/commands/mastermind/adr.md

@@ -1,4 +1,6 @@
-<!-- Draft an Architecture Decision Record from accumulated decision markers in this session's prompts -->
+---
+description: Draft an Architecture Decision Record from accumulated decision markers in this session's prompts
+---
 
 Run the hook handler's adr-draft action to scan `.monomind/decisions.jsonl` for the last 7 days of decision markers (e.g. "let's go with X", "we chose Y", "decision: Z") and produce an ADR template in `docs/adrs/`.
 

package/.claude/commands/mastermind/brain.md

@@ -1,4 +1,6 @@
-<!-- Mastermind brain management — inspect, compact, refresh, and control the three-tier business memory (memory store + Monograph) -->
+---
+description: Mastermind brain management — inspect, compact, refresh, and control the three-tier business memory (memory store + Monograph)
+---
 
 **First — extract repeat flags:** Follow the REPEAT PREAMBLE from `mastermind-repeat/SKILL.md`. Extracts `--repeat`, `--tillend`, `--maxruns`, `--wait`, `--rep`, `--loop` from `$ARGUMENTS` before all other parsing.
 

package/.claude/commands/mastermind/budget.md

@@ -1,4 +1,6 @@
-<!-- Show current budget status — today, month, limits, autotuned flag -->
+---
+description: Show current budget status — today, month, limits, autotuned flag
+---
 
 ```bash
 node "$CLAUDE_PROJECT_DIR/.claude/helpers/hook-handler.cjs" budget-status

package/.claude/commands/mastermind/code-review.md

@@ -1,4 +1,6 @@
-<!-- "Monomind — Multi-agent iterative review loop: runs reviewer, Security Engineer, and domain specialists in parallel, auto-fixes findings each iteration, and captures human-in-loop items to a dated file." -->
+---
+description: "Monomind — Multi-agent iterative review loop: runs reviewer, Security Engineer, and domain specialists in parallel, auto-fixes findings each iteration, and captures human-in-loop items to a dated file."
+---
 
 **First — extract repeat flags:** Follow the REPEAT PREAMBLE from `mastermind-repeat/SKILL.md`. Extracts `--repeat`, `--tillend`, `--maxruns`, `--wait`, `--rep`, `--loop` from `$ARGUMENTS` before all other parsing. If `is_continuation = true`, skip the empty-arguments check below.
 

package/.claude/commands/mastermind/content.md

@@ -1,4 +1,6 @@
-<!-- Mastermind content domain — blog posts, threads, docs, newsletters. Default mode: confirm. -->
+---
+description: "Mastermind content domain — blog posts, threads, docs, newsletters. Default mode: confirm."
+---
 
 **First — extract repeat flags:** Follow the REPEAT PREAMBLE from `mastermind-repeat/SKILL.md`. Extracts `--repeat`, `--tillend`, `--maxruns`, `--wait`, `--rep`, `--loop` from `$ARGUMENTS` before all other parsing. If `is_continuation = true`, skip the empty-prompt check and intake below.
 

package/.claude/commands/mastermind/createorg.md

@@ -1,4 +1,6 @@
-<!-- Define and save an autonomous agent organization (Org Runtime v2) — roles and hierarchy. Suggest or confirm roles, then persist the org config for `monomind org run`/`serve`. -->
+---
+description: Define and save an autonomous agent organization (Org Runtime) — roles and hierarchy. Suggest or confirm roles, then persist the org config for `monomind org run`/`serve`.
+---
 
 **If $ARGUMENTS is empty:** Output the following and wait.
 
@@ -6,7 +8,7 @@
 
 **MASTERMIND: CREATE ORG**
 
-An org is a named, persistent agent team, run by the Org Runtime v2 daemon — not a one-shot Mastermind run and not a Task-tool-spawned boss. Once created, every role in the config starts as its own live agent session the moment you run the org; roles message each other directly (no shared task board) and the org keeps running until stopped.
+An org is a named, persistent agent team, run by the Org Runtime daemon — not a one-shot Mastermind run and not a Task-tool-spawned boss. Once created, every role in the config starts as its own live agent session the moment you run the org; roles message each other directly (no shared task board) and the org keeps running until stopped.
 
 Use orgs when the work is ongoing, not single-shot. A content team that ships 10 posts a month. A research squad that runs competitive scans weekly. A dev team with a permanent backlog.
 

package/.claude/commands/mastermind/createtask.md

@@ -1,4 +1,6 @@
-<!-- "Mastermind — Ingest a prompt, file, or folder, deeply understand it, generate agent-optimized tasks saved to docs/tasks/ (default) or to a monotask board (--monotask flag)" -->
+---
+description: Mastermind — Ingest a prompt, file, or folder, deeply understand it, generate agent-optimized tasks saved to docs/tasks/ (default) or to a monotask board (--monotask flag)
+---
 
 If `$ARGUMENTS` is empty, output this and STOP:
 
@@ -55,7 +57,7 @@ Bundle everything into `FULL_CONTEXT`.
 
 **File mode (default, `USE_MONOTASK=false`):**
 
-Compute the output path (follow slug rules from `mastermind-taskfile/SKILL.md`):
+Compute the output path (follow slug rules from `.claude/commands/mastermind/_taskfile.md`):
 ```bash
 DATE=$(date +%Y-%m-%d)
 SLUG=$(echo "$INPUT_LABEL" | tr '[:upper:]' '[:lower:]' | sed 's/[^a-z0-9]/-/g; s/-\+/-/g; s/^-//; s/-$//' | cut -c1-40)
@@ -161,7 +163,7 @@ Store as `TASKS` array.
 
 **File mode (default, `USE_MONOTASK=false`):**
 
-Write the file using the Write tool. Follow the task section format from `mastermind-taskfile/SKILL.md` exactly. The file structure:
+Write the file using the Write tool. Follow the task section format from `.claude/commands/mastermind/_taskfile.md` exactly. The file structure:
 
 1. **Write frontmatter + header** (once):
 ```markdown
@@ -180,7 +182,7 @@ recommended_mode: <parallel|minimal|sequential>
 ---
 ```
 
-2. **Append one section per task** (initial status = `todo` if no prerequisites, `backlog` if has prerequisites), following the task section format from `mastermind-taskfile/SKILL.md`:
+2. **Append one section per task** (initial status = `todo` if no prerequisites, `backlog` if has prerequisites), following the task section format from `.claude/commands/mastermind/_taskfile.md`:
 ```markdown
 ## Task N: <title>
 

package/.claude/commands/mastermind/debug.md

@@ -1,4 +1,6 @@
-<!-- Systematic root-cause debugging — use before ANY fix attempt for bugs, test failures, unexpected behavior, build failures, or performance regressions. Enforces Phase 1 root-cause investigation before proposing fixes. -->
+---
+description: Systematic root-cause debugging — use before ANY fix attempt for bugs, test failures, unexpected behavior, build failures, or performance regressions. Enforces Phase 1 root-cause investigation before proposing fixes.
+---
 
 **First — extract repeat flags:** Follow the REPEAT PREAMBLE from `mastermind-repeat/SKILL.md`.
 

package/.claude/commands/mastermind/design.md

@@ -1,4 +1,6 @@
-<!-- Collaborative design session — explore intent, clarify requirements, propose approaches, and produce an approved spec before any implementation begins. -->
+---
+description: Collaborative design session — explore intent, clarify requirements, propose approaches, and produce an approved spec before any implementation begins.
+---
 
 **First — extract repeat flags:** Follow REPEAT PREAMBLE from `mastermind-repeat/SKILL.md`.