awnode 0.3.4__tar.gz → 0.3.5__tar.gz

{awnode-0.3.4 → awnode-0.3.5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: awnode
-Version: 0.3.4
+Version: 0.3.5
 Summary: awnode — lightweight local gateway for AitherOS
 Project-URL: Homepage, https://aitherium.com
 Author-email: Aitherium <hello@aitherium.com>
@@ -15,13 +15,10 @@ Requires-Dist: mcp<3.0.0,>=1.0.0
 Requires-Dist: pyyaml>=6.0
 Requires-Dist: uvicorn[standard]>=0.27.0
 Provides-Extra: all
-Requires-Dist: ollama>=0.1.0; extra == 'all'
 Requires-Dist: starlette>=0.27.0; extra == 'all'
 Provides-Extra: dev
 Requires-Dist: pytest-asyncio>=0.21.0; extra == 'dev'
 Requires-Dist: pytest>=7.0; extra == 'dev'
-Provides-Extra: ollama
-Requires-Dist: ollama>=0.1.0; extra == 'ollama'
 Provides-Extra: sse
 Requires-Dist: starlette>=0.27.0; extra == 'sse'
 Description-Content-Type: text/markdown
@@ -32,7 +29,7 @@ Description-Content-Type: text/markdown
 
 **[Docs](https://aitherium.github.io/awnode/)**  ·  [Source](https://github.com/Aitherium/awnode)  ·  `pip install awnode`  ·  [The Aither World](https://aitherium.github.io/)
 
-> **The Aither World** is an operating system for agents — a Linux you can hand to one, the runtimes it works in, and the tools it works with. [awnix](https://github.com/Aitherium/awnix) is the Linux underneath it; **awnode** is one of its 67 bricks — each installs on its own, runs offline, and needs no account.
+> **The Aither World** is an operating system for agents — a Linux you can hand to one, the runtimes it works in, and the tools it works with. [awnix](https://github.com/Aitherium/awnix) is the Linux underneath it; **awnode** is one of its 68 bricks — each installs on its own, runs offline, and needs no account.
 >
 > **Start here:** Run it once and point one app at it instead of a vendor URL.
 
@@ -54,7 +51,7 @@ curl -fsSL https://launch.aitherium.com | bash
 ## Quick Start
 
 ```bash
-# Start with auto-detection (finds Genesis, vLLM, llama.cpp, LM Studio, Ollama, or cloud)
+# Start with auto-detection (finds Genesis, vLLM, llama.cpp, LM Studio, or cloud)
 awnode start
 
 # Run the published container
@@ -83,8 +80,7 @@ awnode deploy my-service --target docker --strategy rolling
 3. **llama.cpp** (localhost:8200) / **Bonsai** (localhost:8092) — OpenAI-compatible llama.cpp servers
 4. **LM Studio** (127.0.0.1:1234) — LM Studio's local server (OpenAI-compatible; start it from LM Studio's Developer tab)
 5. **Elysium** (cloud) — hosted AitherOS (requires API key)
-6. **Ollama** (localhost:11434) — local CPU/GPU inference
-7. **Standalone** — no LLM, tools-only mode
+6. **Standalone** — no LLM, tools-only mode
 
 OpenAI-compatible backends (vLLM, llama.cpp, Bonsai, LM Studio, custom) are detected by a
 200 on `/v1/models`, never by `/health`, so an unrelated service on the same port is not
@@ -123,7 +119,7 @@ the read-only set; for full tool access from an IDE use `awnode mcp`.
 |---|---|---|
 | `AITHER_URL` | `http://localhost:8001` | Genesis URL |
 | `AITHER_VLLM_URL` | `http://localhost:8120` | vLLM URL |
-| `OLLAMA_HOST` | `http://localhost:11434` | Ollama URL |
+| `LLAMACPP_URL` | `http://localhost:8200` | llama.cpp `llama-server` URL (OpenAI-compatible) |
 | `AITHERNODE_LMSTUDIO_URL` | `http://127.0.0.1:1234` | LM Studio server URL (`LMSTUDIO_URL` also read) |
 | `AWNODE_EXTENSION_IDS` | | Extra extension ids admitted on the extension routes (comma-separated) |
 | `AWNODE_BROWSER_FULL_TOOLS` | | `1` lets direct loopback callers (browser/extension included) run every tool without a credential; tunnelled/remote callers stay read-only |
@@ -173,6 +169,7 @@ Each installs on its own, works offline, and needs no account.
 | [awrelay](https://github.com/Aitherium/awrelay) | a SaaS in the middle of your agents | findings, alerts and coordination over your own transport |
 | [awask](https://github.com/Aitherium/awask) | that anyone read the paragraph where you asked | the ask itself, with a button that steers the session that raised it |
 | [awmail](https://github.com/Aitherium/awmail) | a mailbox somebody else can read | mail your agents send and receive over your own server |
+| [awdiscord](https://github.com/Aitherium/awdiscord) | a self-bot on someone's account, or a bot that acts without asking | slash commands, rate limits, and approval buttons only your operators can press |
 | [awswarm](https://github.com/Aitherium/awswarm) | that a model either fits your GPU or it doesn't run at all | a placement plan and an acquisition-probability estimate before you spend on a run |
 | [awfind](https://github.com/Aitherium/awfind) | one vendor's idea of the web | results from whichever providers you configured |
 | [awbrowse](https://github.com/Aitherium/awbrowse) | that the page said what you were told | the render, the DOM and the requests it made |
@@ -243,6 +240,7 @@ Every repository here is public. Each publishes an `aither-manifest.json` beside
 | [awrelay](https://github.com/Aitherium/awrelay) | Portable agent messaging — findings, alerts, coordination | [docs](https://aitherium.github.io/awrelay/) |
 | [awask](https://github.com/Aitherium/awask) | Your agent asks you a question — and acts on your answer | [docs](https://aitherium.github.io/awask/) |
 | [awmail](https://github.com/Aitherium/awmail) | Give an agent an email address — send, and actually receive | [docs](https://aitherium.github.io/awmail/) |
+| [awdiscord](https://github.com/Aitherium/awdiscord) | Put any agent on Discord as a real bot, with an operator in the loop | — |
 | [awnet](https://github.com/Aitherium/awnet) | The agentic web — agents host a mesh, and agents join one | [docs](https://aitherium.github.io/awnet/) |
 | [awswarm](https://github.com/Aitherium/awswarm) | Run one model too big for any single GPU across a pool of small ones | — |
 | [awfind](https://github.com/Aitherium/awfind) | A portable search client — query, results, ranking | [docs](https://aitherium.github.io/awfind/) |

{awnode-0.3.4 → awnode-0.3.5}/README.md

@@ -4,7 +4,7 @@
 
 **[Docs](https://aitherium.github.io/awnode/)**  ·  [Source](https://github.com/Aitherium/awnode)  ·  `pip install awnode`  ·  [The Aither World](https://aitherium.github.io/)
 
-> **The Aither World** is an operating system for agents — a Linux you can hand to one, the runtimes it works in, and the tools it works with. [awnix](https://github.com/Aitherium/awnix) is the Linux underneath it; **awnode** is one of its 67 bricks — each installs on its own, runs offline, and needs no account.
+> **The Aither World** is an operating system for agents — a Linux you can hand to one, the runtimes it works in, and the tools it works with. [awnix](https://github.com/Aitherium/awnix) is the Linux underneath it; **awnode** is one of its 68 bricks — each installs on its own, runs offline, and needs no account.
 >
 > **Start here:** Run it once and point one app at it instead of a vendor URL.
 
@@ -26,7 +26,7 @@ curl -fsSL https://launch.aitherium.com | bash
 ## Quick Start
 
 ```bash
-# Start with auto-detection (finds Genesis, vLLM, llama.cpp, LM Studio, Ollama, or cloud)
+# Start with auto-detection (finds Genesis, vLLM, llama.cpp, LM Studio, or cloud)
 awnode start
 
 # Run the published container
@@ -55,8 +55,7 @@ awnode deploy my-service --target docker --strategy rolling
 3. **llama.cpp** (localhost:8200) / **Bonsai** (localhost:8092) — OpenAI-compatible llama.cpp servers
 4. **LM Studio** (127.0.0.1:1234) — LM Studio's local server (OpenAI-compatible; start it from LM Studio's Developer tab)
 5. **Elysium** (cloud) — hosted AitherOS (requires API key)
-6. **Ollama** (localhost:11434) — local CPU/GPU inference
-7. **Standalone** — no LLM, tools-only mode
+6. **Standalone** — no LLM, tools-only mode
 
 OpenAI-compatible backends (vLLM, llama.cpp, Bonsai, LM Studio, custom) are detected by a
 200 on `/v1/models`, never by `/health`, so an unrelated service on the same port is not
@@ -95,7 +94,7 @@ the read-only set; for full tool access from an IDE use `awnode mcp`.
 |---|---|---|
 | `AITHER_URL` | `http://localhost:8001` | Genesis URL |
 | `AITHER_VLLM_URL` | `http://localhost:8120` | vLLM URL |
-| `OLLAMA_HOST` | `http://localhost:11434` | Ollama URL |
+| `LLAMACPP_URL` | `http://localhost:8200` | llama.cpp `llama-server` URL (OpenAI-compatible) |
 | `AITHERNODE_LMSTUDIO_URL` | `http://127.0.0.1:1234` | LM Studio server URL (`LMSTUDIO_URL` also read) |
 | `AWNODE_EXTENSION_IDS` | | Extra extension ids admitted on the extension routes (comma-separated) |
 | `AWNODE_BROWSER_FULL_TOOLS` | | `1` lets direct loopback callers (browser/extension included) run every tool without a credential; tunnelled/remote callers stay read-only |
@@ -145,6 +144,7 @@ Each installs on its own, works offline, and needs no account.
 | [awrelay](https://github.com/Aitherium/awrelay) | a SaaS in the middle of your agents | findings, alerts and coordination over your own transport |
 | [awask](https://github.com/Aitherium/awask) | that anyone read the paragraph where you asked | the ask itself, with a button that steers the session that raised it |
 | [awmail](https://github.com/Aitherium/awmail) | a mailbox somebody else can read | mail your agents send and receive over your own server |
+| [awdiscord](https://github.com/Aitherium/awdiscord) | a self-bot on someone's account, or a bot that acts without asking | slash commands, rate limits, and approval buttons only your operators can press |
 | [awswarm](https://github.com/Aitherium/awswarm) | that a model either fits your GPU or it doesn't run at all | a placement plan and an acquisition-probability estimate before you spend on a run |
 | [awfind](https://github.com/Aitherium/awfind) | one vendor's idea of the web | results from whichever providers you configured |
 | [awbrowse](https://github.com/Aitherium/awbrowse) | that the page said what you were told | the render, the DOM and the requests it made |
@@ -215,6 +215,7 @@ Every repository here is public. Each publishes an `aither-manifest.json` beside
 | [awrelay](https://github.com/Aitherium/awrelay) | Portable agent messaging — findings, alerts, coordination | [docs](https://aitherium.github.io/awrelay/) |
 | [awask](https://github.com/Aitherium/awask) | Your agent asks you a question — and acts on your answer | [docs](https://aitherium.github.io/awask/) |
 | [awmail](https://github.com/Aitherium/awmail) | Give an agent an email address — send, and actually receive | [docs](https://aitherium.github.io/awmail/) |
+| [awdiscord](https://github.com/Aitherium/awdiscord) | Put any agent on Discord as a real bot, with an operator in the loop | — |
 | [awnet](https://github.com/Aitherium/awnet) | The agentic web — agents host a mesh, and agents join one | [docs](https://aitherium.github.io/awnet/) |
 | [awswarm](https://github.com/Aitherium/awswarm) | Run one model too big for any single GPU across a pool of small ones | — |
 | [awfind](https://github.com/Aitherium/awfind) | A portable search client — query, results, ranking | [docs](https://aitherium.github.io/awfind/) |

{awnode-0.3.4 → awnode-0.3.5}/aither-constellation.js

@@ -42,6 +42,7 @@
     { id: "awrecover", label: "awrecover", blurb: "Labelled snapshots with an all-or-nothing..." },
     { id: "awrelay", label: "awrelay", blurb: "Portable agent messaging — findings, alerts,..." },
     { id: "awmail", label: "awmail", blurb: "Give an agent an email address — send, and..." },
+    { id: "awdiscord", label: "awdiscord", blurb: "Put any agent on Discord as a real bot,..." },
     { id: "awnet", label: "awnet", blurb: "The agentic web — agents host a mesh, and..." },
     { id: "awfind", label: "awfind", blurb: "A portable search client — query, results,..." },
     { id: "awbrowse", label: "awbrowse", blurb: "A portable browser client — navigate,..." },

{awnode-0.3.4 → awnode-0.3.5}/awnode/__init__.py

@@ -7,7 +7,7 @@ The bridge between local apps (Shell, Desktop, Connect) and AitherOS services.
 Modes:
 1. **Local Genesis** — proxy to localhost:8001 (full AitherOS Docker stack)
 2. **Cloud** — proxy to mcp.aitherium.com (SaaS inference + MCP tools)
-3. **Standalone** — local Ollama + basic tools (no Genesis needed)
+3. **Standalone** — local llama.cpp + basic tools (no Genesis needed)
 4. **Hybrid** — local tools + cloud LLM (best of both)
 
 Install:
@@ -17,4 +17,4 @@ Install:
     aithernode status              # Check what's connected
 """
 
-__version__ = "0.3.4"
+__version__ = "0.3.5"

{awnode-0.3.4 → awnode-0.3.5}/awnode/cli.py

@@ -72,7 +72,7 @@ def main():
     start_p.add_argument("--cloud", action="store_true", help="Force cloud mode (Elysium)")
     start_p.add_argument("--local", action="store_true", help="Force local mode (no cloud)")
     start_p.add_argument("--vllm-url", type=str, help="vLLM URL (default: http://localhost:8120)")
-    start_p.add_argument("--ollama-url", type=str, help="Ollama URL (default: http://localhost:11434)")
+    start_p.add_argument("--llamacpp-url", type=str, help="llama.cpp server URL (default: http://localhost:8200)")
     start_p.add_argument("--genesis-url", type=str, help="Genesis URL (default: http://localhost:8001)")
 
     # status
@@ -105,7 +105,8 @@ def main():
              "AitherOS fabric. Omit for the default GPU/backend setup.",
     )
     setup_p.add_argument("--cloud", action="store_true", help="Cloud-only mode (skip local GPU)")
-    setup_p.add_argument("--model", type=str, help="Specific Ollama model to pull")
+    setup_p.add_argument("--model", type=str,
+                         help="GGUF model id the local llama.cpp server should serve")
     setup_p.add_argument("--skip-model", action="store_true", help="Skip model download")
     # `setup comfyui` flags (ignored by the bare `setup` path)
     setup_p.add_argument("--profile", type=str, default="studio",
@@ -191,8 +192,8 @@ def main():
             os.environ["AITHERNODE_MODE"] = "local"
         if args.vllm_url:
             os.environ["AITHER_VLLM_URL"] = args.vllm_url
-        if args.ollama_url:
-            os.environ["OLLAMA_HOST"] = args.ollama_url
+        if args.llamacpp_url:
+            os.environ["LLAMACPP_URL"] = args.llamacpp_url
         if args.genesis_url:
             os.environ["AITHER_URL"] = args.genesis_url
         os.environ["AITHERNODE_PORT"] = str(args.port)
@@ -347,7 +348,7 @@ async def _doctor(*, emit_json: bool = False, timeout: float = 3.0) -> int:
     backends = {
         "genesis": os.environ.get("AITHER_URL", "http://localhost:8001"),
         "vllm": os.environ.get("AITHER_VLLM_URL", "http://localhost:8120"),
-        "ollama": os.environ.get("OLLAMA_HOST", "http://localhost:11434"),
+        "llamacpp": os.environ.get("LLAMACPP_URL", "http://localhost:8200"),
         # LM Studio has no /health; its OpenAI-compatible /v1/models is the liveness surface.
         "lmstudio": os.environ.get(
             "AITHERNODE_LMSTUDIO_URL", os.environ.get("LMSTUDIO_URL", "http://127.0.0.1:1234")
@@ -356,7 +357,7 @@ async def _doctor(*, emit_json: bool = False, timeout: float = 3.0) -> int:
     }
     async with httpx.AsyncClient(timeout=timeout) as client:
         for name, url in backends.items():
-            health_path = {"ollama": "/api/tags", "lmstudio": "/v1/models"}.get(name, "/health")
+            health_path = {"lmstudio": "/v1/models"}.get(name, "/health")
             try:
                 r = await client.get(url.rstrip("/") + health_path)
                 if r.status_code < 500:
@@ -364,7 +365,7 @@ async def _doctor(*, emit_json: bool = False, timeout: float = 3.0) -> int:
                 else:
                     record(name, "warn", f"{url} -> {r.status_code}")
             except Exception as e:
-                # gateway/vllm/ollama unreachable is a warn (not everyone runs them)
+                # gateway/vllm/llamacpp unreachable is a warn (not everyone runs them)
                 record(name, "warn", f"{url} unreachable: {type(e).__name__}")
 
     # Compute exit code
@@ -403,7 +404,7 @@ async def _status():
     checks = {
         "Genesis": os.environ.get("AITHER_URL", "http://localhost:8001"),
         "vLLM": os.environ.get("AITHER_VLLM_URL", os.environ.get("VLLM_URL", "http://localhost:8120")),
-        "Ollama": os.environ.get("OLLAMA_HOST", "http://localhost:11434"),
+        "llama.cpp": os.environ.get("LLAMACPP_URL", "http://localhost:8200"),
         "Elysium": "https://mcp.aitherium.com",
     }
 
@@ -412,7 +413,7 @@ async def _status():
     for name, url in checks.items():
         try:
             async with httpx.AsyncClient(timeout=3.0) as c:
-                health_path = "/api/tags" if name == "Ollama" else "/health"
+                health_path = "/health"
                 r = await c.get(f"{url.rstrip('/')}{health_path}")
                 status = "UP" if r.status_code == 200 else f"HTTP {r.status_code}"
         except Exception:
@@ -684,7 +685,7 @@ async def _connect(
         try:
             result = await gw.register_agent(
                 name=f"node-{hostname}",
-                capabilities=["chat", "code", "filesystem", "git", "vllm", "ollama"],
+                capabilities=["chat", "code", "filesystem", "git", "vllm", "llamacpp"],
                 description=f"awnode on {hostname} ({platform.system()})",
             )
             print(f"Connected! Agent ID: {result.get('agent_id', '?')}")
@@ -796,42 +797,25 @@ async def _setup(cloud_only: bool = False, model: str = None, skip_model: bool =
         print("  Local models will use your GPU for fast inference.")
     elif gpu_vram >= 4096:
         cfg["mode"] = "auto"
-        print(f"  Recommended: Ollama with small models ({gpu_vram}MB VRAM)")
+        print(f"  Recommended: llama.cpp with a small GGUF model ({gpu_vram}MB VRAM)")
     else:
         cfg["mode"] = "auto"
-        print("  Recommended: Ollama CPU mode or cloud")
-
-    # -- Ollama Detection & Model Pull --
-    has_ollama = shutil.which("ollama") is not None
-    if has_ollama:
-        print("  Ollama: installed")
-        cfg["ollama"] = {"available": True}
-
-        if not skip_model and not cloud_only:
-            target_model = model
-            if not target_model:
-                if gpu_vram >= 16384:
-                    target_model = "llama3.1:8b"
-                elif gpu_vram >= 8192:
-                    target_model = "llama3.2:3b"
-                else:
-                    target_model = "llama3.2:1b"
-
-            print(f"\n  Pulling model: {target_model}...")
-            pull_result = subprocess.run(
-                ["ollama", "pull", target_model],
-                timeout=600,
-            )
-            if pull_result.returncode == 0:
-                cfg["ollama"]["default_model"] = target_model
-                print(f"  Model ready: {target_model}")
-            else:
-                print(f"  Model pull failed (exit {pull_result.returncode})")
+        print("  Recommended: llama.cpp CPU mode or cloud")
+
+    # -- llama.cpp Detection --
+    # Weights are GGUF files from artifact.aitherium.com (awrtifact); awnode does not
+    # download them -- it records which model the local llama-server should serve.
+    has_llamacpp = shutil.which("llama-server") is not None
+    cfg["llamacpp"] = {"available": has_llamacpp}
+    if has_llamacpp:
+        print("  llama.cpp: llama-server installed")
     else:
-        print("  Ollama: not installed")
+        print("  llama.cpp: llama-server not on PATH")
         if not cloud_only:
-            print("    Install: https://ollama.com/download")
-        cfg["ollama"] = {"available": False}
+            print("    Install: https://github.com/ggml-org/llama.cpp/releases")
+    if model and not skip_model and not cloud_only:
+        cfg["llamacpp"]["default_model"] = model
+        print(f"  Default local model: {model}")
 
     # -- vLLM Detection --
     import httpx

{awnode-0.3.4 → awnode-0.3.5}/awnode/extensions.py

@@ -9,8 +9,8 @@ alongside awnode on customer hardware.
 `awdk.adk.addon_manager` reads component manifests from three sources (a pip-installed
 brick's `aither.components` entry point, `~/.aither/components/*.yaml`, the bundled
 `adk/addon_manifests/`) and this manager renders them in the shape `awnode ext` has
-always shown. The nine entries that used to be hardcoded in BUILTIN_CATALOG (canvas,
-comfyui, comfyui-3d, muse, sd-webui, ollama, iris, design, autorig) are manifests now,
+always shown. The entries that used to be hardcoded in BUILTIN_CATALOG (canvas,
+comfyui, comfyui-3d, muse, sd-webui, iris, design, autorig) are manifests now,
 so `awnode ext start canvas` runs the same declaration awdk, the aitheros launcher, awsh
 and the Living Desktop read. Two hosts with two catalogs of "what can run here" was the
 silence this ended: an extension present in one and absent from the other looked like

{awnode-0.3.4 → awnode-0.3.5}/awnode/hub.py

@@ -90,6 +90,18 @@ DEFAULT_BRICKS: List[Dict[str, Any]] = [
     {"name": "awfind", "main": "awfind.cli:main", "args": ["mcp"]},
     {"name": "awprism", "main": "awprism.cli:main", "args": ["mcp"]},
     {"name": "awsh", "module": "adk.harnesses.mcp_stdio"},
+    # REA reverse-engineering toolkit. The heavy toolchain (Node 22, Ghidra, the
+    # pinned rea-agents CLI) lives in the policy-gated ``aitheros-rea`` awnix
+    # sidecar, never in this process; this brick is a thin stdio->HTTP bridge to
+    # that sidecar's MCP endpoint, so every rea_* call is served THROUGH the
+    # sidecar's in-process policy proxy (lib.agents.packs.rea.policy.authorize:
+    # target_class + per-tool grant, fail-closed). Disabled by default so boots
+    # on a host without the sidecar never try to start it; enable per host via
+    # ~/.aither/aw-hub config ``{"bricks": {"rea": {"disabled": false}}}`` or the
+    # AITHER_REA_SIDECAR_URL env the argv reads.
+    {"name": "rea", "disabled": True, "module": "awnode.rea_sidecar_bridge",
+     "args": ["--url", os.environ.get("AITHER_REA_SIDECAR_URL",
+                                      "https://aitheros-rea:8791/mcp")]},
 ]
 
 _CREATE_NO_WINDOW = 0x08000000

{awnode-0.3.4 → awnode-0.3.5}/awnode/license.py

@@ -13,7 +13,7 @@ Flow:
 
 If the subscription lapses:
   - Tools/agents/extensions that require a paid tier stop working
-  - Free-tier tools (filesystem, git, search, Ollama) keep working
+  - Free-tier tools (filesystem, git, search, local llama.cpp) keep working
   - Clear error messages tell the user to renew
 
 If offline:
@@ -49,8 +49,8 @@ PLAN_FEATURES: Dict[str, Dict[str, Any]] = {
     "free": {
         "tools": {"filesystem", "git_tools", "search", "commands", "context",
                   "memory", "knowledge", "notebooks", "documents", "mermaid",
-                  "http_client", "ollama", "orchestrator", "embeddings"},
-        "extensions": {"ollama"},
+                  "http_client", "orchestrator", "embeddings"},
+        "extensions": set(),
         "agents": set(),
         "max_agents": 1,
         "image_generation": False,
@@ -60,9 +60,9 @@ PLAN_FEATURES: Dict[str, Dict[str, Any]] = {
     "developer": {
         "tools": {"filesystem", "git_tools", "search", "commands", "context",
                   "memory", "knowledge", "notebooks", "documents", "mermaid",
-                  "http_client", "ollama", "orchestrator", "embeddings",
+                  "http_client", "orchestrator", "embeddings",
                   "generation", "extensions", "vision"},
-        "extensions": {"ollama", "canvas", "comfyui"},
+        "extensions": {"canvas", "comfyui"},
         "agents": {"iris"},
         "max_agents": 3,
         "image_generation": True,
@@ -71,7 +71,7 @@ PLAN_FEATURES: Dict[str, Dict[str, Any]] = {
     },
     "pro": {
         "tools": {"*"},  # All tools
-        "extensions": {"ollama", "canvas", "comfyui", "comfyui-3d", "muse",
+        "extensions": {"canvas", "comfyui", "comfyui-3d", "muse",
                        "sd-webui", "iris", "design"},
         "agents": {"iris", "demiurge", "atlas", "hydra", "viviane"},
         "max_agents": 15,

{awnode-0.3.4 → awnode-0.3.5}/awnode/proxy.py

@@ -57,8 +57,8 @@ _DEFAULT_SERVICES: Dict[str, str] = {
     "genesis": os.environ.get("AITHER_URL", "http://127.0.0.1:8001"),
     # Bonsai-27B llama.cpp server (PrismML fork) — OpenAI-compatible /v1
     "bonsai": os.environ.get("AITHER_BONSAI_URL", "http://127.0.0.1:8092"),
-    # Ollama — generic local model host
-    "ollama": os.environ.get("OLLAMA_HOST", "http://127.0.0.1:11434"),
+    # llama.cpp llama-server — generic local OpenAI-compatible model host
+    "llamacpp": os.environ.get("LLAMACPP_URL", "http://127.0.0.1:8200"),
     # AitherMarketplace — unified catalog (agent/skill/tool packs, apps,
     # tools, plugins, extensions) with the fail-closed entitlement gate.
     "marketplace": os.environ.get("AITHER_MARKETPLACE_URL", "https://127.0.0.1:8260"),
@@ -80,7 +80,7 @@ def _load_service_map() -> Dict[str, str]:
 
     def _norm(url: str) -> str:
         url = url.strip().rstrip("/")
-        # OLLAMA_HOST is commonly "0.0.0.0:11434" (no scheme) — normalize so
+        # A host env is commonly "0.0.0.0:<port>" (no scheme) — normalize so
         # httpx gets an absolute URL, and rewrite wildcard binds to loopback.
         if url and "://" not in url:
             url = f"http://{url}"