{
"markdown": "# Ask LLM\n\n\n\n[](https://github.com/Lykhoyda/ask-llm/actions/workflows/ci.yml)\n[](https://github.com/Lykhoyda/ask-llm/actions/workflows/release.yml)\n[](https://github.com/Lykhoyda/ask-llm/releases)\n[](https://opensource.org/licenses/MIT)\n\n| Package | Type | Version | Downloads |\n|---------|------|---------|-----------|\n| [`@ask-llm/gemini-mcp`](https://www.npmjs.com/package/@ask-llm/gemini-mcp) | MCP Server | [](https://www.npmjs.com/package/@ask-llm/gemini-mcp) | [](https://www.npmjs.com/package/@ask-llm/gemini-mcp) |\n| [`@ask-llm/codex-mcp`](https://www.npmjs.com/package/@ask-llm/codex-mcp) | MCP Server | [](https://www.npmjs.com/package/@ask-llm/codex-mcp) | [](https://www.npmjs.com/package/@ask-llm/codex-mcp) |\n| [`@ask-llm/claude-mcp`](https://www.npmjs.com/package/@ask-llm/claude-mcp) | MCP Server | [](https://www.npmjs.com/package/@ask-llm/claude-mcp) | [](https://www.npmjs.com/package/@ask-llm/claude-mcp) |\n| [`@ask-llm/grok-mcp`](https://www.npmjs.com/package/@ask-llm/grok-mcp) | MCP Server | [](https://www.npmjs.com/package/@ask-llm/grok-mcp) | [](https://www.npmjs.com/package/@ask-llm/grok-mcp) |\n| [`@ask-llm/ollama-mcp`](https://www.npmjs.com/package/@ask-llm/ollama-mcp) | MCP Server | [](https://www.npmjs.com/package/@ask-llm/ollama-mcp) | [](https://www.npmjs.com/package/@ask-llm/ollama-mcp) |\n| [`@ask-llm/antigravity-mcp`](https://www.npmjs.com/package/@ask-llm/antigravity-mcp) | MCP Server | [](https://www.npmjs.com/package/@ask-llm/antigravity-mcp) | [](https://www.npmjs.com/package/@ask-llm/antigravity-mcp) |\n| [`@ask-llm/mcp`](https://www.npmjs.com/package/@ask-llm/mcp) | MCP Server | [](https://www.npmjs.com/package/@ask-llm/mcp) | [](https://www.npmjs.com/package/@ask-llm/mcp) |\n| [`@ask-llm/plugin`](https://www.npmjs.com/package/@ask-llm/plugin) | Claude Code + Cursor Agent + Pi Host Package | [](https://www.npmjs.com/package/@ask-llm/plugin) | [](https://www.npmjs.com/package/@ask-llm/plugin) |\n\n**MCP servers + Claude Code/Cursor Agent/Pi host package for AI-to-AI collaboration**\n\n
\n\n**Get a second opinion before you ship.** Ask LLM lets your AI assistant — Claude Code, Codex CLI, Cursor, Claude Desktop, or any of [40+ MCP clients](https://modelcontextprotocol.io/clients) — consult a _second_ model to review your code, debate a plan, or catch a bug it might have missed. Pick the reviewer that fits: OpenAI **Codex** (GPT-5.6 Sol → Terra), Anthropic **Claude** (Opus → Sonnet), xAI **Grok** 4.6 via API or Grok CLI (no fallback), Google **Antigravity** (`agy`), a local **Ollama** model, or **Gemini** (1M+ token context). Standard [MCP](https://modelcontextprotocol.io/), no prompt hacks.\n\n> **⚠️ Gemini CLI goes enterprise-only on 2026-06-18:** From that date Google restricts Gemini CLI to **Gemini Code Assist Standard/Enterprise** seats, and free, Google AI Pro, and Ultra accounts lose access. `@ask-llm/gemini-mcp` still installs, but a non-enterprise account then surfaces actionable guidance instead of output. Free/Pro users: switch to **`ask-antigravity`** (the Google-sanctioned successor, subscription-backed via Google AI Pro/Ultra), **`ask-codex`**, **`ask-claude`**, or **`ask-ollama`**. [Announcement](https://developers.googleblog.com/an-important-update-transitioning-gemini-cli-to-antigravity-cli/)\n\n## Why a second opinion?\n\nYour primary AI is confident — but confidence isn't correctness. A second model, with no stake in the first one's answer, catches what it missed.\n\n- **Second opinion on code** — before you commit to an approach, have another model review it independently.\n- **Debate a plan** — send an architecture proposal for critique, alternatives, and trade-off analysis.\n- **Review a diff** — have a different model analyze your changes to surface issues your primary AI glossed over.\n- **Read more than fits** — Gemini and Antigravity's large context windows ingest whole codebases at once.\n- **Keep it local** — run reviews through Ollama when nothing can leave your machine.\n\n## In action\n\n```text\nYou: ask codex to review src/auth.ts for security issues\nCodex: ⚠ verifyToken() compares tokens with === — not timing-safe (line 42)\n ⚠ the session cookie is missing a SameSite attribute\nClaude: Good catches — applying both fixes to src/auth.ts.\n```\n\nOne prompt. A second model reviews independently; your assistant applies the fix — no copy-paste between tools.\n\n## Quick Start\n\n### Claude Code\n\n```bash\n# All-in-one — auto-detects installed providers\nclaude mcp add --scope user ask-llm -- npx -y @ask-llm/mcp\n```\n\n\nOr install providers individually
\n\n```bash\nclaude mcp add --scope user gemini -- npx -y @ask-llm/gemini-mcp\nclaude mcp add --scope user codex -- npx -y @ask-llm/codex-mcp\nclaude mcp add --scope user grok -e XAI_API_KEY=\"$XAI_API_KEY\" -- npx -y @ask-llm/grok-mcp\nclaude mcp add --scope user ollama -- npx -y @ask-llm/ollama-mcp\nclaude mcp add --scope user antigravity -- npx -y @ask-llm/antigravity-mcp\n```\n\n \n\n### Pi\n\nPi is a host harness, not another consulted provider. Install the canonical multi-host package; it exposes the shared skills plus native provider tools because Pi intentionally has no built-in MCP client:\n\n```bash\npi install npm:@ask-llm/plugin\npi list\n```\n\nInvoke `/skill:codex-review`, `/skill:multi-review`, `/skill:compare`, `/skill:brainstorm`, or describe the workflow naturally. The package registers `ask-codex`, `ask-gemini`, `ask-grok`, `ask-ollama`, `ask-antigravity`, model-neutral `ask-cursor-agent`, and deterministic concurrent `ask-multi` tools. Provider CLI authentication is unchanged and separate from Pi's host-model login.\n\nPi codex-pair requires the repository marker, Pi project trust, **and** interactive user-owned consent via `/codex-pair`; a committed marker alone never authorizes source transfer or cost. Pi surfaces findings non-blockingly and does not claim Claude's blocking Stop-gate or one-shot print parity. Independent `fable-review` remains Claude Code-only and is excluded from Pi.\n\nUpdate/remove with `pi update npm:@ask-llm/plugin` and `pi remove npm:@ask-llm/plugin`. See the [Pi host guide](https://lykhoyda.github.io/ask-llm/plugin/pi) for temporary/project-local installs, trust, data transfer, troubleshooting, and the full compatibility matrix.\n\n### Claude Desktop\n\nAdd to `claude_desktop_config.json`:\n\n```json\n{\n \"mcpServers\": {\n \"ask-llm\": {\n \"command\": \"npx\",\n \"args\": [\"-y\", \"@ask-llm/mcp\"]\n }\n }\n}\n```\n\n\nOr install providers individually
\n\n```json\n{\n \"mcpServers\": {\n \"gemini\": {\n \"command\": \"npx\",\n \"args\": [\"-y\", \"@ask-llm/gemini-mcp\"]\n },\n \"codex\": {\n \"command\": \"npx\",\n \"args\": [\"-y\", \"@ask-llm/codex-mcp\"]\n },\n \"ollama\": {\n \"command\": \"npx\",\n \"args\": [\"-y\", \"@ask-llm/ollama-mcp\"]\n }\n }\n}\n```\n\n \n\n\nCursor, Codex CLI, OpenCode, and other clients
\n\n**Cursor** (`.cursor/mcp.json`):\n```json\n{\n \"mcpServers\": {\n \"ask-llm\": { \"command\": \"npx\", \"args\": [\"-y\", \"@ask-llm/mcp\"] }\n }\n}\n```\n\n**Codex CLI** (`~/.codex/config.toml`):\n```toml\n[mcp_servers.ask-llm]\ncommand = \"npx\"\nargs = [\"-y\", \"@ask-llm/mcp\"]\n```\n\nFor the focused Codex → Claude second-opinion path:\n\n```bash\ncodex mcp add claude -- npx -y @ask-llm/claude-mcp\n```\n\n**Any MCP Client** (STDIO transport):\n```json\n{ \"command\": \"npx\", \"args\": [\"-y\", \"@ask-llm/mcp\"] }\n```\n\nReplace `@ask-llm/mcp` with `@ask-llm/codex-mcp`, `@ask-llm/claude-mcp`, `@ask-llm/grok-mcp`, `@ask-llm/antigravity-mcp`, `@ask-llm/ollama-mcp`, or `@ask-llm/gemini-mcp` for a single provider.\n\n \n\n### Migrating from the old package names\n\nAll public MCP packages now live in the `@ask-llm` npm organization. The old\npackage names are deprecated, but their executable names are unchanged. Update\nthe package argument in your MCP config; commands such as `ask-codex-mcp` and\n`ask-llm-mcp doctor` keep working after a global install.\n\n| Old package | Use instead |\n|-------------|-------------|\n| `ask-gemini-mcp` | `@ask-llm/gemini-mcp` |\n| `ask-codex-mcp` | `@ask-llm/codex-mcp` |\n| `@anton-lykhoyda/ask-claude-mcp` | `@ask-llm/claude-mcp` |\n| `ask-ollama-mcp` | `@ask-llm/ollama-mcp` |\n| `ask-antigravity-mcp` | `@ask-llm/antigravity-mcp` |\n| `ask-llm-mcp` | `@ask-llm/mcp` |\n\nSee the [installation guide](https://lykhoyda.github.io/ask-llm/installation) for\nthe complete package-to-executable mapping.\n\n## Choose your reviewer\n\n| Provider | Best for | Model (default → fallback) | Notes |\n|----------|----------|----------------------------|-------|\n| **Codex** | Code reasoning, targeted reviews, architecture critique | `gpt-5.6-sol` → `gpt-5.6-terra` | Requires an OpenAI/Codex account |\n| **Claude** | Independent review from Codex or another non-Claude host | `opus` → `sonnet` | Claude Code CLI; native sessions; read-only tools |\n| **Grok** | Grok 4.6 critique through xAI API or official Grok CLI | API `grok-4.6`, reasoning `high` (no fallback) | Harness selected separately; exact harness catalog ID sent unchanged |\n| **Antigravity** | A subscription-backed second opinion; larger-context reads | `gemini-3.1-pro` → `gemini-3.5-flash` (both at `--effort high`) | Google AI Pro/Ultra plan; one-shot, experimental |\n| **Ollama** | Private/local review, zero cost, offline | `qwen3.8:27b` (no auto-fallback) | Runs entirely on your machine |\n| **Gemini** | Whole-codebase reads (1M+ tokens) | `gemini-3.1-pro-preview` → `gemini-3.7-flash` | ⚠️ Enterprise-gated from 2026-06-18 |\n| **Unified (`ask-llm`)** | One install for all of the above; fan out in parallel | routes per call | **Recommended** |\n\n## Host Package: Claude Code, Cursor Agent, and Pi\n\n`@ask-llm/plugin` is one package, version, release lifecycle, and canonical skill corpus. Claude Code loads its existing marketplace agents/hooks; Cursor Agent loads the adapted `/codex-pair` and `/grok-pair` skills through its Agent Skills surface plus `mcp.json` (`agent --plugin-dir ./packages/claude-plugin`; see the [Cursor Agent host guide](https://lykhoyda.github.io/ask-llm/plugin/cursor)); Pi loads explicit native tools, portable skill adapters, and a thin lifecycle extension.\n\n| Capability | Claude Code | Cursor Agent | Codex CLI host | Pi |\n|---|---:|---:|---:|---:|\n| Provider transport | MCP | MCP (`mcp.json`, unified `ask-llm` only) | MCP | native Ask LLM tools (no built-in MCP) |\n| Review/compare/brainstorm skills | yes | Agent Skills | tools only | `/skill:` + natural language |\n| Isolated reviewer contexts / Fable | yes | no; `fable-review` excluded | no | no; `fable-review` excluded |\n| codex-pair | hooks | on-demand persisted session | no | lifecycle extension |\n| `/grok-pair` | yes (explicit Cursor/xAI/CLI route) | direct xAI/CLI routes via pinned unified `ask-llm` (or user-installed `ask-grok`) | no | excluded |\n| Blocking HIGH Stop gate | opt-in | no | no | no; surfaced non-blockingly |\n| Async pairing in one-shot print | n/a | on-demand skill | no | unsupported |\n\n### Claude Code Plugin\n\nThe **Ask LLM plugin** adds multi-provider code review, brainstorming, and automated hooks directly into Claude Code:\n\n```\n/plugin marketplace add Lykhoyda/ask-llm\n/plugin install ask-llm@ask-llm-plugins\n```\n\n### What You Get\n\n| Feature | Description |\n|:---|:---|\n| `/multi-review` | Parallel Antigravity + Codex review with 4-phase validation pipeline and consensus highlighting (gemini via `/gemini-review`) |\n| `/gemini-review` | Gemini-only review with confidence filtering |\n| `/codex-review` | Codex-only review with confidence filtering |\n| `/fable-review` | Isolated, read-only review that requests the native Fable model and discloses runtime verification limits |\n| `/sol-review` | Model-pinned GPT-5.6 Sol review through Codex |\n| `/grok-review` | Metered Grok review through xAI with exact model attribution and no fallback |\n| `/ollama-review` | Local review — no data leaves your machine |\n| `/antigravity-review` | Subscription-backed review via Google Antigravity (`agy`) — experimental |\n| `/brainstorm` | Multi-LLM brainstorm: Claude Opus researches the topic against real files in parallel with external providers (Gemini/Codex/Grok/Ollama/Antigravity), then synthesizes all findings with verified findings weighted higher; also supports an exact no-Gemini Grok + GPT-5.6 Sol panel routed through Cursor Agent |\n| `/compare` | Side-by-side raw responses from multiple providers, no synthesis — for when you want to see how each provider phrases the same answer |\n| **`codex-pair` hook** | Opt-in continuous review — runs Codex against every Edit/Write/MultiEdit when a `.codex-pair/context.md` marker is present in the project |\n\nThe review agents use a 4-phase pipeline inspired by [Anthropic's code-review plugin](https://github.com/anthropics/claude-code/tree/main/plugins/code-review): context gathering, prompt construction with explicit false-positive exclusions, synthesis, and source-level validation of each finding.\n\nSee the [plugin docs](https://lykhoyda.github.io/ask-llm/plugin/overview) for details.\n\n## Prerequisites\n\n- **[Node.js](https://nodejs.org/)** v20.0.0 or higher (LTS)\n- **At least one provider:**\n - [Codex CLI](https://github.com/openai/codex) — installed and authenticated\n - [Claude Code CLI](https://docs.anthropic.com/en/docs/claude-code/getting-started) — installed and authenticated (for Codex/other clients consulting Claude)\n - [xAI API](https://docs.x.ai/) — set `XAI_API_KEY` for the default metered Grok harness; or install/authenticate official Grok Build (pin `harness: \"grok-cli\"` per request, or set `ASK_GROK_HARNESS=grok-cli` as the default; no failover)\n - [Cursor CLI](https://cursor.com/docs/cli) — optional model-neutral harness; authenticate and choose an exact ID from `agent --list-models`\n - [Antigravity CLI](https://antigravity.google) (`agy`) >=1.1.5 — installed and logged in once (Google AI Pro/Ultra); verify with `agy --version`\n - [Ollama](https://ollama.com) — running locally with a model pulled (`ollama pull qwen3.8:27b`)\n - [Gemini CLI](https://github.com/google-gemini/gemini-cli) — `npm install -g @google/gemini-cli && gemini login` (enterprise-gated from 2026-06-18)\n\n## MCP Tools\n\n| Tool | Package | Purpose |\n|------|---------|---------|\n| `ask-gemini` | @ask-llm/gemini-mcp | Send prompts to Gemini CLI with `@` file syntax. 1M+ token context. Live progressive output via `stream-json` |\n| `ask-gemini-edit` | @ask-llm/gemini-mcp | Get structured OLD/NEW code edit blocks from Gemini |\n| `fetch-chunk` | @ask-llm/gemini-mcp | Retrieve chunks from cached large responses |\n| `ask-codex` | @ask-llm/codex-mcp | Send prompts to Codex CLI. GPT-5.6 Sol with Terra fallback; omit `sessionId` for ephemeral use, or pass `sessionId: \"\"` first to persist and resume |\n| `ask-claude` | @ask-llm/claude-mcp | Send prompts to Claude Code CLI. Opus with Sonnet fallback; native sessions; Read/Glob/Grep-only workspace access |\n| `ask-grok` | @ask-llm/grok-mcp | Send a one-shot Grok prompt through explicit `xai-api` (default) or `grok-cli`; exact harness model ID; no harness/model fallback |\n| `ask-cursor-agent` | @ask-llm/mcp | Model-neutral Cursor Agent harness: separate provider (`claude`, `codex`, `gemini`, `grok`) + exact Cursor catalog model verified against that family, read-only ask mode, no force/trust/spend changes/fallback |\n| `ask-ollama` | @ask-llm/ollama-mcp | Send prompts to local Ollama. Fully private, zero cost. Server-side conversation replay via `sessionId` |\n| `ask-antigravity` | @ask-llm/antigravity-mcp | Send a prompt to Google Antigravity (`agy`) for a subscription-backed second opinion. Experimental; one-shot |\n| `ask-llm` | @ask-llm/mcp | Unified orchestrator — pick provider per call. Fan out to all installed providers |\n| `multi-llm` | @ask-llm/mcp | Dispatch the same prompt to multiple providers in parallel; returns per-provider responses + usage in one call |\n| `get-usage-stats` | all | Per-session token totals, fallback counts, breakdowns by provider/model — all in-memory, no persistence |\n| `diagnose` | @ask-llm/mcp | Self-diagnosis: Node version, PATH resolution, provider CLI presence + versions. Read-only |\n| `ping` | all | Connection test — verify MCP setup |\n\nSession-capable `ask-*` tools accept an optional `sessionId` parameter and return a structured `AskResponse` (provider, response, model, sessionId, usage) via MCP `outputSchema` alongside the human-readable text. Codex requires `sessionId: \"\"` on the first call for a resumable thread; omitting it makes that call ephemeral. The orchestrator (`@ask-llm/mcp`) also exposes `usage://current-session` as an MCP Resource for live JSON snapshots.\n\n### Usage Examples\n\n```\nask codex to review the changes in src/auth.ts for security issues\nask claude for an independent opinion on this architecture (from Codex or another MCP client)\nask antigravity to debate this architecture plan in docs/design.md\nask ollama to explain src/config.ts (runs locally, no data sent anywhere)\nask gemini to summarize @. the current directory (1M+ context, @ is Gemini-only)\nuse multi-llm to compare what codex and gemini think about this approach\n```\n\n## CLI Subcommands\n\nThe orchestrator binary (`@ask-llm/mcp`, installed as `ask-llm-mcp`) starts the MCP server only when run with no arguments. `ask-llm-mcp --help` is the canonical command reference (`--version` prints the package version; unsupported commands or arguments print usage and exit nonzero). The two most common commands:\n\n```bash\n# Interactive multi-provider REPL — switch providers, persist sessions, see usage live\nnpx @ask-llm/mcp repl\n\n# Diagnose your setup — Node version, PATH, provider CLI versions, env vars\nnpx @ask-llm/mcp doctor # human-readable\nnpx @ask-llm/mcp doctor --json # established full JSON, exit 1 on error\nnpx @ask-llm/mcp doctor --format toon # bounded, versioned agent-facing TOON pilot\nnpx @ask-llm/mcp doctor --format toon --full # full TOON escape hatch\n```\n\nThe REPL ships sessions per provider (`/provider gemini`, `/provider codex`, `/new`, `/sessions`, `/usage`) and inherits all the executor behavior (quota fallback, stream-json output for Gemini, native session resume).\n\n## Models\n\n| Provider | Default | Fallback |\n|----------|---------|----------|\n| Gemini | `gemini-3.1-pro-preview` | `gemini-3.7-flash` (on quota) |\n| Codex | `gpt-5.6-sol` | `gpt-5.6-terra` (on quota) |\n| Claude | `opus` | `sonnet` (on overload/unavailability) |\n| Grok | API: `grok-4.6`; CLI: exact `grok models` ID (effort `high`) | — (explicit harness/model selection; no fallback) |\n| Antigravity | `gemini-3.1-pro` (`--effort high`) | `gemini-3.5-flash` (on rate limit); model-less recovery if a shipped slug is rejected |\n| Ollama | `qwen3.8:27b` | — (local; errors if the model isn't pulled) |\n\nGemini, Codex, Claude, and Antigravity automatically fall back under their documented provider-specific conditions. Grok and Ollama never substitute a model: Grok preserves the explicit API/CLI harness catalog selection, while Ollama preserves the locally pulled model — if the requested model isn't pulled, it returns a clear `ollama pull` error.\n\n## Documentation\n\n- **Docs site:** [lykhoyda.github.io/ask-llm](https://lykhoyda.github.io/ask-llm/)\n- **AI-readable:** [llms.txt](https://lykhoyda.github.io/ask-llm/llms.txt) | [llms-full.txt](https://lykhoyda.github.io/ask-llm/llms-full.txt)\n\n## Contributing\n\nContributions are welcome! See [open issues](https://github.com/Lykhoyda/ask-llm/issues) for things to work on.\n\n## License\n\nMIT License. See [LICENSE](LICENSE) for details.\n\n**Disclaimer:** This is an unofficial, third-party tool and is not affiliated with, endorsed, or sponsored by Anthropic, Google, OpenAI, or xAI.\n",
"bytes": 21251,
"sha": "03f34bfbd398678574357832b6637c2f3304700bf9580260c150e1ccec9996e6",
"repo_slug": "lykhoyda/ask-llm",
"fonte": "repo",
"truncated": false,
"api": "https://agentalog.com/api/listings/mcp_io_github_lykhoyda_ask_grok_dff9ea81/readme"
}