@claude-flow/cli 3.0.0-alpha.161 β†’ 3.0.0-alpha.162

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/README.md CHANGED
@@ -1,7 +1,10 @@
1
- # Claude-Flow v3: Enterprise AI Orchestration Platform
1
+ # 🌊 Claude-Flow v3: Enterprise AI Orchestration Platform
2
2
 
3
3
  <div align="center">
4
4
 
5
+ ![Claude-Flow Banner](https://repository-images.githubusercontent.com/995029641/b9acbe16-0f49-420d-804f-468ba2a73ace)
6
+
7
+
5
8
 
6
9
  [![Star on GitHub](https://img.shields.io/github/stars/ruvnet/claude-flow?style=for-the-badge&logo=github&color=gold)](https://github.com/ruvnet/claude-flow)
7
10
  [![Monthly Downloads](https://img.shields.io/npm/dm/claude-flow?style=for-the-badge&logo=npm&color=blue&label=Monthly%20Downloads)](https://www.npmjs.com/package/claude-flow)
@@ -16,7 +19,7 @@
16
19
 
17
20
  **Production-ready multi-agent AI orchestration for Claude Code**
18
21
 
19
- *Deploy 54+ specialized agents in coordinated swarms with self-learning capabilities, fault-tolerant consensus, and enterprise-grade security.*
22
+ *Deploy 60+ specialized agents in coordinated swarms with self-learning capabilities, fault-tolerant consensus, and enterprise-grade security.*
20
23
 
21
24
  </div>
22
25
 
@@ -59,7 +62,7 @@ flowchart TB
59
62
  CLM[Claims<br/>Human-Agent Coord]
60
63
  end
61
64
 
62
- subgraph AGENTS["πŸ€– 54+ Agents"]
65
+ subgraph AGENTS["πŸ€– 60+ Agents"]
63
66
  AG1[coder]
64
67
  AG2[tester]
65
68
  AG3[reviewer]
@@ -152,7 +155,7 @@ npx claude-flow@v3alpha init
152
155
  ---
153
156
  ### Key Capabilities
154
157
 
155
- πŸ€– **54+ Specialized Agents** - Ready-to-use AI agents for coding, code review, testing, security audits, documentation, and DevOps. Each agent is optimized for its specific role.
158
+ πŸ€– **60+ Specialized Agents** - Ready-to-use AI agents for coding, code review, testing, security audits, documentation, and DevOps. Each agent is optimized for its specific role.
156
159
 
157
160
  🐝 **Coordinated Agent Teams** - Run unlimited agents simultaneously in organized swarms. Agents spawn sub-workers, communicate, share context, and divide work automatically using hierarchical (queen/workers) or mesh (peer-to-peer) patterns.
158
161
 
@@ -177,7 +180,7 @@ Every request flows through four layers: from your CLI or Claude Code interface,
177
180
  |-------|------------|--------------|
178
181
  | User | Claude Code, CLI | Your interface to control and run commands |
179
182
  | Orchestration | MCP Server, Router, Hooks | Routes requests to the right agents |
180
- | Agents | 54+ types | Specialized workers (coder, tester, reviewer...) |
183
+ | Agents | 60+ types | Specialized workers (coder, tester, reviewer...) |
181
184
  | Providers | Anthropic, OpenAI, Google, Ollama | AI models that power reasoning |
182
185
 
183
186
  </details>
@@ -257,6 +260,124 @@ Smart routing skips expensive LLM calls when possible. Simple edits use WASM (fr
257
260
 
258
261
  </details>
259
262
 
263
+ <details>
264
+ <summary>⚑ <strong>Agent Booster (WASM)</strong> β€” 352x faster code transforms, skip LLM entirely</summary>
265
+
266
+ Agent Booster uses WebAssembly to handle simple code transformations without calling the LLM at all. When the hooks system detects a simple task, it routes directly to Agent Booster for instant results.
267
+
268
+ **Supported Transform Intents:**
269
+
270
+ | Intent | What It Does | Example |
271
+ |--------|--------------|---------|
272
+ | `var-to-const` | Convert var/let to const | `var x = 1` β†’ `const x = 1` |
273
+ | `add-types` | Add TypeScript type annotations | `function foo(x)` β†’ `function foo(x: string)` |
274
+ | `add-error-handling` | Wrap in try/catch | Adds proper error handling |
275
+ | `async-await` | Convert promises to async/await | `.then()` chains β†’ `await` |
276
+ | `add-logging` | Add console.log statements | Adds debug logging |
277
+ | `remove-console` | Strip console.* calls | Removes all console statements |
278
+
279
+ **Hook Signals:**
280
+
281
+ When you see these in hook output, the system is telling you how to optimize:
282
+
283
+ ```bash
284
+ # Agent Booster available - skip LLM entirely
285
+ [AGENT_BOOSTER_AVAILABLE] Intent: var-to-const
286
+ β†’ Use Edit tool directly, 352x faster than LLM
287
+
288
+ # Model recommendation for Task tool
289
+ [TASK_MODEL_RECOMMENDATION] Use model="haiku"
290
+ β†’ Pass model="haiku" to Task tool for cost savings
291
+ ```
292
+
293
+ **Performance:**
294
+
295
+ | Metric | Agent Booster | LLM Call |
296
+ |--------|---------------|----------|
297
+ | Latency | <1ms | 2-5s |
298
+ | Cost | $0 | $0.0002-$0.015 |
299
+ | Speedup | **352x faster** | baseline |
300
+
301
+ </details>
302
+
303
+ <details>
304
+ <summary>πŸ’° <strong>Token Optimizer</strong> β€” 30-50% token reduction</summary>
305
+
306
+ The Token Optimizer integrates agentic-flow optimizations to reduce API costs by compressing context and caching results.
307
+
308
+ **Savings Breakdown:**
309
+
310
+ | Optimization | Token Savings | How It Works |
311
+ |--------------|---------------|--------------|
312
+ | ReasoningBank retrieval | -32% | Fetches relevant patterns instead of full context |
313
+ | Agent Booster edits | -15% | Simple edits skip LLM entirely |
314
+ | Cache (95% hit rate) | -10% | Reuses embeddings and patterns |
315
+ | Optimal batch size | -20% | Groups related operations |
316
+ | **Combined** | **30-50%** | Stacks multiplicatively |
317
+
318
+ **Usage:**
319
+
320
+ ```typescript
321
+ import { getTokenOptimizer } from '@claude-flow/integration';
322
+ const optimizer = await getTokenOptimizer();
323
+
324
+ // Get compact context (32% fewer tokens)
325
+ const ctx = await optimizer.getCompactContext("auth patterns");
326
+
327
+ // Optimized edit (352x faster for simple transforms)
328
+ await optimizer.optimizedEdit(file, oldStr, newStr, "typescript");
329
+
330
+ // Optimal config for swarm (100% success rate)
331
+ const config = optimizer.getOptimalConfig(agentCount);
332
+ ```
333
+
334
+ </details>
335
+
336
+ <details>
337
+ <summary>πŸ›‘οΈ <strong>Anti-Drift Swarm Configuration</strong> β€” Prevent goal drift in multi-agent work</summary>
338
+
339
+ Complex swarms can drift from their original goals. Claude-Flow V3 includes anti-drift defaults that prevent agents from going off-task.
340
+
341
+ **Recommended Configuration:**
342
+
343
+ ```javascript
344
+ // Anti-drift defaults (ALWAYS use for coding tasks)
345
+ swarm_init({
346
+ topology: "hierarchical", // Single coordinator enforces alignment
347
+ maxAgents: 8, // Smaller team = less drift surface
348
+ strategy: "specialized" // Clear roles reduce ambiguity
349
+ })
350
+ ```
351
+
352
+ **Why This Prevents Drift:**
353
+
354
+ | Setting | Anti-Drift Benefit |
355
+ |---------|-------------------|
356
+ | `hierarchical` | Coordinator validates each output against goal, catches divergence early |
357
+ | `maxAgents: 6-8` | Fewer agents = less coordination overhead, easier alignment |
358
+ | `specialized` | Clear boundaries - each agent knows exactly what to do, no overlap |
359
+ | `raft` consensus | Leader maintains authoritative state, no conflicting decisions |
360
+
361
+ **Additional Anti-Drift Measures:**
362
+
363
+ - Frequent checkpoints via `post-task` hooks
364
+ - Shared memory namespace for all agents
365
+ - Short task cycles with verification gates
366
+ - Hierarchical coordinator reviews all outputs
367
+
368
+ **Task β†’ Agent Routing (Anti-Drift):**
369
+
370
+ | Code | Task Type | Recommended Agents |
371
+ |------|-----------|-------------------|
372
+ | 1 | Bug Fix | coordinator, researcher, coder, tester |
373
+ | 3 | Feature | coordinator, architect, coder, tester, reviewer |
374
+ | 5 | Refactor | coordinator, architect, coder, reviewer |
375
+ | 7 | Performance | coordinator, perf-engineer, coder |
376
+ | 9 | Security | coordinator, security-architect, auditor |
377
+ | 11 | Memory | coordinator, memory-specialist, perf-engineer |
378
+
379
+ </details>
380
+
260
381
  ### Claude Code: With vs Without Claude-Flow
261
382
 
262
383
  | Capability | Claude Code Alone | Claude Code + Claude-Flow |
@@ -402,7 +523,7 @@ Claude-Flow v3 introduces **self-learning neural capabilities** that no other ag
402
523
  | **Code Transforms** | βœ… Agent Booster (352x) | β›” | β›” | β›” | β›” |
403
524
  | **Input Validation** | βœ… Zod + Path security | β›” | β›” | β›” | β›” |
404
525
 
405
- <sub>*Comparison updated January 15, 2026*</sub>
526
+ <sub>*Comparison updated January 23, 2026*</sub>
406
527
 
407
528
  <details>
408
529
  <summary>πŸš€ <strong>Key Differentiators</strong> β€” Self-learning, memory optimization, fault tolerance</summary>
@@ -519,7 +640,7 @@ flowchart TB
519
640
 
520
641
  subgraph Agents["πŸ€– Agent Layer"]
521
642
  Queen[Queen Coordinator]
522
- Workers[54+ Specialized Agents]
643
+ Workers[60+ Specialized Agents]
523
644
  Swarm[Swarm Manager]
524
645
  end
525
646
 
@@ -703,7 +824,7 @@ Connect Claude-Flow to your development environment.
703
824
  <details>
704
825
  <summary>πŸ”Œ <strong>MCP Setup</strong> β€” Connect Claude-Flow to Any AI Environment</summary>
705
826
 
706
- Claude-Flow runs as an MCP (Model Context Protocol) server, allowing you to connect it to any MCP-compatible AI client. This means you can use Claude-Flow's 54+ agents, swarm coordination, and self-learning capabilities from Claude Desktop, VS Code, Cursor, Windsurf, ChatGPT, and more.
827
+ Claude-Flow runs as an MCP (Model Context Protocol) server, allowing you to connect it to any MCP-compatible AI client. This means you can use Claude-Flow's 60+ agents, swarm coordination, and self-learning capabilities from Claude Desktop, VS Code, Cursor, Windsurf, ChatGPT, and more.
707
828
 
708
829
  ### Quick Add Command
709
830
 
@@ -985,12 +1106,12 @@ echo "ANTHROPIC_API_KEY=sk-ant-..." >> .env
985
1106
  Comprehensive capabilities for enterprise-grade AI agent orchestration.
986
1107
 
987
1108
  <details>
988
- <summary>πŸ“¦ <strong>Features</strong> β€” 54+ Agents, Swarm Topologies, MCP Tools & Security</summary>
1109
+ <summary>πŸ“¦ <strong>Features</strong> β€” 60+ Agents, Swarm Topologies, MCP Tools & Security</summary>
989
1110
 
990
1111
  Comprehensive feature set for enterprise-grade AI agent orchestration.
991
1112
 
992
1113
  <details open>
993
- <summary>πŸ€– <strong>Agent Ecosystem</strong> β€” 54+ specialized agents across 8 categories</summary>
1114
+ <summary>πŸ€– <strong>Agent Ecosystem</strong> β€” 60+ specialized agents across 8 categories</summary>
994
1115
 
995
1116
  Pre-built agents for every development task, from coding to security audits.
996
1117
 
@@ -1134,6 +1255,40 @@ Build custom plugins with the fluent builder API. Create MCP tools, hooks, worke
1134
1255
 
1135
1256
  **Plugin Performance:** Load <20ms, Hook execution <0.5ms, Worker spawn <50ms
1136
1257
 
1258
+ ### πŸ“¦ Available Optional Plugins
1259
+
1260
+ Install these optional plugins to extend Claude Flow capabilities:
1261
+
1262
+ | Plugin | Version | Description | Install Command |
1263
+ |--------|---------|-------------|-----------------|
1264
+ | **@claude-flow/plugin-agentic-qe** | 3.0.0-alpha.2 | Quality Engineering with 58 AI agents across 12 DDD contexts. TDD, coverage analysis, security scanning, chaos engineering, accessibility testing. | `npm install @claude-flow/plugin-agentic-qe` |
1265
+ | **@claude-flow/plugin-prime-radiant** | 0.1.4 | Mathematical AI interpretability with 6 engines: sheaf cohomology, spectral analysis, causal inference, quantum topology, category theory, HoTT proofs. | `npm install @claude-flow/plugin-prime-radiant` |
1266
+
1267
+ **Agentic-QE Plugin Features:**
1268
+ - 58 specialized QE agents across 13 bounded contexts
1269
+ - 16 MCP tools: `aqe/generate-tests`, `aqe/tdd-cycle`, `aqe/analyze-coverage`, `aqe/security-scan`, `aqe/chaos-inject`, etc.
1270
+ - London-style TDD with red-green-refactor cycles
1271
+ - O(log n) coverage gap detection with Johnson-Lindenstrauss
1272
+ - OWASP/SANS compliance auditing
1273
+
1274
+ **Prime-Radiant Plugin Features:**
1275
+ - 6 mathematical engines for AI interpretability
1276
+ - 6 MCP tools: `pr_coherence_check`, `pr_spectral_analyze`, `pr_causal_infer`, `pr_consensus_verify`, `pr_quantum_topology`, `pr_memory_gate`
1277
+ - Sheaf Laplacian coherence detection (<5ms)
1278
+ - Do-calculus causal inference
1279
+ - Hallucination prevention via consensus verification
1280
+
1281
+ ```bash
1282
+ # Install Quality Engineering plugin
1283
+ npm install @claude-flow/plugin-agentic-qe
1284
+
1285
+ # Install AI Interpretability plugin
1286
+ npm install @claude-flow/plugin-prime-radiant
1287
+
1288
+ # Register plugins in your project
1289
+ npx claude-flow plugins list --installed
1290
+ ```
1291
+
1137
1292
  </details>
1138
1293
 
1139
1294
  <details>
@@ -1772,19 +1927,34 @@ Real-time development status display for Claude Code integration showing DDD pro
1772
1927
 
1773
1928
  **Usage:**
1774
1929
  ```bash
1775
- # V3 statusline (Node.js)
1776
- node v3/@claude-flow/hooks/bin/statusline.js
1930
+ # Default: Safe multi-line (avoids Claude Code collision zone)
1931
+ npx claude-flow@v3alpha hooks statusline
1932
+
1933
+ # Single-line mode (completely avoids collision)
1934
+ npx claude-flow@v3alpha hooks statusline --single
1935
+
1936
+ # Legacy multi-line (original behavior, may have bleeding)
1937
+ npx claude-flow@v3alpha hooks statusline --legacy
1777
1938
 
1778
1939
  # JSON output for scripting
1779
- node v3/@claude-flow/hooks/bin/statusline.js --json
1940
+ npx claude-flow@v3alpha hooks statusline --json
1780
1941
 
1781
1942
  # Compact JSON (single line)
1782
- node v3/@claude-flow/hooks/bin/statusline.js --compact
1783
-
1784
- # Help
1785
- node v3/@claude-flow/hooks/bin/statusline.js --help
1943
+ npx claude-flow@v3alpha hooks statusline --compact
1786
1944
  ```
1787
1945
 
1946
+ **Collision Zone Fix (Issue #985):**
1947
+
1948
+ Claude Code writes internal status (e.g., `7s β€’ 1p`) at absolute terminal coordinates (columns 15-25 on the second-to-last line). The safe mode pads the collision line with spaces to push content past column 25, preventing character bleeding.
1949
+
1950
+ | Option | Description |
1951
+ |--------|-------------|
1952
+ | (default) | Safe multi-line with collision zone avoidance |
1953
+ | `--single` | Single-line output (complete collision avoidance) |
1954
+ | `--legacy` | Original multi-line (may cause bleeding) |
1955
+ | `--json` | JSON output with pretty printing |
1956
+ | `--compact` | JSON output without formatting |
1957
+
1788
1958
  **Claude Code Integration:**
1789
1959
 
1790
1960
  Add to `.claude/settings.json`:
@@ -1792,7 +1962,7 @@ Add to `.claude/settings.json`:
1792
1962
  {
1793
1963
  "statusLine": {
1794
1964
  "type": "command",
1795
- "command": "node v3/@claude-flow/hooks/bin/statusline.js"
1965
+ "command": "npx claude-flow@v3alpha hooks statusline --single"
1796
1966
  }
1797
1967
  }
1798
1968
  ```
@@ -1907,6 +2077,189 @@ Shell-based daemons for monitoring (Linux/macOS only):
1907
2077
  .claude/helpers/worker-manager.sh status
1908
2078
  ```
1909
2079
 
2080
+ </details>
2081
+
2082
+ <details>
2083
+ <summary>⌨️ <strong>V3 CLI Commands</strong> β€” 26 commands with 140+ subcommands</summary>
2084
+
2085
+ Complete command-line interface for all Claude-Flow operations.
2086
+
2087
+ **Core Commands:**
2088
+
2089
+ | Command | Subcommands | Description |
2090
+ |---------|-------------|-------------|
2091
+ | `init` | 4 | Project initialization with wizard, presets, skills, hooks |
2092
+ | `agent` | 8 | Agent lifecycle (spawn, list, status, stop, metrics, pool, health, logs) |
2093
+ | `swarm` | 6 | Multi-agent swarm coordination and orchestration |
2094
+ | `memory` | 11 | AgentDB memory with vector search (150x-12,500x faster) |
2095
+ | `mcp` | 9 | MCP server management and tool execution |
2096
+ | `task` | 6 | Task creation, assignment, and lifecycle |
2097
+ | `session` | 7 | Session state management and persistence |
2098
+ | `config` | 7 | Configuration management and provider setup |
2099
+ | `status` | 3 | System status monitoring with watch mode |
2100
+ | `start` | 3 | Service startup and quick launch |
2101
+ | `workflow` | 6 | Workflow execution and template management |
2102
+ | `hooks` | 17 | Self-learning hooks + 12 background workers |
2103
+ | `hive-mind` | 6 | Queen-led Byzantine fault-tolerant consensus |
2104
+
2105
+ **Advanced Commands:**
2106
+
2107
+ | Command | Subcommands | Description |
2108
+ |---------|-------------|-------------|
2109
+ | `daemon` | 5 | Background worker daemon (start, stop, status, trigger, enable) |
2110
+ | `neural` | 5 | Neural pattern training (train, status, patterns, predict, optimize) |
2111
+ | `security` | 6 | Security scanning (scan, audit, cve, threats, validate, report) |
2112
+ | `performance` | 5 | Performance profiling (benchmark, profile, metrics, optimize, report) |
2113
+ | `providers` | 5 | AI providers (list, add, remove, test, configure) |
2114
+ | `plugins` | 5 | Plugin management (list, install, uninstall, enable, disable) |
2115
+ | `deployment` | 5 | Deployment management (deploy, rollback, status, environments, release) |
2116
+ | `embeddings` | 4 | Vector embeddings (embed, batch, search, init) - 75x faster with agentic-flow |
2117
+ | `claims` | 4 | Claims-based authorization (check, grant, revoke, list) |
2118
+ | `migrate` | 5 | V2 to V3 migration with rollback support |
2119
+ | `process` | 4 | Background process management |
2120
+ | `doctor` | 1 | System diagnostics with health checks |
2121
+ | `completions` | 4 | Shell completions (bash, zsh, fish, powershell) |
2122
+
2123
+ **Quick Examples:**
2124
+
2125
+ ```bash
2126
+ # Initialize project with wizard
2127
+ npx claude-flow@v3alpha init --wizard
2128
+
2129
+ # Start daemon with background workers
2130
+ npx claude-flow@v3alpha daemon start
2131
+
2132
+ # Spawn an agent with specific type
2133
+ npx claude-flow@v3alpha agent spawn -t coder --name my-coder
2134
+
2135
+ # Initialize swarm with V3 mode
2136
+ npx claude-flow@v3alpha swarm init --v3-mode
2137
+
2138
+ # Search memory (HNSW-indexed, 150x faster)
2139
+ npx claude-flow@v3alpha memory search -q "authentication patterns"
2140
+
2141
+ # Run security scan
2142
+ npx claude-flow@v3alpha security scan --depth full
2143
+
2144
+ # Performance benchmark
2145
+ npx claude-flow@v3alpha performance benchmark --suite all
2146
+ ```
2147
+
2148
+ </details>
2149
+
2150
+ <details>
2151
+ <summary>🩺 <strong>Doctor Health Checks</strong> β€” System diagnostics with auto-fix</summary>
2152
+
2153
+ Run `npx claude-flow@v3alpha doctor` to diagnose and fix common issues.
2154
+
2155
+ **Health Checks Performed:**
2156
+
2157
+ | Check | Requirement | Auto-Fix |
2158
+ |-------|-------------|----------|
2159
+ | **Node.js version** | 20+ | ❌ Manual upgrade required |
2160
+ | **npm version** | 9+ | ❌ Manual upgrade required |
2161
+ | **Git installation** | Any version | ❌ Manual install required |
2162
+ | **Config file validity** | Valid JSON/YAML | βœ… Regenerates defaults |
2163
+ | **Daemon status** | Running | βœ… Restarts daemons |
2164
+ | **Memory database** | SQLite writable | βœ… Recreates if corrupt |
2165
+ | **API keys** | Valid format | ❌ Manual configuration |
2166
+ | **MCP servers** | Responsive | βœ… Restarts unresponsive servers |
2167
+ | **Disk space** | >100MB free | ❌ Manual cleanup required |
2168
+ | **TypeScript** | Installed | βœ… Installs if missing |
2169
+
2170
+ **Commands:**
2171
+
2172
+ ```bash
2173
+ # Run full diagnostics
2174
+ npx claude-flow@v3alpha doctor
2175
+
2176
+ # Run diagnostics with auto-fix
2177
+ npx claude-flow@v3alpha doctor --fix
2178
+
2179
+ # Check specific component
2180
+ npx claude-flow@v3alpha doctor --component memory
2181
+
2182
+ # Verbose output
2183
+ npx claude-flow@v3alpha doctor --verbose
2184
+ ```
2185
+
2186
+ **Output Example:**
2187
+
2188
+ ```
2189
+ 🩺 Claude-Flow Doctor v3.0.0-alpha
2190
+
2191
+ βœ… Node.js 20.11.0 (required: 20+)
2192
+ βœ… npm 10.2.4 (required: 9+)
2193
+ βœ… Git 2.43.0
2194
+ βœ… Config Valid claude-flow.config.json
2195
+ βœ… Daemon Running (PID: 12345)
2196
+ βœ… Memory SQLite healthy, 1.2MB
2197
+ ⚠️ API Keys ANTHROPIC_API_KEY set, OPENAI_API_KEY missing
2198
+ βœ… MCP Server Responsive (45ms latency)
2199
+ βœ… Disk Space 2.4GB available
2200
+
2201
+ Summary: 9/10 checks passed
2202
+ ```
2203
+
2204
+ </details>
2205
+
2206
+ <details>
2207
+ <summary>πŸ“¦ <strong>Embeddings Package v3</strong> β€” Cross-platform ONNX with hyperbolic support</summary>
2208
+
2209
+ The embeddings package (v3.0.0-alpha.12) provides high-performance vector embeddings with multiple backends.
2210
+
2211
+ **Key Features:**
2212
+
2213
+ | Feature | Description | Performance |
2214
+ |---------|-------------|-------------|
2215
+ | **sql.js backend** | Cross-platform SQLite (WASM) | No native compilation needed |
2216
+ | **Document chunking** | Configurable overlap and size | Handles large documents |
2217
+ | **Normalization** | L2, L1, min-max, z-score | 4 normalization methods |
2218
+ | **Hyperbolic embeddings** | PoincarΓ© ball model | Better hierarchical representation |
2219
+ | **agentic-flow ONNX** | Integrated ONNX runtime | 75x faster than API calls |
2220
+ | **Neural substrate** | RuVector integration | Full learning pipeline |
2221
+
2222
+ **Models Available:**
2223
+
2224
+ | Model | Dimensions | Speed | Quality |
2225
+ |-------|------------|-------|---------|
2226
+ | `all-MiniLM-L6-v2` | 384 | Fast | Good |
2227
+ | `all-mpnet-base-v2` | 768 | Medium | Better |
2228
+
2229
+ **Usage:**
2230
+
2231
+ ```bash
2232
+ # Initialize embeddings system
2233
+ npx claude-flow@v3alpha embeddings init
2234
+
2235
+ # Generate embedding for text
2236
+ npx claude-flow@v3alpha embeddings embed "authentication patterns"
2237
+
2238
+ # Batch embed multiple texts
2239
+ npx claude-flow@v3alpha embeddings batch --file texts.txt
2240
+
2241
+ # Search with semantic similarity
2242
+ npx claude-flow@v3alpha embeddings search "login flow" --top-k 5
2243
+ ```
2244
+
2245
+ **Programmatic:**
2246
+
2247
+ ```typescript
2248
+ import { createEmbeddingServiceAsync } from '@claude-flow/embeddings';
2249
+
2250
+ const service = await createEmbeddingServiceAsync({
2251
+ model: 'all-MiniLM-L6-v2',
2252
+ hyperbolic: true, // Enable PoincarΓ© ball embeddings
2253
+ cacheSize: 256
2254
+ });
2255
+
2256
+ // Generate embedding
2257
+ const embedding = await service.embed("authentication flow");
2258
+
2259
+ // Search similar patterns
2260
+ const results = await service.search("login", { topK: 5 });
2261
+ ```
2262
+
1910
2263
  </details>
1911
2264
  </details>
1912
2265
 
@@ -2012,6 +2365,35 @@ Hooks intercept operations (file edits, commands, tasks) and learn from outcomes
2012
2365
  150x faster success/fail compression memory lock
2013
2366
  ```
2014
2367
 
2368
+ ### Hook Signals (ADR-026 Model Routing)
2369
+
2370
+ When hooks run, they emit signals that guide routing decisions. Watch for these in hook output:
2371
+
2372
+ | Signal | Meaning | Action |
2373
+ |--------|---------|--------|
2374
+ | `[AGENT_BOOSTER_AVAILABLE]` | Simple transform detected, skip LLM | Use Edit tool directly (352x faster, $0) |
2375
+ | `[TASK_MODEL_RECOMMENDATION] Use model="haiku"` | Low complexity task | Pass `model: "haiku"` to Task tool |
2376
+ | `[TASK_MODEL_RECOMMENDATION] Use model="sonnet"` | Medium complexity task | Pass `model: "sonnet"` to Task tool |
2377
+ | `[TASK_MODEL_RECOMMENDATION] Use model="opus"` | High complexity task | Pass `model: "opus"` to Task tool |
2378
+
2379
+ **Agent Booster Intents** (handled without LLM):
2380
+ - `var-to-const` - Convert var/let to const
2381
+ - `add-types` - Add TypeScript type annotations
2382
+ - `add-error-handling` - Wrap in try/catch
2383
+ - `async-await` - Convert promises to async/await
2384
+ - `add-logging` - Add console.log statements
2385
+ - `remove-console` - Strip console.* calls
2386
+
2387
+ **Example Hook Output:**
2388
+ ```bash
2389
+ $ npx claude-flow@v3alpha hooks pre-task --description "convert var to const in utils.ts"
2390
+
2391
+ [AGENT_BOOSTER_AVAILABLE] Intent: var-to-const
2392
+ Recommendation: Use Edit tool directly
2393
+ Performance: <1ms (352x faster than LLM)
2394
+ Cost: $0
2395
+ ```
2396
+
2015
2397
  ### All 27 Hooks by Category
2016
2398
 
2017
2399
  #### πŸ”§ Tool Lifecycle Hooks (6 hooks)
@@ -5853,7 +6235,7 @@ export CLAUDE_FLOW_HNSW_EF=100
5853
6235
  β”‚ Pattern Matching β”‚ Self-learning (ReasoningBank) β”‚
5854
6236
  β”‚ Security β”‚ CVE remediation + strict validation β”‚
5855
6237
  β”‚ Modular Architecture β”‚ 18 @claude-flow/* packages β”‚
5856
- β”‚ Agent Coordination β”‚ 54+ specialized agents β”‚
6238
+ β”‚ Agent Coordination β”‚ 60+ specialized agents β”‚
5857
6239
  β”‚ Token Efficiency β”‚ 32% reduction with optimization β”‚
5858
6240
  β””β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”΄β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”€β”˜
5859
6241
  ```
@@ -1 +1 @@
1
- {"version":3,"file":"discovery.d.ts","sourceRoot":"","sources":["../../../../src/plugins/store/discovery.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAKH,OAAO,KAAK,EACV,cAAc,EACd,mBAAmB,EACnB,iBAAiB,EAElB,MAAM,YAAY,CAAC;AAGpB;;GAEG;AACH,eAAO,MAAM,2BAA2B,EAAE,iBAiCzC,CAAC;AAEF;;GAEG;AACH,MAAM,WAAW,qBAAqB;IACpC,OAAO,EAAE,OAAO,CAAC;IACjB,QAAQ,CAAC,EAAE,cAAc,CAAC;IAC1B,GAAG,CAAC,EAAE,MAAM,CAAC;IACb,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,SAAS,CAAC,EAAE,OAAO,CAAC;IACpB,KAAK,CAAC,EAAE,MAAM,CAAC;CAChB;AAED;;GAEG;AACH,qBAAa,sBAAsB;IACjC,OAAO,CAAC,MAAM,CAAoB;IAClC,OAAO,CAAC,KAAK,CAA2E;gBAE5E,MAAM,GAAE,OAAO,CAAC,iBAAiB,CAAM;IAInD;;OAEG;IACG,gBAAgB,CAAC,YAAY,CAAC,EAAE,MAAM,GAAG,OAAO,CAAC,qBAAqB,CAAC;IAqE7E;;OAEG;IACH,OAAO,CAAC,kBAAkB;IA8D1B;;OAEG;IACH,OAAO,CAAC,cAAc;IA+UtB;;OAEG;IACH,OAAO,CAAC,uBAAuB;IAQ/B;;OAEG;IACH,cAAc,IAAI,mBAAmB,EAAE;IAIvC;;OAEG;IACH,WAAW,CAAC,QAAQ,EAAE,mBAAmB,GAAG,IAAI;IAIhD;;OAEG;IACH,cAAc,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO;IASrC;;OAEG;IACH,UAAU,IAAI,IAAI;IAIlB;;OAEG;IACH,aAAa,IAAI;QAAE,OAAO,EAAE,MAAM,CAAC;QAAC,UAAU,EAAE,MAAM,EAAE,CAAA;KAAE;CAM3D;AAED;;GAEG;AACH,wBAAgB,4BAA4B,CAC1C,MAAM,CAAC,EAAE,OAAO,CAAC,iBAAiB,CAAC,GAClC,sBAAsB,CAExB"}
1
+ {"version":3,"file":"discovery.d.ts","sourceRoot":"","sources":["../../../../src/plugins/store/discovery.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AAKH,OAAO,KAAK,EACV,cAAc,EACd,mBAAmB,EACnB,iBAAiB,EAElB,MAAM,YAAY,CAAC;AAGpB;;GAEG;AACH,eAAO,MAAM,2BAA2B,EAAE,iBAiCzC,CAAC;AAEF;;GAEG;AACH,MAAM,WAAW,qBAAqB;IACpC,OAAO,EAAE,OAAO,CAAC;IACjB,QAAQ,CAAC,EAAE,cAAc,CAAC;IAC1B,GAAG,CAAC,EAAE,MAAM,CAAC;IACb,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,SAAS,CAAC,EAAE,OAAO,CAAC;IACpB,KAAK,CAAC,EAAE,MAAM,CAAC;CAChB;AAED;;GAEG;AACH,qBAAa,sBAAsB;IACjC,OAAO,CAAC,MAAM,CAAoB;IAClC,OAAO,CAAC,KAAK,CAA2E;gBAE5E,MAAM,GAAE,OAAO,CAAC,iBAAiB,CAAM;IAInD;;OAEG;IACG,gBAAgB,CAAC,YAAY,CAAC,EAAE,MAAM,GAAG,OAAO,CAAC,qBAAqB,CAAC;IAqE7E;;OAEG;IACH,OAAO,CAAC,kBAAkB;IA8D1B;;OAEG;IACH,OAAO,CAAC,cAAc;IAqctB;;OAEG;IACH,OAAO,CAAC,uBAAuB;IAQ/B;;OAEG;IACH,cAAc,IAAI,mBAAmB,EAAE;IAIvC;;OAEG;IACH,WAAW,CAAC,QAAQ,EAAE,mBAAmB,GAAG,IAAI;IAIhD;;OAEG;IACH,cAAc,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO;IASrC;;OAEG;IACH,UAAU,IAAI,IAAI;IAIlB;;OAEG;IACH,aAAa,IAAI;QAAE,OAAO,EAAE,MAAM,CAAC;QAAC,UAAU,EAAE,MAAM,EAAE,CAAA;KAAE;CAM3D;AAED;;GAEG;AACH,wBAAgB,4BAA4B,CAC1C,MAAM,CAAC,EAAE,OAAO,CAAC,iBAAiB,CAAC,GAClC,sBAAsB,CAExB"}