tachibot-mcp

- Tool descriptions no longer name-drop model versions.19 descriptions were rewritten to name the provider and the task instead of pinned strings like "GPT-5.2", "Kimi K3" or "424B MoE" — those silently rot on every model bump and then actively mislead the routing model. Two were outright wrong:openai_reasonandopenai_searchadvertised "GPT-5.2" while the code calls the current OpenAI flagship tier.
- grok_searchfinally says why to pick it.Its entire description was "Web search" — no way to choose it over the four other search tools. It now states its live X/news grounding edge and cross-referencesgrok_search_liteas the cheaper path. The five OpenRouter reasoners (deepseek_reason,glm_reason,stepfun_reason,ernie_reason,qwen_reason) likewise differed only by vendor trivia; each now carries an actual routing rule for when to pick it.
- Claude Desktop manifest corrected— it advertised 65 tools (actual: 66), omitted theheavy_codingprofile from theTACHIBOT_PROFILEhelp, and itsOPENAI_API_KEYhint claimed that key powers the Qwen and QwQ tools (those route via OpenRouter).
- Packaging no longer breaks the working tree.scripts/package-extension.shrannpm install --production, pruning devDependencies and leavingtscunable to build afterward. It now restores the full dependency tree when it finishes.
- The server introduced itself as v2.0.0.serverInfo.versionwas hardcoded from a 12-tool era ~28 minor versions ago, so Claude Desktop's connector panel misidentified every install. It now reads the real version frompackage.json. The/setupwizard's profile sizes were stale in the same way and now match the six real profiles.
- Correction to v2.30.0:grok_search_liteis not broken, and is staying.That release claimedgrok-4.3"does not invoke web search either" — which would have made the cheap search tier pointless. Re-probed Aug 15 asking today's date:grok-4.3 ran 2 web searches and returned a citation in 10.4s; grok-4.6 did the same in 36.4s.Lite is grounded, cheaper, and 3–4x faster — prefer it for high-volume lookups.grok-4.5is still the one that doesn't ground. The cheap tier now has its own grounding test, since nothing previously covered it.
- No tools added or removed — the count stays 66.

- grok_searchwas not searching.It ran ongrok-4.5, which never invokes theweb_searchtool on xAI's Agent Tools API — so it answered from training data while still rendering a source footer and a "Search used up to N sources" cost line computed locally frommax_search_results, not from real usage. Probed Aug 14: asked today's date,grok-4.5replied"October 10, 2025"with zeroweb_search_callentries;grok-4.6replied correctly with two search calls and a citation.grok_searchnow runsgrok-4.6(same $2/$6 and 500K context as 4.5). The regression test assertsgrounding— aweb_search_calland ≥1 annotation — never the answer text, because a plausible ungrounded answer is precisely what hid this.
- Correction to the v2.27.0 note below:grok_search_litenever ran ongrok-4-1-fastat $0.20/$0.50. xAI retired that id and silently servedgrok-4.3— HTTP 200, the swap disclosed only in the response body'smodelfield, so nothing threw and no fallback fired. The advertised "~10x cheaper" was really ~1.6x.
- 13console.logcalls were corrupting the MCP protocol stream.stdout is the JSON-RPC channel on a stdio server; diagnostics now go to stderr, guarded by a test that also rejectsconsole.info,console.debugandprocess.stdout.write.
- create_workflowsilently destroyed existing workflow files— the existence check guarded thedirectory, not thefile. It now refuses to clobber unlessoverwrite: true, and reports the real.tachibot/workflows/path instead of.tachi/workflows/.
- Eight prompt maps injected the literal string"undefined"into system prompts for any unrecognisedapproach/task, because the||fallback satinsidethe index. Fixed acrossgrok_reason,grok_code,kimi_thinking,qwen_reason,deepseek_reason,glm_reason,stepfun_reason,ernie_reason.
- Workflow steps ran the wrong model— theqwen_algostep called QwQ-32B with a generic prompt instead of Qwen3.8-Max, andqwq_reasonlost its 4-persona deliberation entirely. Both now delegate to the real tools. Separately,planner_maker'sqwen_coderstep sent a parameter the schema rejects, so every execution of it failed validation.
- Test suite130 → 158. The live search probe is now opt-in behindRUN_LIVE_TESTS=1, so an ordinarynpm testno longer makes a billed API call.

- Qwen3.8 Max(Aug 3,qwen/qwen3.8-max) — Alibaba's new flagship, GA the same day, now powersqwen_algo,qwen_reason, and theqwen_reasonjuror.1M context (up from 262K), multimodal (text+image+video in), and the first Qwen exposingconfigurable reasoning effort. $2/$6 per M.
- Measured on a range-query algorithm problem, 3.8 Max was the only one of five Qwen candidates to surface both the offline-vs-online tradeoff and the strict-inequality-with-duplicates edge case. The outgoingqwen3-235b-a22b-thinking-2507was correct but shallow;qwen3.7-maxcost 1.6x for twice the wall time;qwen3-max-thinkingwas rejected outright — it returns zero reasoning tokens.
- Effort is pinned tomedium, deliberately.This model'sdefaulteffort behaves likehigh: 302s and $0.09 on a singleqwen_algocall. Atmediumthe same call answers at equal depth in 18–48s for $0.006–0.018 — cheaperand3.5x faster than the model it replaces, which took 169s and $0.036 for a shorter answer.lowstarts dropping alternatives and is not used.
- reasoning_effortpass-through for OpenRouter—callOpenRouternow forwards the parameter; OpenRouter drops it for models that don't list it, so the quota fallback chain (3.8 Max → 3.7 Max → 235B Thinking) stays safe. Qwen3.8/3.7-Max also join the 600s extended-timeout bucket.
- qwen_coder,qwen_competitive,testgenstay on Qwen3-Coder-Next— it is coding-specialized and ~16x cheaper ($0.12/$0.80); 3.8 Max is the reasoning tier, not the codegen tier.

- Kimi K3(Jul 16,moonshotai/kimi-k3) now powers every Kimi tool, thekimijuror, and the Kimi seat ondiff_review. 2.8T open-weight MoE — the largest open model shipped — with a1M context (up from 262K), native multimodal input, and long-horizon agentic coding that beats Opus 4.8 and GPT-5.5 on coding/agent benchmarks. Note the price:$3/$15 per M, 4x K2.7-Code— K2.7-Code stays as the automatic fallback (K3 → K2.7-Code → K2.6).
- Gemini 3.6 Flash(Jul 21,gemini-3.6-flash) is the new search/workhorse tier behindgemini_search— 1M context at$1.50/$7.50, down from $9 output, and ~17% fewer output tokens than 3.5 Flash. Flash-Lite moves togemini-3.5-flash-lite($0.30/$2.50).
- Everything else re-verified, unchanged— a full provider audit (Jul 26) confirms GPT-5.6, Grok 4.5, DeepSeek V4 Pro, GLM-5.2, MiniMax M3, StepFun 3.7, ERNIE 4.5 VL, Qwen Coder-Next and Perplexity Sonar are all still the newest available. Gemini 3.1 Pro remains the reasoning/judge default: Google shipped three models on Jul 21 andskipped 3.5 Pro. xAI's staged rollout finished (EU access Jul 17), so thegrok-4.3fallback is now quota/region insurance rather than a rollout workaround.

- Grok 4.5(Jul 8 flagship, "Opus-class") now powers all Grok tools, with automatic fallback togrok-4.3while xAI's region-staged rollout completes (EU mid-July) — tools keep working everywhere, and 4.5 activates by itself.
- grok_search_lite(new tool, 65 total) — the same Grok live search ongrok-4-1-fast($0.20/$0.50, 2M ctx), ~10x cheaper thangrok_search. Use it for high-volume lookups and jury/council fan-outs.
- GPT-5.6 tiers—openai_*tools move togpt-5.6-sol(flagship, same $5/$30 as 5.5 but stronger),terrafor code (5.5-level at half price),lunafor explanations ($1/$6). The $30/$180gpt-5.5-protier is replaced by sol + reasoning effort; a permission fallback (sol → terra → 5.5) covers org-gated accounts.

Gateway Mode (Recommended)— 2 keys, all providers:

{ "mcpServers": { "tachibot": { "command": "tachibot", "env": { "OPENROUTER_API_KEY": "sk-or-xxx", "PERPLEXITY_API_KEY": "pplx-xxx", "USE_OPENROUTER_GATEWAY": "true" } } } }

{ "mcpServers": { "tachibot": { "command": "tachibot", "env": { "PERPLEXITY_API_KEY": "your-key", "GROK_API_KEY": "your-key", "OPENAI_API_KEY": "your-key", "GOOGLE_API_KEY": "your-key", "OPENROUTER_API_KEY": "your-key" } } } }

perplexity_ask·perplexity_reason·grok_search·grok_search_lite·openai_search·gemini_search

grok_reason·openai_reason·qwen_reason·qwq_reason·kimi_thinking·kimi_decompose·deepseek_reason·glm_reason·stepfun_reason·ernie_reason·planner_maker·planner_runner·list_plans·spec_writer

kimi_code·grok_code·grok_debug·qwen_coder·qwen_algo·qwen_competitive·deepseek_algo·minimax_code·minimax_agent·testgen·debug_triage

gemini_analyze_text·gemini_analyze_code·gemini_judge·jury·diff_review·plan_critique·gemini_brainstorm·openai_brainstorm·openai_code_review·openai_explain·grok_brainstorm·grok_architect·security_review·kimi_long_context

think·nextThought·focus·tachi·doctor·usage_stats

workflow·workflow_start·continue_workflow·list_workflows·create_workflow·visualize_workflow·workflow_status·validate_workflow·validate_workflow_file

list_prompt_techniques·preview_prompt_technique·execute_prompt_technique·refine_prompt

- Challenger— Critical analysis with multi-model fact-checking
- Verifier— Multi-model consensus verification
- Scout— Hybrid intelligence gathering

// Create a plan with multi-model council planner_maker({ task: "Build a REST API with auth and tests", mode: "start" }) // → Grok searches → Qwen analyzes → Kimi decomposes → GPT critiques → Gemini synthesizes // Execute with checkpoints planner_runner({ plan: planContent, mode: "step", stepNum: 1 }) // → Automatic verification at 50%, 80% (kimi_decompose), and 100%

kimi_decompose({ task: "Migrate monolith to microservices", depth: 3, outputFormat: "dependencies" }) // → Structured subtasks with IDs, parallel flags, acceptance criteria

kimi_code({ task: "review", code: "function processPayment(amount, card) { ... }", language: "typescript" }) // → SWE-Bench 76.8% quality analysis

This is a web browser that enables your coding agent, such as Claude Code, to visit websites on your behalf and assist you in identifying bugs or creating UI test cases.

Deepseek Thinking & Claude 3.5 Sonnet

Combines DeepSeek's reasoning capabilities with Claude 3.5 Sonnet's response generation through OpenRouter.

An MCP server that orchestrates Google Gemini and Claude Code models via the OpenRouter API.

Delegate bounded work from Claude to any OpenAI-compatible LLM endpoint (LM Studio, Ollama, OpenRouter), preserving your Claude context and quota.

next-devtools-mcp is a MCP server that provides Next.js development tools and utilities for AI coding assistants like Claude and Cursor.

A demonstration server for ActionKit, providing access to Slack actions via Claude Desktop.

MCP server that lets Claude Code agents delegate tasks to agents in other project directories, with parallel dispatch, sessions, and async jobs.

Universal tool adapter — @tool decorator exports Python functions to OpenAI, Claude, Gemini, MCP, JSON Schema. Audit token costs.

Standing review layer for coding agents: Claude, GPT and Gemini debate each answer and return one recommendation plus the strongest dissent.