switchboard-router 0.5.20 → 0.5.21

package/app/.next/standalone/.next-cli-build/standalone/switchboard/skills/switchboard/SKILL.md

@@ -1,79 +0,0 @@
----
-name: switchboard
-description: Entry point for Switchboard — intelligent model-routing gateway with one OpenAI-compatible /v1 endpoint, multi-provider accounts, and combo strategies (fallback, round-robin, fusion, Auto). Use when the user mentions Switchboard, SWITCHBOARD_URL, combos, Auto routing, or wants chat/code through a local gateway without provider boilerplate.
----
-
-# Switchboard
-
-Local AI routing gateway: one OpenAI-compatible endpoint, many providers, model combos with Auto (router → worker), fallback, round-robin, and fusion. Dashboard for keys, combos, usage, and routing insights.
-
-## Setup
-
-```bash
-export SWITCHBOARD_URL="http://localhost:20128"      # or VPS / tunnel URL
-export SWITCHBOARD_KEY="sk_switchboard"              # from Dashboard → Endpoint & keys (only if requireApiKey=true)
-```
-
-All requests: `${SWITCHBOARD_URL}/v1/...` with header `Authorization: Bearer ${SWITCHBOARD_KEY}` (omit if auth disabled).
-
-Verify: `curl $SWITCHBOARD_URL/api/health` → `{"ok":true}`
-
-## Point any OpenAI-compatible client
-
-```bash
-export OPENAI_BASE_URL="$SWITCHBOARD_URL/v1"
-export OPENAI_API_KEY="$SWITCHBOARD_KEY"
-```
-
-Works with Claude Code, Codex, Cline, Cursor, Continue, and similar tools.
-
-## Discover models
-
-```bash
-# Chat / LLM models (and combos with owned_by:"combo")
-curl $SWITCHBOARD_URL/v1/models | jq '.data[].id'
-
-# Per-model metadata
-curl "$SWITCHBOARD_URL/v1/models/info?id=openai/gpt-4o"
-```
-
-Use `data[].id` as the `model` field. Prefer a **combo name** (e.g. `auto`) when the dashboard has Auto/fallback configured — Switchboard picks or fails over workers for you.
-
-Response shape:
-
-```json
-{
-  "object": "list",
-  "data": [
-    { "id": "openai/gpt-5", "object": "model", "owned_by": "openai", "created": 1735000000 },
-    { "id": "auto", "object": "model", "owned_by": "combo", "created": 1735000000 }
-  ]
-}
-```
-
-## Combos & Auto routing
-
-Combos are dashboard-defined groups of models:
-
-| Strategy | Behavior |
-|----------|----------|
-| **fallback** | Try models in order until one succeeds |
-| **round-robin** | Rotate traffic across models |
-| **fusion** | Parallel panel + judge merge |
-| **Auto** | Router LLM picks one worker from the pool every request; outcomes feed learning |
-
-Use the combo’s `name` as `model` in chat requests. Configure router, objective, and pool under **Dashboard → Combos**. Routing insights / relearn: **Combos → Routing insights**.
-
-## Capability skill
-
-| Capability | Agent URL (with server running) |
-|------------|----------------------------------|
-| Chat / code-gen (OpenAI + Anthropic formats) | `$SWITCHBOARD_URL/api/skills/switchboard-chat` |
-
-On disk: `skills/switchboard-chat/SKILL.md`.
-
-## Errors
-
-- **401** → set/refresh `SWITCHBOARD_KEY` (Dashboard → Endpoint & keys)
-- **400** `Invalid model format` → check `model` exists in `/v1/models`
-- **503** `All accounts unavailable` / all combo workers failed → wait `retry-after` or add provider accounts

package/app/.next/standalone/.next-cli-build/standalone/switchboard/skills/switchboard-chat/SKILL.md

@@ -1,116 +0,0 @@
----
-name: switchboard-chat
-description: Chat / code generation via Switchboard using OpenAI /v1/chat/completions, Anthropic /v1/messages, or /v1/responses — with streaming, tools, and combo strategies (fallback / Auto). Use when the user wants to ask an LLM, generate code, run agent tools, or route prompts through Switchboard.
----
-
-# Switchboard — Chat
-
-Requires `SWITCHBOARD_URL` (and `SWITCHBOARD_KEY` if auth enabled). See `$SWITCHBOARD_URL/api/skills/switchboard` for setup.
-
-## Endpoints
-
-| Endpoint | Format |
-|----------|--------|
-| `POST $SWITCHBOARD_URL/v1/chat/completions` | OpenAI Chat Completions |
-| `POST $SWITCHBOARD_URL/v1/messages` | Anthropic Messages |
-| `POST $SWITCHBOARD_URL/v1/responses` | OpenAI Responses API |
-
-## Discover models
-
-```bash
-curl $SWITCHBOARD_URL/v1/models | jq '.data[] | {id, owned_by}'
-curl "$SWITCHBOARD_URL/v1/models/info?id=openai/gpt-4o"
-```
-
-- Single models: `provider/model-id` (e.g. `openai/gpt-4o`, `cc/claude-sonnet-4-6`)
-- **Combos**: dashboard name as `model` (e.g. `auto`) — Switchboard applies fallback / round-robin / fusion / Auto
-
-## Prefer a combo for reliability
-
-If the user has an Auto or fallback combo, use that name as `model` so Switchboard routes (and learns, for Auto):
-
-```bash
-curl -X POST $SWITCHBOARD_URL/v1/chat/completions \
-  -H "Authorization: Bearer $SWITCHBOARD_KEY" \
-  -H "Content-Type: application/json" \
-  -d '{"model":"auto","messages":[{"role":"user","content":"Hi"}],"stream":false}'
-```
-
-## OpenAI format
-
-```bash
-curl -X POST $SWITCHBOARD_URL/v1/chat/completions \
-  -H "Authorization: Bearer $SWITCHBOARD_KEY" \
-  -H "Content-Type: application/json" \
-  -d '{"model":"openai/gpt-4o","messages":[{"role":"user","content":"Hi"}],"stream":false}'
-```
-
-JS (OpenAI SDK):
-
-```js
-import OpenAI from "openai";
-const client = new OpenAI({
-  baseURL: `${process.env.SWITCHBOARD_URL}/v1`,
-  apiKey: process.env.SWITCHBOARD_KEY,
-});
-const res = await client.chat.completions.create({
-  model: "auto", // or "openai/gpt-4o"
-  messages: [{ role: "user", content: "Hi" }],
-  stream: true,
-});
-for await (const chunk of res) {
-  process.stdout.write(chunk.choices[0]?.delta?.content || "");
-}
-```
-
-## Anthropic format
-
-```bash
-curl -X POST $SWITCHBOARD_URL/v1/messages \
-  -H "Authorization: Bearer $SWITCHBOARD_KEY" \
-  -H "anthropic-version: 2023-06-01" \
-  -H "Content-Type: application/json" \
-  -d '{"model":"cc/claude-sonnet-4-6","max_tokens":1024,"messages":[{"role":"user","content":"Hi"}]}'
-```
-
-## Response shape
-
-OpenAI (`/v1/chat/completions`):
-
-```json
-{
-  "id": "chatcmpl-...",
-  "object": "chat.completion",
-  "model": "openai/gpt-4o",
-  "choices": [
-    {
-      "index": 0,
-      "message": { "role": "assistant", "content": "Hello!" },
-      "finish_reason": "stop"
-    }
-  ],
-  "usage": { "prompt_tokens": 8, "completion_tokens": 2, "total_tokens": 10 }
-}
-```
-
-Streaming (`stream: true`) emits SSE: `data: {choices:[{delta:{content:"..."}}]}\n\n` … `data: [DONE]\n\n`.
-
-Anthropic (`/v1/messages`):
-
-```json
-{
-  "id": "msg_...",
-  "type": "message",
-  "role": "assistant",
-  "model": "cc/claude-sonnet-4-6",
-  "content": [{ "type": "text", "text": "Hello!" }],
-  "stop_reason": "end_turn",
-  "usage": { "input_tokens": 8, "output_tokens": 2 }
-}
-```
-
-## Tips
-
-- Tools / function calling: send OpenAI or Anthropic tool schemas as usual — Switchboard translates per upstream.
-- Vision: include image parts in messages when the worker model supports vision.
-- Auto combos re-pick a worker every request; do not assume sticky model identity across turns.

package/app/.next/standalone/cli/app/.next/standalone/skills/switchboard/SKILL.md

@@ -1,79 +0,0 @@
----
-name: switchboard
-description: Entry point for Switchboard — intelligent model-routing gateway with one OpenAI-compatible /v1 endpoint, multi-provider accounts, and combo strategies (fallback, round-robin, fusion, Auto). Use when the user mentions Switchboard, SWITCHBOARD_URL, combos, Auto routing, or wants chat/code through a local gateway without provider boilerplate.
----
-
-# Switchboard
-
-Local AI routing gateway: one OpenAI-compatible endpoint, many providers, model combos with Auto (router → worker), fallback, round-robin, and fusion. Dashboard for keys, combos, usage, and routing insights.
-
-## Setup
-
-```bash
-export SWITCHBOARD_URL="http://localhost:20128"      # or VPS / tunnel URL
-export SWITCHBOARD_KEY="sk_switchboard"              # from Dashboard → Endpoint & keys (only if requireApiKey=true)
-```
-
-All requests: `${SWITCHBOARD_URL}/v1/...` with header `Authorization: Bearer ${SWITCHBOARD_KEY}` (omit if auth disabled).
-
-Verify: `curl $SWITCHBOARD_URL/api/health` → `{"ok":true}`
-
-## Point any OpenAI-compatible client
-
-```bash
-export OPENAI_BASE_URL="$SWITCHBOARD_URL/v1"
-export OPENAI_API_KEY="$SWITCHBOARD_KEY"
-```
-
-Works with Claude Code, Codex, Cline, Cursor, Continue, and similar tools.
-
-## Discover models
-
-```bash
-# Chat / LLM models (and combos with owned_by:"combo")
-curl $SWITCHBOARD_URL/v1/models | jq '.data[].id'
-
-# Per-model metadata
-curl "$SWITCHBOARD_URL/v1/models/info?id=openai/gpt-4o"
-```
-
-Use `data[].id` as the `model` field. Prefer a **combo name** (e.g. `auto`) when the dashboard has Auto/fallback configured — Switchboard picks or fails over workers for you.
-
-Response shape:
-
-```json
-{
-  "object": "list",
-  "data": [
-    { "id": "openai/gpt-5", "object": "model", "owned_by": "openai", "created": 1735000000 },
-    { "id": "auto", "object": "model", "owned_by": "combo", "created": 1735000000 }
-  ]
-}
-```
-
-## Combos & Auto routing
-
-Combos are dashboard-defined groups of models:
-
-| Strategy | Behavior |
-|----------|----------|
-| **fallback** | Try models in order until one succeeds |
-| **round-robin** | Rotate traffic across models |
-| **fusion** | Parallel panel + judge merge |
-| **Auto** | Router LLM picks one worker from the pool every request; outcomes feed learning |
-
-Use the combo’s `name` as `model` in chat requests. Configure router, objective, and pool under **Dashboard → Combos**. Routing insights / relearn: **Combos → Routing insights**.
-
-## Capability skill
-
-| Capability | Agent URL (with server running) |
-|------------|----------------------------------|
-| Chat / code-gen (OpenAI + Anthropic formats) | `$SWITCHBOARD_URL/api/skills/switchboard-chat` |
-
-On disk: `skills/switchboard-chat/SKILL.md`.
-
-## Errors
-
-- **401** → set/refresh `SWITCHBOARD_KEY` (Dashboard → Endpoint & keys)
-- **400** `Invalid model format` → check `model` exists in `/v1/models`
-- **503** `All accounts unavailable` / all combo workers failed → wait `retry-after` or add provider accounts

package/app/.next/standalone/cli/app/.next/standalone/skills/switchboard-chat/SKILL.md

@@ -1,116 +0,0 @@
----
-name: switchboard-chat
-description: Chat / code generation via Switchboard using OpenAI /v1/chat/completions, Anthropic /v1/messages, or /v1/responses — with streaming, tools, and combo strategies (fallback / Auto). Use when the user wants to ask an LLM, generate code, run agent tools, or route prompts through Switchboard.
----
-
-# Switchboard — Chat
-
-Requires `SWITCHBOARD_URL` (and `SWITCHBOARD_KEY` if auth enabled). See `$SWITCHBOARD_URL/api/skills/switchboard` for setup.
-
-## Endpoints
-
-| Endpoint | Format |
-|----------|--------|
-| `POST $SWITCHBOARD_URL/v1/chat/completions` | OpenAI Chat Completions |
-| `POST $SWITCHBOARD_URL/v1/messages` | Anthropic Messages |
-| `POST $SWITCHBOARD_URL/v1/responses` | OpenAI Responses API |
-
-## Discover models
-
-```bash
-curl $SWITCHBOARD_URL/v1/models | jq '.data[] | {id, owned_by}'
-curl "$SWITCHBOARD_URL/v1/models/info?id=openai/gpt-4o"
-```
-
-- Single models: `provider/model-id` (e.g. `openai/gpt-4o`, `cc/claude-sonnet-4-6`)
-- **Combos**: dashboard name as `model` (e.g. `auto`) — Switchboard applies fallback / round-robin / fusion / Auto
-
-## Prefer a combo for reliability
-
-If the user has an Auto or fallback combo, use that name as `model` so Switchboard routes (and learns, for Auto):
-
-```bash
-curl -X POST $SWITCHBOARD_URL/v1/chat/completions \
-  -H "Authorization: Bearer $SWITCHBOARD_KEY" \
-  -H "Content-Type: application/json" \
-  -d '{"model":"auto","messages":[{"role":"user","content":"Hi"}],"stream":false}'
-```
-
-## OpenAI format
-
-```bash
-curl -X POST $SWITCHBOARD_URL/v1/chat/completions \
-  -H "Authorization: Bearer $SWITCHBOARD_KEY" \
-  -H "Content-Type: application/json" \
-  -d '{"model":"openai/gpt-4o","messages":[{"role":"user","content":"Hi"}],"stream":false}'
-```
-
-JS (OpenAI SDK):
-
-```js
-import OpenAI from "openai";
-const client = new OpenAI({
-  baseURL: `${process.env.SWITCHBOARD_URL}/v1`,
-  apiKey: process.env.SWITCHBOARD_KEY,
-});
-const res = await client.chat.completions.create({
-  model: "auto", // or "openai/gpt-4o"
-  messages: [{ role: "user", content: "Hi" }],
-  stream: true,
-});
-for await (const chunk of res) {
-  process.stdout.write(chunk.choices[0]?.delta?.content || "");
-}
-```
-
-## Anthropic format
-
-```bash
-curl -X POST $SWITCHBOARD_URL/v1/messages \
-  -H "Authorization: Bearer $SWITCHBOARD_KEY" \
-  -H "anthropic-version: 2023-06-01" \
-  -H "Content-Type: application/json" \
-  -d '{"model":"cc/claude-sonnet-4-6","max_tokens":1024,"messages":[{"role":"user","content":"Hi"}]}'
-```
-
-## Response shape
-
-OpenAI (`/v1/chat/completions`):
-
-```json
-{
-  "id": "chatcmpl-...",
-  "object": "chat.completion",
-  "model": "openai/gpt-4o",
-  "choices": [
-    {
-      "index": 0,
-      "message": { "role": "assistant", "content": "Hello!" },
-      "finish_reason": "stop"
-    }
-  ],
-  "usage": { "prompt_tokens": 8, "completion_tokens": 2, "total_tokens": 10 }
-}
-```
-
-Streaming (`stream: true`) emits SSE: `data: {choices:[{delta:{content:"..."}}]}\n\n` … `data: [DONE]\n\n`.
-
-Anthropic (`/v1/messages`):
-
-```json
-{
-  "id": "msg_...",
-  "type": "message",
-  "role": "assistant",
-  "model": "cc/claude-sonnet-4-6",
-  "content": [{ "type": "text", "text": "Hello!" }],
-  "stop_reason": "end_turn",
-  "usage": { "input_tokens": 8, "output_tokens": 2 }
-}
-```
-
-## Tips
-
-- Tools / function calling: send OpenAI or Anthropic tool schemas as usual — Switchboard translates per upstream.
-- Vision: include image parts in messages when the worker model supports vision.
-- Auto combos re-pick a worker every request; do not assume sticky model identity across turns.

package/app/.next/standalone/cli/app/cli/app/.next/standalone/skills/switchboard/SKILL.md

@@ -1,79 +0,0 @@
----
-name: switchboard
-description: Entry point for Switchboard — intelligent model-routing gateway with one OpenAI-compatible /v1 endpoint, multi-provider accounts, and combo strategies (fallback, round-robin, fusion, Auto). Use when the user mentions Switchboard, SWITCHBOARD_URL, combos, Auto routing, or wants chat/code through a local gateway without provider boilerplate.
----
-
-# Switchboard
-
-Local AI routing gateway: one OpenAI-compatible endpoint, many providers, model combos with Auto (router → worker), fallback, round-robin, and fusion. Dashboard for keys, combos, usage, and routing insights.
-
-## Setup
-
-```bash
-export SWITCHBOARD_URL="http://localhost:20128"      # or VPS / tunnel URL
-export SWITCHBOARD_KEY="sk_switchboard"              # from Dashboard → Endpoint & keys (only if requireApiKey=true)
-```
-
-All requests: `${SWITCHBOARD_URL}/v1/...` with header `Authorization: Bearer ${SWITCHBOARD_KEY}` (omit if auth disabled).
-
-Verify: `curl $SWITCHBOARD_URL/api/health` → `{"ok":true}`
-
-## Point any OpenAI-compatible client
-
-```bash
-export OPENAI_BASE_URL="$SWITCHBOARD_URL/v1"
-export OPENAI_API_KEY="$SWITCHBOARD_KEY"
-```
-
-Works with Claude Code, Codex, Cline, Cursor, Continue, and similar tools.
-
-## Discover models
-
-```bash
-# Chat / LLM models (and combos with owned_by:"combo")
-curl $SWITCHBOARD_URL/v1/models | jq '.data[].id'
-
-# Per-model metadata
-curl "$SWITCHBOARD_URL/v1/models/info?id=openai/gpt-4o"
-```
-
-Use `data[].id` as the `model` field. Prefer a **combo name** (e.g. `auto`) when the dashboard has Auto/fallback configured — Switchboard picks or fails over workers for you.
-
-Response shape:
-
-```json
-{
-  "object": "list",
-  "data": [
-    { "id": "openai/gpt-5", "object": "model", "owned_by": "openai", "created": 1735000000 },
-    { "id": "auto", "object": "model", "owned_by": "combo", "created": 1735000000 }
-  ]
-}
-```
-
-## Combos & Auto routing
-
-Combos are dashboard-defined groups of models:
-
-| Strategy | Behavior |
-|----------|----------|
-| **fallback** | Try models in order until one succeeds |
-| **round-robin** | Rotate traffic across models |
-| **fusion** | Parallel panel + judge merge |
-| **Auto** | Router LLM picks one worker from the pool every request; outcomes feed learning |
-
-Use the combo’s `name` as `model` in chat requests. Configure router, objective, and pool under **Dashboard → Combos**. Routing insights / relearn: **Combos → Routing insights**.
-
-## Capability skill
-
-| Capability | Agent URL (with server running) |
-|------------|----------------------------------|
-| Chat / code-gen (OpenAI + Anthropic formats) | `$SWITCHBOARD_URL/api/skills/switchboard-chat` |
-
-On disk: `skills/switchboard-chat/SKILL.md`.
-
-## Errors
-
-- **401** → set/refresh `SWITCHBOARD_KEY` (Dashboard → Endpoint & keys)
-- **400** `Invalid model format` → check `model` exists in `/v1/models`
-- **503** `All accounts unavailable` / all combo workers failed → wait `retry-after` or add provider accounts

package/app/.next/standalone/cli/app/cli/app/.next/standalone/skills/switchboard-chat/SKILL.md

@@ -1,116 +0,0 @@
----
-name: switchboard-chat
-description: Chat / code generation via Switchboard using OpenAI /v1/chat/completions, Anthropic /v1/messages, or /v1/responses — with streaming, tools, and combo strategies (fallback / Auto). Use when the user wants to ask an LLM, generate code, run agent tools, or route prompts through Switchboard.
----
-
-# Switchboard — Chat
-
-Requires `SWITCHBOARD_URL` (and `SWITCHBOARD_KEY` if auth enabled). See `$SWITCHBOARD_URL/api/skills/switchboard` for setup.
-
-## Endpoints
-
-| Endpoint | Format |
-|----------|--------|
-| `POST $SWITCHBOARD_URL/v1/chat/completions` | OpenAI Chat Completions |
-| `POST $SWITCHBOARD_URL/v1/messages` | Anthropic Messages |
-| `POST $SWITCHBOARD_URL/v1/responses` | OpenAI Responses API |
-
-## Discover models
-
-```bash
-curl $SWITCHBOARD_URL/v1/models | jq '.data[] | {id, owned_by}'
-curl "$SWITCHBOARD_URL/v1/models/info?id=openai/gpt-4o"
-```
-
-- Single models: `provider/model-id` (e.g. `openai/gpt-4o`, `cc/claude-sonnet-4-6`)
-- **Combos**: dashboard name as `model` (e.g. `auto`) — Switchboard applies fallback / round-robin / fusion / Auto
-
-## Prefer a combo for reliability
-
-If the user has an Auto or fallback combo, use that name as `model` so Switchboard routes (and learns, for Auto):
-
-```bash
-curl -X POST $SWITCHBOARD_URL/v1/chat/completions \
-  -H "Authorization: Bearer $SWITCHBOARD_KEY" \
-  -H "Content-Type: application/json" \
-  -d '{"model":"auto","messages":[{"role":"user","content":"Hi"}],"stream":false}'
-```
-
-## OpenAI format
-
-```bash
-curl -X POST $SWITCHBOARD_URL/v1/chat/completions \
-  -H "Authorization: Bearer $SWITCHBOARD_KEY" \
-  -H "Content-Type: application/json" \
-  -d '{"model":"openai/gpt-4o","messages":[{"role":"user","content":"Hi"}],"stream":false}'
-```
-
-JS (OpenAI SDK):
-
-```js
-import OpenAI from "openai";
-const client = new OpenAI({
-  baseURL: `${process.env.SWITCHBOARD_URL}/v1`,
-  apiKey: process.env.SWITCHBOARD_KEY,
-});
-const res = await client.chat.completions.create({
-  model: "auto", // or "openai/gpt-4o"
-  messages: [{ role: "user", content: "Hi" }],
-  stream: true,
-});
-for await (const chunk of res) {
-  process.stdout.write(chunk.choices[0]?.delta?.content || "");
-}
-```
-
-## Anthropic format
-
-```bash
-curl -X POST $SWITCHBOARD_URL/v1/messages \
-  -H "Authorization: Bearer $SWITCHBOARD_KEY" \
-  -H "anthropic-version: 2023-06-01" \
-  -H "Content-Type: application/json" \
-  -d '{"model":"cc/claude-sonnet-4-6","max_tokens":1024,"messages":[{"role":"user","content":"Hi"}]}'
-```
-
-## Response shape
-
-OpenAI (`/v1/chat/completions`):
-
-```json
-{
-  "id": "chatcmpl-...",
-  "object": "chat.completion",
-  "model": "openai/gpt-4o",
-  "choices": [
-    {
-      "index": 0,
-      "message": { "role": "assistant", "content": "Hello!" },
-      "finish_reason": "stop"
-    }
-  ],
-  "usage": { "prompt_tokens": 8, "completion_tokens": 2, "total_tokens": 10 }
-}
-```
-
-Streaming (`stream: true`) emits SSE: `data: {choices:[{delta:{content:"..."}}]}\n\n` … `data: [DONE]\n\n`.
-
-Anthropic (`/v1/messages`):
-
-```json
-{
-  "id": "msg_...",
-  "type": "message",
-  "role": "assistant",
-  "model": "cc/claude-sonnet-4-6",
-  "content": [{ "type": "text", "text": "Hello!" }],
-  "stop_reason": "end_turn",
-  "usage": { "input_tokens": 8, "output_tokens": 2 }
-}
-```
-
-## Tips
-
-- Tools / function calling: send OpenAI or Anthropic tool schemas as usual — Switchboard translates per upstream.
-- Vision: include image parts in messages when the worker model supports vision.
-- Auto combos re-pick a worker every request; do not assume sticky model identity across turns.

package/app/.next/standalone/cli/app/cli/app/skills/switchboard/SKILL.md

@@ -1,79 +0,0 @@
----
-name: switchboard
-description: Entry point for Switchboard — intelligent model-routing gateway with one OpenAI-compatible /v1 endpoint, multi-provider accounts, and combo strategies (fallback, round-robin, fusion, Auto). Use when the user mentions Switchboard, SWITCHBOARD_URL, combos, Auto routing, or wants chat/code through a local gateway without provider boilerplate.
----
-
-# Switchboard
-
-Local AI routing gateway: one OpenAI-compatible endpoint, many providers, model combos with Auto (router → worker), fallback, round-robin, and fusion. Dashboard for keys, combos, usage, and routing insights.
-
-## Setup
-
-```bash
-export SWITCHBOARD_URL="http://localhost:20128"      # or VPS / tunnel URL
-export SWITCHBOARD_KEY="sk_switchboard"              # from Dashboard → Endpoint & keys (only if requireApiKey=true)
-```
-
-All requests: `${SWITCHBOARD_URL}/v1/...` with header `Authorization: Bearer ${SWITCHBOARD_KEY}` (omit if auth disabled).
-
-Verify: `curl $SWITCHBOARD_URL/api/health` → `{"ok":true}`
-
-## Point any OpenAI-compatible client
-
-```bash
-export OPENAI_BASE_URL="$SWITCHBOARD_URL/v1"
-export OPENAI_API_KEY="$SWITCHBOARD_KEY"
-```
-
-Works with Claude Code, Codex, Cline, Cursor, Continue, and similar tools.
-
-## Discover models
-
-```bash
-# Chat / LLM models (and combos with owned_by:"combo")
-curl $SWITCHBOARD_URL/v1/models | jq '.data[].id'
-
-# Per-model metadata
-curl "$SWITCHBOARD_URL/v1/models/info?id=openai/gpt-4o"
-```
-
-Use `data[].id` as the `model` field. Prefer a **combo name** (e.g. `auto`) when the dashboard has Auto/fallback configured — Switchboard picks or fails over workers for you.
-
-Response shape:
-
-```json
-{
-  "object": "list",
-  "data": [
-    { "id": "openai/gpt-5", "object": "model", "owned_by": "openai", "created": 1735000000 },
-    { "id": "auto", "object": "model", "owned_by": "combo", "created": 1735000000 }
-  ]
-}
-```
-
-## Combos & Auto routing
-
-Combos are dashboard-defined groups of models:
-
-| Strategy | Behavior |
-|----------|----------|
-| **fallback** | Try models in order until one succeeds |
-| **round-robin** | Rotate traffic across models |
-| **fusion** | Parallel panel + judge merge |
-| **Auto** | Router LLM picks one worker from the pool every request; outcomes feed learning |
-
-Use the combo’s `name` as `model` in chat requests. Configure router, objective, and pool under **Dashboard → Combos**. Routing insights / relearn: **Combos → Routing insights**.
-
-## Capability skill
-
-| Capability | Agent URL (with server running) |
-|------------|----------------------------------|
-| Chat / code-gen (OpenAI + Anthropic formats) | `$SWITCHBOARD_URL/api/skills/switchboard-chat` |
-
-On disk: `skills/switchboard-chat/SKILL.md`.
-
-## Errors
-
-- **401** → set/refresh `SWITCHBOARD_KEY` (Dashboard → Endpoint & keys)
-- **400** `Invalid model format` → check `model` exists in `/v1/models`
-- **503** `All accounts unavailable` / all combo workers failed → wait `retry-after` or add provider accounts