/keirouter

name: keirouter
description: Entry point for KeiRouter — local/remote AI gateway with OpenAI-compatible REST for chat, image, TTS, embeddings, web search, web fetch. Use when the user mentions KeiRouter, KEIROUTER_URL, or wants AI without writing provider boilerplate. This skill covers setup + indexes capability skills; fetch the relevant capability SKILL.md from the URLs below when needed.

KeiRouter

Local/remote AI gateway exposing OpenAI-compatible REST. One key, many providers, auto-fallback.

Setup

export KEIROUTER_URL="http://localhost:20180"      # or VPS / tunnel URL
export KEIROUTER_KEY="sk-..."                      # from Dashboard → Keys (only if auth enabled)

All requests: `${KEIROUTER_URL}/v1/...` with header `Authorization: Bearer ${KEIROUTER_KEY}` (omit if auth disabled).

Verify: `curl $KEIROUTER_URL/healthz` → `{"ok":true}`

Discover models

curl $KEIROUTER_URL/v1/models                  # chat/LLM (default)
curl $KEIROUTER_URL/v1/models/image            # image-gen
curl $KEIROUTER_URL/v1/models/tts              # text-to-speech
curl $KEIROUTER_URL/v1/models/embedding        # embeddings
curl $KEIROUTER_URL/v1/models/web              # web search + fetch (entries have `kind` field)
curl $KEIROUTER_URL/v1/models/stt              # speech-to-text

Use `data[].id` as `model` field in requests. Combos appear with `owned_by:"combo"`.

Response shape:

{ "object": "list", "data": [
  { "id": "openai/gpt-5", "object": "model", "owned_by": "openai", "created": 1735000000 },
  { "id": "tavily/search", "object": "model", "kind": "webSearch", "owned_by": "tavily", "created": 1735000000 }
]}

Capability skills

When the user needs a specific capability, fetch that skill's `SKILL.md` from its raw URL:

| Capability | Raw URL | |---|---| | Chat / code-gen | https://raw.githubusercontent.com/mydisha/keirouter/main/skills/keirouter-chat/SKILL.md | | Image generation | https://raw.githubusercontent.com/mydisha/keirouter/main/skills/keirouter-image/SKILL.md | | Text-to-speech | https://raw.githubusercontent.com/mydisha/keirouter/main/skills/keirouter-tts/SKILL.md | | Speech-to-text | https://raw.githubusercontent.com/mydisha/keirouter/main/skills/keirouter-stt/SKILL.md | | Embeddings | https://raw.githubusercontent.com/mydisha/keirouter/main/skills/keirouter-embeddings/SKILL.md | | Web search | https://raw.githubusercontent.com/mydisha/keirouter/main/skills/keirouter-web-search/SKILL.md | | Web fetch (URL → markdown) | https://raw.githubusercontent.com/mydisha/keirouter/main/skills/keirouter-web-fetch/SKILL.md |

Supported providers (highlights)

| Provider | ID | Alias | Notes | |---|---|---|---| | OpenAI | `openai` | `openai` | GPT-5, GPT-4o, DALL-E, Whisper, TTS | | Anthropic | `anthropic` | `anthropic` | Claude Opus, Sonnet, Haiku | | Claude Code | `claude` | `cc` | Claude via Claude Code subscription | | Google Gemini | `gemini` | `gemini` | Gemini 2.5 (chat and vision) | | Groq | `groq` | `groq` | Fast inference, Whisper | | DeepSeek | `deepseek` | `ds` | DeepSeek V3, Coder | | OpenRouter | `openrouter` | `openrouter` | 100+ models via single key | | Mistral | `mistral` | `mistral` | Mistral Large, Medium | | xAI | `xai` | `xai` | Grok models | | NVIDIA NIM | `nvidia` | `nvidia` | Nemotron, Llama | | Ollama Local | `ollama-local` | `ollama-local` | Local models, no auth | | Custom OpenAI | `custom-openai` | `custom-openai` | Any OpenAI-compatible endpoint | | Custom Anthropic | `custom-anthropic` | `custom-anthropic` | Any Anthropic-compatible endpoint |

Use `provider/model` format: `openai/gpt-5`, `anthropic/claude-opus-4-7`, `groq/whisper-large-v3`.

Errors

• 401 → set/refresh `KEIROUTER_KEY` (Dashboard → Keys)
• 400 `Invalid model format` → check `model` exists in `/v1/models/<kind>`
• 503 `All accounts unavailable` → wait `retry-after` or add another provider account
• 429 `Budget exhausted` → budget limit reached, check Dashboard → Budgets