feat: ModelRouter — unified OpenAI-compatible multi-source LLM gateway

- Lua adapters per upstream (transform_request/response/stream_chunk, build_headers signing hooks)
- AUTO priority routing with per-model kind (chat/image), explicit source/model routing
- Per-source concurrency caps with queueing, exponential backoff, AUTO failover
- OpenAI-compatible API: chat completions, SSE streaming, image generations, models
- Gateway key auth, web UI for adapter/source management, runtime persistence
- e2e test running the real binary against mocked upstreams

internal/lua/adapters/anthropic.lua

@@ -0,0 +1,86 @@
+local adapter = {}
+
+adapter.name = "anthropic"
+adapter.version = "2.0.0"
+adapter.endpoint = "/v1/messages"
+adapter.headers = {
+    ["anthropic-version"] = "2023-06-01"
+}
+
+function adapter.transform_request(raw_body)
+    local ok, req = pcall(json.decode, raw_body)
+    if not ok then return raw_body end
+
+    local msgs = {}
+    local system = ""
+    for _, m in ipairs(req.messages or {}) do
+        if m.role == "system" then
+            system = system .. m.content .. "\n"
+        else
+            table.insert(msgs, { role = m.role, content = m.content })
+        end
+    end
+
+    local anthropic_req = {
+        model = req.model or "claude-sonnet-4-20250514",
+        max_tokens = req.max_tokens or 4096,
+        messages = msgs,
+        stream = req.stream or false,
+    }
+
+    if not req.disable_thinking then
+        anthropic_req.thinking = { type = "enabled", budget_tokens = 4096 }
+    end
+
+    if system ~= "" then
+        anthropic_req.system = system
+    end
+
+    return json.encode(anthropic_req)
+end
+
+function adapter.transform_response(raw_body)
+    local ok, resp = pcall(json.decode, raw_body)
+    if not ok then return raw_body end
+
+    local unified = {
+        content = "",
+        finish_reason = "",
+        token_usage = { prompt = 0, completion = 0, total = 0 }
+    }
+
+    if resp.usage then
+        unified.token_usage.prompt = resp.usage.input_tokens or 0
+        unified.token_usage.completion = resp.usage.output_tokens or 0
+        unified.token_usage.total = (resp.usage.input_tokens or 0) + (resp.usage.output_tokens or 0)
+    end
+
+    if resp.content and #resp.content > 0 then
+        for _, block in ipairs(resp.content) do
+            if block.type == "text" then
+                unified.content = unified.content .. (block.text or "")
+            end
+        end
+    end
+    unified.finish_reason = resp.stop_reason or ""
+
+    return json.encode(unified)
+end
+
+function adapter.transform_stream_chunk(raw_chunk)
+    local ok, chunk = pcall(json.decode, raw_chunk)
+    if not ok then return "" end
+    if chunk.type == "message_start" then return "" end
+    if chunk.type == "message_delta" then
+        return json.encode({ content = "", done = (chunk.delta and chunk.delta.stop_reason ~= nil) })
+    end
+    if chunk.type == "content_block_delta" and chunk.delta then
+        return json.encode({ content = chunk.delta.text or "", done = false })
+    end
+    if chunk.type == "message_stop" then
+        return json.encode({ content = "", done = true })
+    end
+    return ""
+end
+
+return adapter

internal/lua/adapters/deepseek.lua

@@ -0,0 +1,78 @@
+local adapter = {}
+
+adapter.name = "deepseek"
+adapter.version = "2.1.0"
+adapter.endpoint = "/chat/completions"
+adapter.headers = {}
+
+function adapter.transform_request(raw_body)
+    local ok, req = pcall(json.decode, raw_body)
+    if not ok then return raw_body end
+
+    req.model = req.model or "deepseek-chat"
+    req.stream = req.stream or false
+    if req.disable_thinking then
+        req.extra_body = req.extra_body or {}
+        req.extra_body.thinking = { type = "disabled" }
+    end
+    req.disable_thinking = nil
+    return json.encode(req)
+end
+
+function adapter.transform_response(raw_body)
+    local ok, resp = pcall(json.decode, raw_body)
+    if not ok or resp == nil then return raw_body end
+
+    local unified = {
+        content = "",
+        finish_reason = "",
+        token_usage = { prompt = 0, completion = 0, total = 0 }
+    }
+
+    if type(resp.usage) == "table" then
+        unified.token_usage.prompt = resp.usage.prompt_tokens or 0
+        unified.token_usage.completion = resp.usage.completion_tokens or 0
+        unified.token_usage.total = resp.usage.total_tokens or 0
+    end
+
+    if type(resp.choices) == "table" and #resp.choices > 0 then
+        local ch = resp.choices[1]
+        if type(ch.message) == "table" then
+            unified.content = ch.message.content or ""
+            if ch.message.reasoning_content then
+                unified.reasoning_content = ch.message.reasoning_content
+            end
+            if type(ch.message.tool_calls) == "table" then
+                local tcs = {}
+                for _, tc in ipairs(ch.message.tool_calls) do
+                    local args_ok, args = pcall(json.decode, tc["function"].arguments)
+                    if not args_ok then args = {} end
+                    table.insert(tcs, {
+                        id = tc.id,
+                        type = tc.type or "function",
+                        name = tc["function"].name,
+                        arguments = args
+                    })
+                end
+                unified.tool_calls = tcs
+            end
+        end
+        unified.finish_reason = ch.finish_reason or ""
+    end
+
+    return json.encode(unified)
+end
+
+function adapter.transform_stream_chunk(raw_chunk)
+    local ok, chunk = pcall(json.decode, raw_chunk)
+    if not ok then return "" end
+    if not chunk.choices or #chunk.choices == 0 then return "" end
+    local delta = chunk.choices[1].delta or {}
+    local fr = chunk.choices[1].finish_reason
+    return json.encode({
+        content = delta.content or "",
+        done = (fr ~= nil)
+    })
+end
+
+return adapter

internal/lua/adapters/gemini.lua

@@ -0,0 +1,89 @@
+local adapter = {}
+
+adapter.name = "gemini"
+adapter.version = "2.0.0"
+adapter.endpoint = "/v1/models"
+adapter.headers = {}
+
+-- Gemini API: POST /v1/models/{model}:generateContent
+-- Auth: API key in query param ?key=XXX or Authorization: Bearer XXX
+function adapter.transform_request(raw_body)
+    local ok, req = pcall(json.decode, raw_body)
+    if not ok then return raw_body end
+
+    local contents = {}
+    for _, m in ipairs(req.messages or {}) do
+        table.insert(contents, {
+            role = (m.role == "assistant") and "model" or m.role,
+            parts = { { text = m.content } }
+        })
+    end
+
+    local gemini_req = {
+        contents = contents,
+        generationConfig = {
+            temperature = req.temperature or 0.7,
+            maxOutputTokens = req.max_tokens or 4096,
+        }
+    }
+
+    if req.stream then
+        gemini_req.stream = true
+    end
+
+    return json.encode(gemini_req)
+end
+
+-- Gemini 的 endpoint 动态拼接：/v1/models/{model}:generateContent
+function adapter.transform_response(raw_body)
+    local ok, resp = pcall(json.decode, raw_body)
+    if not ok then return raw_body end
+
+    local unified = {
+        content = "",
+        finish_reason = "",
+        token_usage = { prompt = 0, completion = 0, total = 0 }
+    }
+
+    if resp.usageMetadata then
+        unified.token_usage.prompt = resp.usageMetadata.promptTokenCount or 0
+        unified.token_usage.completion = resp.usageMetadata.candidatesTokenCount or 0
+        unified.token_usage.total = resp.usageMetadata.totalTokenCount or 0
+    end
+
+    if resp.candidates and #resp.candidates > 0 then
+        local cand = resp.candidates[1]
+        if cand.content and cand.content.parts then
+            for _, part in ipairs(cand.content.parts) do
+                if part.text then
+                    unified.content = unified.content .. part.text
+                end
+            end
+        end
+        if cand.finishReason then
+            unified.finish_reason = cand.finishReason
+        end
+    end
+
+    return json.encode(unified)
+end
+
+function adapter.transform_stream_chunk(raw_chunk)
+    local ok, chunk = pcall(json.decode, raw_chunk)
+    if not ok then return "" end
+
+    if not chunk.candidates or #chunk.candidates == 0 then return "" end
+    local cand = chunk.candidates[1]
+    local content = ""
+    if cand.content and cand.content.parts then
+        for _, part in ipairs(cand.content.parts) do
+            content = content .. (part.text or "")
+        end
+    end
+    return json.encode({
+        content = content,
+        done = (cand.finishReason ~= nil)
+    })
+end
+
+return adapter

internal/lua/adapters/github.lua

@@ -0,0 +1,75 @@
+local adapter = {}
+
+adapter.name = "github"
+adapter.version = "2.0.0"
+adapter.endpoint = "/chat/completions"
+adapter.headers = {}
+
+-- GitHub Models: Azure-like endpoint, auth via Bearer token (PAT)
+-- BaseURL example: https://models.inference.ai.azure.com
+function adapter.transform_request(raw_body)
+    local ok, req = pcall(json.decode, raw_body)
+    if not ok then return raw_body end
+    req.model = req.model or "gpt-4o"
+    req.temperature = req.temperature or 0.7
+    req.max_tokens = req.max_tokens or 4096
+    req.stream = req.stream or false
+    req.disable_thinking = nil
+    req.extra_body = nil
+    return json.encode(req)
+end
+
+function adapter.transform_response(raw_body)
+    local ok, resp = pcall(json.decode, raw_body)
+    if not ok or resp == nil then return raw_body end
+
+    local unified = {
+        content = "",
+        finish_reason = "",
+        token_usage = { prompt = 0, completion = 0, total = 0 }
+    }
+
+    if type(resp.usage) == "table" then
+        unified.token_usage.prompt = resp.usage.prompt_tokens or 0
+        unified.token_usage.completion = resp.usage.completion_tokens or 0
+        unified.token_usage.total = resp.usage.total_tokens or 0
+    end
+
+    if type(resp.choices) == "table" and #resp.choices > 0 then
+        local ch = resp.choices[1]
+        if type(ch.message) == "table" then
+            unified.content = ch.message.content or ""
+            if type(ch.message.tool_calls) == "table" then
+                local tcs = {}
+                for _, tc in ipairs(ch.message.tool_calls) do
+                    local args_ok, args = pcall(json.decode, tc["function"].arguments)
+                    if not args_ok then args = {} end
+                    table.insert(tcs, {
+                        id = tc.id,
+                        type = tc.type or "function",
+                        name = tc["function"].name,
+                        arguments = args
+                    })
+                end
+                unified.tool_calls = tcs
+            end
+        end
+        unified.finish_reason = ch.finish_reason or ""
+    end
+
+    return json.encode(unified)
+end
+
+function adapter.transform_stream_chunk(raw_chunk)
+    local ok, chunk = pcall(json.decode, raw_chunk)
+    if not ok then return "" end
+    if not chunk.choices or #chunk.choices == 0 then return "" end
+    local delta = chunk.choices[1].delta or {}
+    local fr = chunk.choices[1].finish_reason
+    return json.encode({
+        content = delta.content or "",
+        done = (fr ~= nil)
+    })
+end
+
+return adapter

internal/lua/adapters/groq.lua

@@ -0,0 +1,74 @@
+local adapter = {}
+
+adapter.name = "groq"
+adapter.version = "2.0.0"
+adapter.endpoint = "/openai/v1/chat/completions"
+adapter.headers = {}
+
+-- Groq API is OpenAI-compatible
+function adapter.transform_request(raw_body)
+    local ok, req = pcall(json.decode, raw_body)
+    if not ok then return raw_body end
+    req.model = req.model or "llama3-70b-8192"
+    req.temperature = req.temperature or 0.7
+    req.max_tokens = req.max_tokens or 4096
+    req.stream = req.stream or false
+    req.disable_thinking = nil
+    req.extra_body = nil
+    return json.encode(req)
+end
+
+function adapter.transform_response(raw_body)
+    local ok, resp = pcall(json.decode, raw_body)
+    if not ok or resp == nil then return raw_body end
+
+    local unified = {
+        content = "",
+        finish_reason = "",
+        token_usage = { prompt = 0, completion = 0, total = 0 }
+    }
+
+    if type(resp.usage) == "table" then
+        unified.token_usage.prompt = resp.usage.prompt_tokens or 0
+        unified.token_usage.completion = resp.usage.completion_tokens or 0
+        unified.token_usage.total = resp.usage.total_tokens or 0
+    end
+
+    if type(resp.choices) == "table" and #resp.choices > 0 then
+        local ch = resp.choices[1]
+        if type(ch.message) == "table" then
+            unified.content = ch.message.content or ""
+            if type(ch.message.tool_calls) == "table" then
+                local tcs = {}
+                for _, tc in ipairs(ch.message.tool_calls) do
+                    local args_ok, args = pcall(json.decode, tc["function"].arguments)
+                    if not args_ok then args = {} end
+                    table.insert(tcs, {
+                        id = tc.id,
+                        type = tc.type or "function",
+                        name = tc["function"].name,
+                        arguments = args
+                    })
+                end
+                unified.tool_calls = tcs
+            end
+        end
+        unified.finish_reason = ch.finish_reason or ""
+    end
+
+    return json.encode(unified)
+end
+
+function adapter.transform_stream_chunk(raw_chunk)
+    local ok, chunk = pcall(json.decode, raw_chunk)
+    if not ok then return "" end
+    if not chunk.choices or #chunk.choices == 0 then return "" end
+    local delta = chunk.choices[1].delta or {}
+    local fr = chunk.choices[1].finish_reason
+    return json.encode({
+        content = delta.content or "",
+        done = (fr ~= nil)
+    })
+end
+
+return adapter

internal/lua/adapters/kimicode.lua

@@ -0,0 +1,107 @@
+local adapter = {}
+
+adapter.name = "kimicode"
+adapter.version = "1.0.0"
+adapter.endpoint = "/v1/chat/completions"
+adapter.headers = {}
+
+-- KimiCode / Kimi K2 属于 OpenAI 兼容协议；但部分云端 API 会校验调用方
+-- "app"（只放行特定 agent），要求每次请求带上按 secret 计算的应用签名。
+-- 这里演示 build_headers 钩子：基于 timestamp + 请求体哈希生成签名头。
+--
+-- 配置要求（source.meta）:
+--   meta:
+--     app_id:   <申请到的 app id>
+--     app_key:  <你的 key（由网关的 base_url 复用 api_key 亦可）>
+--     app_secret: <签名密钥>
+--     app_agent: code-agent   # 若云端要求声明 agent 身份
+function adapter.transform_request(raw_body)
+    local ok, req = pcall(json.decode, raw_body)
+    if not ok then return raw_body end
+    req.model = req.model or "kimi-k2"
+    req.disable_thinking = nil
+    req.extra_body = nil
+    return json.encode(req)
+end
+
+-- 可选的动态签名钩子。meta 由 Go 注入：
+--   meta.url / meta.method / meta.body / meta.api_key / meta.timestamp / meta.source.meta
+function adapter.build_headers(meta)
+    local h = {
+        ["Content-Type"] = "application/json",
+        ["X-App-Id"] = tostring((meta.source.meta or {}).app_id or ""),
+        ["X-Timestamp"] = tostring(meta.timestamp),
+    }
+    local agent = (meta.source.meta or {}).app_agent
+    if agent and agent ~= "" then
+        h["X-Agent"] = agent
+    end
+    -- 校验 app：通常要求 Authorization 用 app secret 派生签名
+    local secret = (meta.source.meta or {}).app_secret
+    local api_key = meta.source.meta and meta.source.meta.api_key or meta.api_key
+    if secret and secret ~= "" then
+        local body_hash = sha256_hex(meta.body)
+        local sign_string = tostring(meta.timestamp) .. meta.method .. meta.url .. body_hash
+        local sign = hmac_sha256_hex(secret, sign_string)
+        h["Authorization"] = "Bearer " .. api_key
+        h["X-App-Sign"] = sign
+    else
+        h["Authorization"] = "Bearer " .. api_key
+    end
+    return h
+end
+
+function adapter.transform_response(raw_body)
+    local ok, resp = pcall(json.decode, raw_body)
+    if not ok or resp == nil then return raw_body end
+
+    local unified = {
+        content = "",
+        finish_reason = "",
+        token_usage = { prompt = 0, completion = 0, total = 0 }
+    }
+    if type(resp.usage) == "table" then
+        unified.token_usage.prompt = resp.usage.prompt_tokens or 0
+        unified.token_usage.completion = resp.usage.completion_tokens or 0
+        unified.token_usage.total = resp.usage.total_tokens or 0
+    end
+    if type(resp.choices) == "table" and #resp.choices > 0 then
+        local ch = resp.choices[1]
+        if type(ch.message) == "table" then
+            unified.content = ch.message.content or ""
+            if ch.message.reasoning_content then
+                unified.reasoning_content = ch.message.reasoning_content
+            end
+            if type(ch.message.tool_calls) == "table" then
+                local tcs = {}
+                for _, tc in ipairs(ch.message.tool_calls) do
+                    local args_ok, args = pcall(json.decode, tc["function"].arguments)
+                    if not args_ok then args = {} end
+                    table.insert(tcs, {
+                        id = tc.id,
+                        type = tc.type or "function",
+                        name = tc["function"].name,
+                        arguments = args
+                    })
+                end
+                unified.tool_calls = tcs
+            end
+        end
+        unified.finish_reason = ch.finish_reason or ""
+    end
+    return json.encode(unified)
+end
+
+function adapter.transform_stream_chunk(raw_chunk)
+    local ok, chunk = pcall(json.decode, raw_chunk)
+    if not ok then return "" end
+    if not chunk.choices or #chunk.choices == 0 then return "" end
+    local delta = chunk.choices[1].delta or {}
+    local fr = chunk.choices[1].finish_reason
+    return json.encode({
+        content = delta.content or "",
+        done = (fr ~= nil)
+    })
+end
+
+return adapter

internal/lua/adapters/mistral.lua

@@ -0,0 +1,74 @@
+local adapter = {}
+
+adapter.name = "mistral"
+adapter.version = "2.0.0"
+adapter.endpoint = "/v1/chat/completions"
+adapter.headers = {}
+
+-- Mistral API is OpenAI-compatible, just passes through
+function adapter.transform_request(raw_body)
+    local ok, req = pcall(json.decode, raw_body)
+    if not ok then return raw_body end
+    req.model = req.model or "mistral-large-latest"
+    req.temperature = req.temperature or 0.7
+    req.max_tokens = req.max_tokens or 4096
+    req.stream = req.stream or false
+    req.disable_thinking = nil
+    req.extra_body = nil
+    return json.encode(req)
+end
+
+function adapter.transform_response(raw_body)
+    local ok, resp = pcall(json.decode, raw_body)
+    if not ok or resp == nil then return raw_body end
+
+    local unified = {
+        content = "",
+        finish_reason = "",
+        token_usage = { prompt = 0, completion = 0, total = 0 }
+    }
+
+    if type(resp.usage) == "table" then
+        unified.token_usage.prompt = resp.usage.prompt_tokens or 0
+        unified.token_usage.completion = resp.usage.completion_tokens or 0
+        unified.token_usage.total = resp.usage.total_tokens or 0
+    end
+
+    if type(resp.choices) == "table" and #resp.choices > 0 then
+        local ch = resp.choices[1]
+        if type(ch.message) == "table" then
+            unified.content = ch.message.content or ""
+            if type(ch.message.tool_calls) == "table" then
+                local tcs = {}
+                for _, tc in ipairs(ch.message.tool_calls) do
+                    local args_ok, args = pcall(json.decode, tc["function"].arguments)
+                    if not args_ok then args = {} end
+                    table.insert(tcs, {
+                        id = tc.id,
+                        type = tc.type or "function",
+                        name = tc["function"].name,
+                        arguments = args
+                    })
+                end
+                unified.tool_calls = tcs
+            end
+        end
+        unified.finish_reason = ch.finish_reason or ""
+    end
+
+    return json.encode(unified)
+end
+
+function adapter.transform_stream_chunk(raw_chunk)
+    local ok, chunk = pcall(json.decode, raw_chunk)
+    if not ok then return "" end
+    if not chunk.choices or #chunk.choices == 0 then return "" end
+    local delta = chunk.choices[1].delta or {}
+    local fr = chunk.choices[1].finish_reason
+    return json.encode({
+        content = delta.content or "",
+        done = (fr ~= nil)
+    })
+end
+
+return adapter

internal/lua/adapters/ollama.lua

@@ -0,0 +1,63 @@
+local adapter = {}
+
+adapter.name = "ollama"
+adapter.version = "2.0.0"
+adapter.endpoint = "/api/chat"
+adapter.headers = {}
+
+-- Ollama API 格式：{ model, messages, stream, options:{temperature,num_predict} }
+function adapter.transform_request(raw_body)
+    local ok, req = pcall(json.decode, raw_body)
+    if not ok then return raw_body end
+
+    local ollama_req = {
+        model = req.model or "llama3",
+        stream = req.stream or false,
+        options = {
+            temperature = req.temperature or 0.7,
+            num_predict = req.max_tokens or 2048
+        }
+    }
+
+    -- 转换 messages 格式（Ollama 兼容 OpenAI 的 messages 格式）
+    if req.messages then
+        local msgs = {}
+        for _, m in ipairs(req.messages) do
+            table.insert(msgs, { role = m.role, content = m.content })
+        end
+        ollama_req.messages = msgs
+    end
+
+    return json.encode(ollama_req)
+end
+
+function adapter.transform_response(raw_body)
+    local ok, resp = pcall(json.decode, raw_body)
+    if not ok then return raw_body end
+
+    local unified = {
+        content = "",
+        finish_reason = resp.done_reason or "",
+        tool_calls = {},
+        usage = { prompt = 0, completion = 0, total = 0 }
+    }
+
+    if resp.message then
+        unified.content = resp.message.content or ""
+    end
+
+    return json.encode(unified)
+end
+
+function adapter.transform_stream_chunk(raw_chunk)
+    local ok, chunk = pcall(json.decode, raw_chunk)
+    if not ok then return "" end
+    if not chunk.message then return "" end
+
+    return json.encode({
+        content = chunk.message.content or "",
+        done = chunk.done or false
+    })
+end
+
+return adapter

internal/lua/adapters/openai.lua

@@ -0,0 +1,80 @@
+local adapter = {}
+
+adapter.name = "openai"
+adapter.version = "2.0.0"
+adapter.endpoint = "/chat/completions"
+adapter.headers = {}
+
+-- OpenAI /chat/completions format (pass-through, strip provider-specific fields)
+function adapter.transform_request(raw_body)
+    local ok, req = pcall(json.decode, raw_body)
+    if not ok then return raw_body end
+    req.disable_thinking = nil
+    req.extra_body = nil
+    if req.messages then
+        for _, msg in ipairs(req.messages) do
+            msg.reasoning_content = nil
+        end
+    end
+    return json.encode(req)
+end
+
+function adapter.transform_response(raw_body)
+    local ok, resp = pcall(json.decode, raw_body)
+    if not ok or resp == nil then return raw_body end
+
+    local unified = {
+        content = "",
+        finish_reason = "",
+        token_usage = { prompt = 0, completion = 0, total = 0 }
+    }
+
+    if type(resp.usage) == "table" then
+        unified.token_usage.prompt = resp.usage.prompt_tokens or 0
+        unified.token_usage.completion = resp.usage.completion_tokens or 0
+        unified.token_usage.total = resp.usage.total_tokens or 0
+    end
+
+    if type(resp.choices) == "table" and #resp.choices > 0 then
+        local ch = resp.choices[1]
+        if type(ch.message) == "table" then
+            unified.content = ch.message.content or ""
+            if ch.message.reasoning_content then
+                unified.reasoning_content = ch.message.reasoning_content
+            end
+            if type(ch.message.tool_calls) == "table" then
+                local tcs = {}
+                for _, tc in ipairs(ch.message.tool_calls) do
+                    local args_ok, args = pcall(json.decode, tc["function"].arguments)
+                    if not args_ok then args = {} end
+                    table.insert(tcs, {
+                        id = tc.id,
+                        type = tc.type or "function",
+                        name = tc["function"].name,
+                        arguments = args
+                    })
+                end
+                unified.tool_calls = tcs
+            end
+        end
+        unified.finish_reason = ch.finish_reason or ""
+    end
+
+    return json.encode(unified)
+end
+
+function adapter.transform_stream_chunk(raw_chunk)
+    local ok, chunk = pcall(json.decode, raw_chunk)
+    if not ok then return "" end
+
+    if not chunk.choices or #chunk.choices == 0 then return "" end
+    local delta = chunk.choices[1].delta or {}
+    local fr = chunk.choices[1].finish_reason
+
+    return json.encode({
+        content = delta.content or "",
+        done = (fr ~= nil)
+    })
+end
+
+return adapter