القائمة
Chat / code generation via flow_router using OpenAI /v1/chat/completions or Anthropic /v1/messages or Gemini /v1beta/models/...:generateContent with streaming + auto-fallback combos.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
Embeddings via flow_router /v1/embeddings. OpenAI-compat shape, routes to active embedding providers (OpenAI, Gemini, Cohere, etc.).
Image generation via flow_router OpenAI-shape /v1/images endpoint. Routes to active media providers (OpenAI, Stability, BFL, Gemini, etc.).
Entry point for flow_router — local/remote AI gateway with OpenAI-compatible REST for chat, image, TTS, embeddings, web search, web fetch. Use when the user mentions flow_router, FLOW_ROUTER_URL, or wants AI without writing provider boilerplate. This skill covers setup + indexes capability skills.
Speech-to-text (audio transcription) via flow_router /v1/audio/transcriptions. OpenAI-compat shape (multipart form), routes to active STT providers.
Text-to-speech via flow_router. OpenAI /v1/audio/speech shape, routes to active TTS providers (OpenAI, ElevenLabs, Gemini TTS, Edge TTS, etc.).
Web fetch (URL → markdown) via flow_router /v1/web/fetch. Pulls a page and returns cleaned markdown text.
| name | flow-router-chat |
| description | Chat / code generation via flow_router using OpenAI /v1/chat/completions or Anthropic /v1/messages or Gemini /v1beta/models/...:generateContent with streaming + auto-fallback combos. |
Requires FLOW_ROUTER_URL (and FLOW_ROUTER_KEY if auth enabled). See flow-router/SKILL.md for setup.
POST $FLOW_ROUTER_URL/v1/chat/completions — OpenAI formatPOST $FLOW_ROUTER_URL/v1/messages — Anthropic formatPOST $FLOW_ROUTER_URL/v1beta/models/{model}:generateContent — Gemini formatPOST $FLOW_ROUTER_URL/v1/responses — OpenAI Responses APIcurl $FLOW_ROUTER_URL/v1/models | jq '.data[].id'
curl "$FLOW_ROUTER_URL/v1/models/info?id=anthropic/claude-haiku-4-5"
Combos (e.g. vip, mycodex) auto-fallback through multiple providers.
curl -X POST $FLOW_ROUTER_URL/v1/chat/completions \
-H "Authorization: Bearer $FLOW_ROUTER_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "claude-haiku-4-5",
"messages": [{"role":"user","content":"Say hi"}],
"stream": false
}'
Set "stream": true to receive SSE chunks (delta format).
curl -X POST $FLOW_ROUTER_URL/v1/messages \
-H "Authorization: Bearer $FLOW_ROUTER_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "claude-haiku-4-5",
"max_tokens": 1024,
"messages": [{"role":"user","content":"Say hi"}]
}'
When RtkTokenSaver is enabled in settings, tool-result messages over 4 KB are auto-compressed by format (git diff, build output, grep, tree, etc.) before being forwarded upstream — typical 40-80% reduction for agent loops.