pi-opencode-go-provider 1.0.15 → 1.0.17

package/.pi/fabric/mesh/state.json

@@ -0,0 +1,117 @@
+{
+  "format": 1,
+  "entries": {
+    "topology/peer-seq": {
+      "key": "topology/peer-seq",
+      "value": {
+        "format": 1,
+        "next": 1
+      },
+      "version": 1,
+      "updatedAt": 1788421061231,
+      "updatedBy": {
+        "id": "session:01a06632-9590-73a7-8e4c-bf4f4d6d8277",
+        "name": "main",
+        "kind": "main",
+        "sessionId": "01a06632-9590-73a7-8e4c-bf4f4d6d8277"
+      }
+    },
+    "topology/hosts/ae5878888fad007023c5ff775ac6356930ec9eaacf6e5b09ceebb18d9d4e54c8": {
+      "key": "topology/hosts/ae5878888fad007023c5ff775ac6356930ec9eaacf6e5b09ceebb18d9d4e54c8",
+      "value": {
+        "format": 1,
+        "id": "session:01a06632-9590-73a7-8e4c-bf4f4d6d8277",
+        "rootId": "session:01a06632-9590-73a7-8e4c-bf4f4d6d8277",
+        "identity": {
+          "id": "session:01a06632-9590-73a7-8e4c-bf4f4d6d8277",
+          "name": "main",
+          "kind": "main",
+          "sessionId": "01a06632-9590-73a7-8e4c-bf4f4d6d8277"
+        },
+        "startedAt": 1788421061220,
+        "updatedAt": 1788422266448,
+        "expiresAt": 1788422281448
+      },
+      "version": 367,
+      "updatedAt": 1788422266450,
+      "updatedBy": {
+        "id": "session:01a06632-9590-73a7-8e4c-bf4f4d6d8277",
+        "name": "main",
+        "kind": "main",
+        "sessionId": "01a06632-9590-73a7-8e4c-bf4f4d6d8277"
+      }
+    },
+    "sessions/01a06632-9590-73a7-8e4c-bf4f4d6d8277": {
+      "key": "sessions/01a06632-9590-73a7-8e4c-bf4f4d6d8277",
+      "value": {
+        "id": "session:01a06632-9590-73a7-8e4c-bf4f4d6d8277",
+        "name": "POGP-1",
+        "label": "POGP-1",
+        "kind": "peer",
+        "status": "running",
+        "runner": "pi",
+        "transport": "host",
+        "cwd": "/Users/monotykamary/VCS/working-remote/open-source/pi-opencode-go-provider",
+        "sessionId": "01a06632-9590-73a7-8e4c-bf4f4d6d8277",
+        "model": "xai/grok-4.6",
+        "thinking": "medium",
+        "startedAt": 1788421061220,
+        "updatedAt": 1788422266448,
+        "pendingMessages": false,
+        "local": false
+      },
+      "version": 367,
+      "updatedAt": 1788422266451,
+      "updatedBy": {
+        "id": "session:01a06632-9590-73a7-8e4c-bf4f4d6d8277",
+        "name": "main",
+        "kind": "main",
+        "sessionId": "01a06632-9590-73a7-8e4c-bf4f4d6d8277"
+      }
+    },
+    "topology/participants/ae5878888fad007023c5ff775ac6356930ec9eaacf6e5b09ceebb18d9d4e54c8": {
+      "key": "topology/participants/ae5878888fad007023c5ff775ac6356930ec9eaacf6e5b09ceebb18d9d4e54c8",
+      "value": {
+        "format": 1,
+        "id": "session:01a06632-9590-73a7-8e4c-bf4f4d6d8277",
+        "kind": "root",
+        "rootId": "session:01a06632-9590-73a7-8e4c-bf4f4d6d8277",
+        "ownerHostId": "session:01a06632-9590-73a7-8e4c-bf4f4d6d8277",
+        "ownerIdentityId": "session:01a06632-9590-73a7-8e4c-bf4f4d6d8277",
+        "name": "main",
+        "status": "running",
+        "runner": "pi",
+        "transport": "host",
+        "capabilities": [
+          "steer",
+          "followUp",
+          "fabric"
+        ],
+        "cwd": "/Users/monotykamary/VCS/working-remote/open-source/pi-opencode-go-provider",
+        "sessionId": "01a06632-9590-73a7-8e4c-bf4f4d6d8277",
+        "model": "xai/grok-4.6",
+        "thinking": "medium",
+        "startedAt": 1788421061220,
+        "updatedAt": 1788422266448,
+        "pendingMessages": false,
+        "controlProtocol": "v1",
+        "label": "POGP-1"
+      },
+      "version": 367,
+      "updatedAt": 1788422266453,
+      "updatedBy": {
+        "id": "session:01a06632-9590-73a7-8e4c-bf4f4d6d8277",
+        "name": "main",
+        "kind": "main",
+        "sessionId": "01a06632-9590-73a7-8e4c-bf4f4d6d8277"
+      }
+    }
+  },
+  "versions": {
+    "topology/peer-seq": 1,
+    "topology/hosts/ae5878888fad007023c5ff775ac6356930ec9eaacf6e5b09ceebb18d9d4e54c8": 367,
+    "sessions/01a06632-9590-73a7-8e4c-bf4f4d6d8277": 367,
+    "topology/participants/ae5878888fad007023c5ff775ac6356930ec9eaacf6e5b09ceebb18d9d4e54c8": 367
+  },
+  "tombstoneOrder": []
+}

package/README.md

@@ -20,6 +20,7 @@ _Go-optimized endpoints for lower latency — 14+ models for [pi](https://github
 - **Fast & Efficient** - Go-optimized endpoints for lower latency
 - **Cost Tracking** with per-model pricing for budget management
 - **Reasoning Models** with thinking level maps for proper effort control
+- **Prompt-cache session affinity** — sends `x-opencode-session` and `x-opencode-client` so OpenCode Go can pin a session to the same cache node
 
 ## Installation
 
@@ -77,7 +78,7 @@ pi
 | GLM-5.3-Flash (2x usage) | Completions | Text + Image | 1.0M | 131K | $0.07 | $0.25 |
 | GPT-5.6 Luna | Responses | Text + Image | 1.1M | 128K | $0.20 | $1.20 |
 | Grok 4.6 | Completions | Text + Image | 500K | 500K | $2.00 | $6.00 |
-| Hy3 | Completions | Text | 256K | 64K | $0.14 | $0.58 |
+| Hy3 | Completions | Text | 256K | 128K | $0.14 | $0.58 |
 | Hy4 preview | Completions | Text | 1.0M | 64K | $0.83 | $2.50 |
 | Kimi K2.6 | Completions | Text + Image | 262K | 66K | $0.95 | $4.00 |
 | Kimi K2.7 Code | Completions | Text + Image | 262K | 262K | $0.95 | $4.00 |
@@ -88,6 +89,7 @@ pi
 | MiniMax-M2.7 | Completions | Text | 205K | 131K | $0.30 | $1.20 |
 | MiniMax-M3 | Anthropic | Text + Image | 1.0M | 131K | $0.30 | $1.20 |
 | Muse Spark 1.2 Contributor | Completions | Text + Image | 1.0M | 131K | $0.10 | $0.20 |
+| Muse Spark 1.3 Contributor | Completions | Text + Image | 1.0M | 131K | $0.10 | $0.20 |
 | Qwen3.6 Plus | Completions | Text + Image | 1.0M | 66K | $0.50 | $3.00 |
 | Qwen3.7 Max | Completions | Text | 1.0M | 66K | $2.50 | $7.50 |
 | Qwen3.7 Plus | Completions | Text + Image | 1.0M | 66K | $0.40 | $1.60 |

package/index.ts

@@ -339,7 +339,40 @@ async function resolveApiKey(modelRegistry: ModelRegistry): Promise<void> {
   cachedApiKey = await modelRegistry.getApiKeyForProvider("opencode-go") ?? undefined;
 }
 
-// ─── Extension Entry Point ────────────────────────────────────────────────────
+// OpenCode Go prompt-cache routing headers.
+// Pi core already injects these in mergeProviderAttributionHeaders for
+// provider "opencode-go" / host opencode.ai (see #4847). This hook is a
+// backstop: it fills them only when missing, so older pi builds and any
+// request path that skipped transformHeaders still pin the session.
+// Per-request via before_provider_headers — not registerProvider.headers —
+// because provider-wide headers leak to raw helper streams
+// (getApiKeyAndHeaders) that must not inherit the main session's cache
+// lineage. No custom streamSimple is needed; headers are applied before
+// provider dispatch.
+const OPENCODE_SESSION_HEADER = "x-opencode-session";
+const OPENCODE_CLIENT_HEADER = "x-opencode-client";
+const OPENCODE_CLIENT = "pi";
+
+function headerLookup(headers: Record<string, string | null | undefined>, name: string): string | undefined {
+  const lower = name.toLowerCase();
+  for (const [key, value] of Object.entries(headers)) {
+    if (key.toLowerCase() === lower) return value ?? undefined;
+  }
+  return undefined;
+}
+
+export function applyOpenCodeSessionHeaders(
+  headers: Record<string, string | null>,
+  sessionId: string | undefined,
+): Record<string, string | null> {
+  if (sessionId && !headerLookup(headers, OPENCODE_SESSION_HEADER)) {
+    headers[OPENCODE_SESSION_HEADER] = sessionId;
+  }
+  if (!headerLookup(headers, OPENCODE_CLIENT_HEADER)) {
+    headers[OPENCODE_CLIENT_HEADER] = OPENCODE_CLIENT;
+  }
+  return headers;
+}
 
 export default function (pi: ExtensionAPI) {
   const embeddedModels = modelsData as JsonModel[];
@@ -369,6 +402,11 @@ export default function (pi: ExtensionAPI) {
     })),
   });
 
+  pi.on("before_provider_headers", (event, ctx) => {
+    if (ctx.model?.provider !== PROVIDER_ID) return;
+    applyOpenCodeSessionHeaders(event.headers, ctx.sessionManager.getSessionId());
+  });
+
   pi.on("session_start", async (_event, ctx) => {
     revalidateAbort?.abort();
     revalidateAbort = new AbortController();

package/models.json

@@ -64,6 +64,34 @@
     "contextWindow": 1048576,
     "maxTokens": 131072
   },
+  {
+    "id": "muse-spark-1.3-contributor",
+    "name": "Muse Spark 1.3 Contributor",
+    "api": "openai-completions",
+    "baseUrl": "https://opencode.ai/zen/go/v1",
+    "reasoning": true,
+    "thinkingLevelMap": {
+      "off": null,
+      "minimal": "minimal",
+      "low": "low",
+      "medium": "medium",
+      "high": "high",
+      "xhigh": "xhigh",
+      "max": null
+    },
+    "input": [
+      "text",
+      "image"
+    ],
+    "cost": {
+      "input": 0.1,
+      "output": 0.2,
+      "cacheRead": 0.002,
+      "cacheWrite": 0
+    },
+    "contextWindow": 1048576,
+    "maxTokens": 131072
+  },
   {
     "id": "deepseek-v4-flash",
     "name": "DeepSeek V4 Flash",
@@ -484,7 +512,7 @@
       "cacheWrite": 0
     },
     "contextWindow": 256000,
-    "maxTokens": 64000
+    "maxTokens": 128000
   },
   {
     "id": "muse-spark-1.2-contributor",

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "pi-opencode-go-provider",
-  "version": "1.0.15",
+  "version": "1.0.17",
   "description": "Opencode Go provider extension for pi - Fast, efficient GLM, Kimi, and MiniMax models through the opencode.ai API",
   "type": "module",
   "main": "index.ts",