[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7CSQGETQEN899AN2HC992TG":3,"skill-same:01M7CSQGETQEN899AN2HC992TG":87},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":24,"commitSha":25,"skillDirectory":26,"forksCount":27,"repositoryHtmlUrl":28,"repositoryHomepageUrl":29,"ownerHtmlUrl":30,"ownerBlogUrl":6,"ownerDescription":31,"updatedAt":32,"install":33,"resources":46},"01M7CSQGETQEN899AN2HC992TG","nemoclaw-user-configure-inference",null,"Connects NemoClaw to a local inference server. Use when setting up Ollama, vLLM, TensorRT-LLM, NIM, or any OpenAI-compatible local model server with NemoClaw. Trigger keywords - nemoclaw local inference, ollama nemoclaw, vllm nemoclaw, local model server, openai compatible endpoint, switch nemoclaw inference model, change inference runtime, nemoclaw additional model, nemoclaw sub-agent model, openclaw sub-agent, agents.list, sessions_spawn, vlm-demo, nemoclaw tool calling, ollama tool calls, vllm tool-call-parser, raw json in tui, nemoclaw inference options, nemoclaw onboarding providers, nemoclaw inference routing.","YYC-Cube\u002FYYC3-Knowledge-Base","YYC-Cube","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F235006543?v=4","01M79J2PH4SWD2X0825X6PN748","yyc3","01M79J2PHVK3F6ZBYT8K2SCZCW",1,"Apache-2.0","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-Skills-技能包\u002FYYC3-Skills-技能包-skills\u002Fskills\u002Fnemoclaw-user-configure-inference\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: \"nemoclaw-user-configure-inference\"\ndescription: \"Connects NemoClaw to a local inference server. Use when setting up Ollama, vLLM, TensorRT-LLM, NIM, or any OpenAI-compatible local model server with NemoClaw. Trigger keywords - nemoclaw local inference, ollama nemoclaw, vllm nemoclaw, local model server, openai compatible endpoint, switch nemoclaw inference model, change inference runtime, nemoclaw additional model, nemoclaw sub-agent model, openclaw sub-agent, agents.list, sessions_spawn, vlm-demo, nemoclaw tool calling, ollama tool calls, vllm tool-call-parser, raw json in tui, nemoclaw inference options, nemoclaw onboarding providers, nemoclaw inference routing.\"\nlicense: \"Apache-2.0\"\n---\n\n\u003C!-- SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved. -->\n\u003C!-- SPDX-License-Identifier: Apache-2.0 -->\n\n# Use a Local Inference Server\n\n## Gotchas\n\n- Ollama is convenient for local chat, but some model\u002Ftemplate combinations can return tool calls as plain text under realistic agent load.\n\n## Prerequisites\n\n- NemoClaw installed.\n- A local model server running, or a supported Ollama, vLLM, or NIM setup that the NemoClaw onboard wizard can use, start, or install.\n\nNemoClaw can route inference to a model server running on your machine instead of a cloud API.\nThis page covers Ollama, compatible-endpoint paths for other servers, and experimental managed options for vLLM and NVIDIA NIM.\n\nAll approaches use the same `inference.local` routing model.\nThe agent inside the sandbox never connects to your model server directly.\nOpenShell intercepts inference traffic and forwards it to the local endpoint you configure.\n\n## Ollama\n\nOllama is the default local inference option.\nThe onboard wizard detects Ollama automatically when it is installed or running on the host.\n\nIf Ollama is installed but not running, NemoClaw starts it for you.\nOn macOS and Linux, the wizard can also offer to install Ollama when it is not present.\nWhen the host Ollama is below the minimum version NemoClaw expects for its starter models (currently `0.7.0`), the wizard surfaces an explicit **Upgrade Ollama** entry in the provider menu instead of silently reusing the older daemon, and the express setup path resolves to that entry.\nThe wizard inspects both the CLI binary (`ollama --version`) and the locally running daemon (`\u002Fapi\u002Fversion` on `:11434`) so the upgrade entry still appears when only one side is stale, for example a fresh user-local binary paired with the original system daemon.\nThe gate skips Windows-host Ollama reached from WSL via `host.docker.internal`; the separate **Use \u002F Start \u002F Install Ollama on Windows host** entries handle that case and run their own actions on the Windows side.\nOn macOS, the wizard runs the platform install or upgrade path with `brew upgrade ollama`.\nOn Linux, the wizard runs the official `https:\u002F\u002Follama.com\u002Finstall.sh` path.\nUpgrades on Linux always take the sudo-driven system path because the sudo-free user-local fallback would leave the existing system daemon on `:11434` serving the stale binary.\nIf sudo is not available in a non-interactive run, NemoClaw refuses to silently downgrade the path and asks you to rerun interactively or upgrade Ollama manually.\nAfter an upgrade finishes, NemoClaw re-probes the running daemon's `\u002Fapi\u002Fversion` and fails the run if the daemon still reports below the minimum.\nFresh installs skip this re-probe because the bundled installers ship a daemon at or above the minimum.\nOn WSL, the wizard can use, start, restart, or install Ollama on the Windows host through PowerShell interop.\n\n### Linux Install Modes\n\nOn native Linux, the install path picks between a system install (under `\u002Fusr\u002Flocal`, via the official `https:\u002F\u002Follama.com\u002Finstall.sh`) and a sudo-free user-local install (under `${HOME}\u002F.local`).\nNemoClaw selects the mode automatically:\n\n- Running as root or with passwordless sudo (`sudo -n true` returns 0) selects the system install.\n- A non-interactive run (`NEMOCLAW_NON_INTERACTIVE=1` or no TTY on stdin) without passwordless sudo selects the user-local install.\n  This is the path that lets headless hosts complete onboarding without prompting for a sudo password.\n- An interactive shell without passwordless sudo selects the system install and lets the official installer prompt for the password as usual.\n\nOverride the detection with `NEMOCLAW_OLLAMA_INSTALL_MODE=system` or `NEMOCLAW_OLLAMA_INSTALL_MODE=user`.\n\nThe user-local install replicates only the binary extraction step of the official installer.\nIt downloads the release tarball, extracts it to `${HOME}\u002F.local`, and launches `${HOME}\u002F.local\u002Fbin\u002Follama serve` once.\nIt does not configure a systemd service, does not create the `ollama` system user, and does not install CUDA drivers, so the daemon must be relaunched manually after a reboot.\nNemoClaw also prints a one-line `PATH` hint if `${HOME}\u002F.local\u002Fbin` is not already on your `PATH`; you can add `export PATH=\"${HOME}\u002F.local\u002Fbin:$PATH\"` to your shell profile to invoke `ollama` directly.\n\nBoth modes rely on `zstd` for archive extraction. On Debian and Ubuntu, the system path uses `sudo apt-get` to install `zstd` automatically and explains the prompt before continuing.\nThe user-local path cannot bootstrap system packages without elevation, so if `zstd` is missing it prints per-distro install hints and exits — install `zstd` manually, then rerun onboarding.\n\nRun the onboard wizard.\n\n```console\n$ nemoclaw onboard\n```\n\nSelect **Local Ollama** from the provider list.\nNemoClaw lists installed models or offers starter models if none are installed.\nOn hosts where the larger starter models fit the currently available GPU memory, the starter list includes `qwen3.6:35b` and selects it by default.\nWhen another GPU workload is using most of the memory at onboard time, NemoClaw downgrades the menu to the largest model that still fits.\nIt pulls the selected model, loads it into memory, and validates it before continuing.\nWhen Ollama reports a loaded-model context length, NemoClaw uses that value for the `contextWindow` baked into `openclaw.json` unless you set `NEMOCLAW_CONTEXT_WINDOW` yourself.\nIf the selected model declares that it does not support tool calling, onboarding stops with guidance to choose a model whose `ollama show \u003Cmodel>` capabilities include `tools`.\nThe validation also requires structured chat-completions tool calls.\nIf the model leaks tool-call JSON as plain message text, onboarding stops so you can choose a model that returns tool calls in the expected response field.\nOn WSL, if you choose the Windows-host Ollama path, NemoClaw uses `host.docker.internal:11434` and pulls missing models through the Ollama HTTP API instead of requiring the `ollama` CLI inside WSL.\n\n### WSL with Windows-Host Ollama\n\nWhen NemoClaw runs inside WSL, the provider menu can include Windows-host Ollama actions:\n\n- Use Ollama on Windows host when the Windows daemon is already reachable.\n- Restart Ollama on Windows host when the daemon is installed but only bound to Windows loopback.\n- Start Ollama on Windows host when Ollama is installed but not running.\n- Install Ollama on Windows host when Windows does not have Ollama installed.\n\nThe install and restart paths set `OLLAMA_HOST=0.0.0.0:11434` on the Windows side so Docker and WSL can reach the daemon through `host.docker.internal`.\nAfter an install or restart action, NemoClaw relaunches Ollama from the detected Windows tray app or verified `ollama.exe` path and waits until `host.docker.internal:11434` responds.\n\nIf the HTTP endpoint is not reachable yet, NemoClaw also checks for the Windows `ollama.exe` process through PowerShell interop so it can offer a start or restart action instead of hiding the Windows-host path.\nIf the daemon does not become reachable, onboarding prints PowerShell commands you can run to inspect the Windows-side process and port state. Use one Ollama instance on port `11434` at a time.\nIf both WSL and Windows-host Ollama are running, pick the intended menu entry during onboarding so NemoClaw validates and pulls models against the right daemon.\n\n**Warning:**\n\nOllama is convenient for local chat, but some model\u002Ftemplate combinations can\nreturn tool calls as plain text under realistic agent load. If the TUI shows raw\nJSON such as `{\"name\":\"memory_search\",\"arguments\":{...}}` instead of running a\ntool, switch to vLLM with `--enable-auto-tool-choice` and the correct\n`--tool-call-parser`. See [Tool-Calling Reliability](references\u002Ftool-calling-reliability.md).\n\n### Authenticated Reverse Proxy\n\nOn non-WSL hosts, NemoClaw keeps Ollama bound to `127.0.0.1:11434` and starts a token-gated reverse proxy on `0.0.0.0:11435`.\nThe native install\u002Fstart paths also reset NemoClaw-managed systemd launches to the loopback binding.\nContainers and other hosts on the local network reach Ollama only through the\nproxy, which validates a Bearer token before forwarding requests.\nOn that native path, NemoClaw never exposes Ollama without authentication.\n\nWSL Ollama paths do not use this proxy.\nWindows-host Ollama uses the Windows daemon through `host.docker.internal`.\n\nFor non-WSL Ollama setups, the onboard wizard manages the proxy automatically:\n\n- Generates a random 24-byte token on first run and stores it in\n  `~\u002F.nemoclaw\u002Follama-proxy-token` with `0600` permissions.\n- Starts the proxy after Ollama and verifies it before continuing.\n- Cleans up stale proxy processes from previous runs.\n- Probes the sandbox Docker network path to the proxy before committing the inference route.\n- Stops matching proxy processes during uninstall before deleting NemoClaw state.\n- Reuses the persisted token after a host reboot so you do not need to re-run\n  onboard.\n\nOn native Linux hosts, a firewall can allow the host proxy health check while still blocking sandbox containers on the OpenShell Docker bridge.\nWhen the sandbox-side proxy probe fails with a TCP error, onboarding exits before it saves the inference route and prints a command like:\n\n```console\n$ sudo ufw allow from \u003Copenshell-docker-subnet> to any port 11435 proto tcp\n$ nemoclaw onboard\n```\n\nIf the probe cannot run, for example because Docker Desktop or WSL uses a different host routing model, onboarding continues and relies on the regular proxy health check.\n\nThe sandbox provider is configured to use proxy port `11435` with the generated\ntoken as its `OPENAI_API_KEY` credential.\nOpenShell's L7 proxy injects the token at egress, so the agent inside the\nsandbox never sees the token directly.\n\nAll proxy endpoints require the Bearer token, including `GET \u002Fapi\u002Ftags`.\nInternal health and reachability checks run via the proxy treat any HTTP\nresponse (including `401`) as proof the proxy is alive — they only fail\nwhen nothing answers at all.\n\nIf Ollama is already running on a non-loopback address when you start onboard,\nthe wizard restarts it on `127.0.0.1:11434` so the proxy is the only network\npath to the model server.\n\n### GPU Memory Cleanup\n\nWhen you switch away from Ollama, stop host services, or destroy an Ollama-backed sandbox, NemoClaw asks Ollama to unload currently loaded models from GPU memory.\nThe cleanup sends `keep_alive: 0` for each model reported by Ollama and runs on a best-effort basis, so shutdown continues if Ollama is already stopped.\nThis does not delete downloaded model files.\n\nLoad [references\u002Fuse-local-inference-details.md](references\u002Fuse-local-inference-details.md) for detailed steps on Non-Interactive Setup.\n\n## OpenAI-Compatible Server\n\nThis option works with any server that implements `\u002Fv1\u002Fchat\u002Fcompletions`, including vLLM, TensorRT-LLM, llama.cpp, LocalAI, and others.\nFor compatible endpoints, NemoClaw uses `\u002Fv1\u002Fchat\u002Fcompletions` by default.\nThis avoids a class of failures where local backends accept `\u002Fv1\u002Fresponses` requests but silently drop the system prompt and tool definitions.\nTo opt in to `\u002Fv1\u002Fresponses`, set `NEMOCLAW_PREFERRED_API=openai-responses` before running onboard.\n\nStart your model server.\nThe examples below use vLLM, but any OpenAI-compatible server works.\n\n```console\n$ vllm serve meta-llama\u002FLlama-3.1-8B-Instruct --port 8000\n```\n\nRun the onboard wizard.\n\n```console\n$ nemoclaw onboard\n```\n\nWhen the wizard asks you to choose an inference provider, select **Other OpenAI-compatible endpoint**.\nEnter the base URL of your local server, for example `http:\u002F\u002Flocalhost:8000\u002Fv1`.\n\nThe wizard prompts for an API key.\nIf your server does not require authentication, enter any non-empty string (for example, `dummy`).\n\nNemoClaw validates the endpoint by sending a test inference request before continuing.\nThe wizard probes `\u002Fv1\u002Fchat\u002Fcompletions` by default for the compatible-endpoint provider.\nIf you set `NEMOCLAW_PREFERRED_API=openai-responses`, NemoClaw probes `\u002Fv1\u002Fresponses` instead and only selects it when the response includes the streaming events OpenClaw requires.\nIf a reasoning model returns only reasoning content before producing a final answer, NemoClaw retries the smoke request with a larger response budget.\nRoute, configuration, and authentication failures still fail immediately.\n\nLoad [references\u002Fuse-local-inference-details.md](references\u002Fuse-local-inference-details.md) for detailed steps on Non-Interactive Setup, Selecting the API Path.\n\n## Anthropic-Compatible Server\n\nLoad [references\u002Fuse-local-inference-details.md](references\u002Fuse-local-inference-details.md) for detailed steps.\n\n## vLLM\n\nWhen vLLM is already running on `localhost:8000`, NemoClaw can detect it automatically and query the `\u002Fv1\u002Fmodels` endpoint to determine the loaded model.\nOn supported Linux hosts with NVIDIA GPUs, the onboard wizard can also install or start a managed vLLM container for you.\n\nFor an already-running vLLM server, run `nemoclaw onboard` and select **Local vLLM [experimental]** from the provider list.\n\n```console\n$ nemoclaw onboard\n```\n\nIf vLLM is already running, NemoClaw detects the running model and validates the endpoint.\nIf vLLM is not running and your host matches a DGX Spark or DGX Station managed profile, NemoClaw shows the **Install vLLM** or **Start vLLM** entry by default.\nGeneric Linux NVIDIA GPU hosts still require `NEMOCLAW_EXPERIMENTAL=1` or `NEMOCLAW_PROVIDER=install-vllm` before the managed entry appears.\nNemoClaw pulls the vLLM image, downloads model weights into `~\u002F.cache\u002Fhuggingface`, starts the `nemoclaw-vllm` container on `localhost:8000`, and prints progress markers while the model loads.\nThe first run can take 10 to 30 minutes.\nLater runs reuse the cached image and model weights.\n\nManaged vLLM uses these profiles:\n\n| Host profile | Default model |\n|---|---|\n| DGX Spark | `Qwen\u002FQwen3.6-27B-FP8` |\n| DGX Station | `Qwen\u002FQwen3.6-27B-FP8` |\n| Linux with an NVIDIA GPU | `nvidia\u002FNVIDIA-Nemotron-3-Nano-4B-FP8` |\n\n**Note:**\n\nNemoClaw forces the `chat\u002Fcompletions` API path for vLLM.\nThe vLLM `\u002Fv1\u002Fresponses` endpoint does not run the `--tool-call-parser`, so tool calls arrive as raw text.\n\nLoad [references\u002Fuse-local-inference-details.md](references\u002Fuse-local-inference-details.md) for detailed steps on Non-Interactive Setup, Override the Managed-vLLM Model.\n\n## NVIDIA NIM (Experimental)\n\nNemoClaw can pull, start, and manage a NIM container on hosts with a NIM-capable NVIDIA GPU.\n\nSet the experimental flag and run onboard.\n\n```console\n$ NEMOCLAW_EXPERIMENTAL=1 nemoclaw onboard\n```\n\nSelect **Local NVIDIA NIM [experimental]** from the provider list.\nNemoClaw filters available models by GPU VRAM, pulls the NIM container image, starts it, and waits for it to become healthy before continuing.\nOn hosts with mixed NVIDIA GPU models, the preflight summary shows each detected GPU model and the total VRAM so you can confirm which device class the model selection used.\n\nNIM container images are hosted on `nvcr.io` and require NGC registry authentication before `docker pull` succeeds.\nIf Docker is not already logged in to `nvcr.io`, onboard prompts for an [NGC API key](https:\u002F\u002Forg.ngc.nvidia.com\u002Fsetup\u002Fapi-key) and runs `docker login nvcr.io` over `--password-stdin` so the key is never written to disk or shell history.\nThe prompt masks the key during input and retries once on a bad key before failing.\nIn non-interactive mode, onboard exits with login instructions if Docker is not already authenticated; run `docker login nvcr.io` yourself, then re-run `nemoclaw onboard --non-interactive`.\nIf `NGC_API_KEY` or `NVIDIA_API_KEY` is already exported, NemoClaw passes it into the managed NIM container through the process environment instead of command-line arguments.\nIf the NIM container exits before the health endpoint becomes ready, onboarding stops early and prints the last container log lines.\n\n**Note:**\n\nNIM uses vLLM internally.\nThe same `chat\u002Fcompletions` API path restriction applies.\n\nLoad [references\u002Fuse-local-inference-details.md](references\u002Fuse-local-inference-details.md) for detailed steps on Non-Interactive Setup.\n\n## Timeout Configuration\n\nLoad [references\u002Fuse-local-inference-details.md](references\u002Fuse-local-inference-details.md) for detailed steps.\n\n## Verify the Configuration\n\nLoad [references\u002Fuse-local-inference-details.md](references\u002Fuse-local-inference-details.md) for detailed steps.\n\n## Switch Models at Runtime\n\nLoad [references\u002Fuse-local-inference-details.md](references\u002Fuse-local-inference-details.md) for detailed steps.\n\n## References\n\n- **Load [references\u002Fswitch-inference-providers.md](references\u002Fswitch-inference-providers.md)** when switching inference providers, changing the model runtime, or reconfiguring inference routing. Changes the active inference model without restarting the sandbox.\n- **Load [references\u002Fset-up-sub-agent.md](references\u002Fset-up-sub-agent.md)** when users ask how to add a second model, configure a sub-agent model, use Omni for vision tasks, configure agents.list, or use sessions_spawn in NemoClaw. Shows the NemoClaw-specific file paths and update flow for adding an auxiliary OpenClaw sub-agent model.\n- **[references\u002Ftool-calling-reliability.md](references\u002Ftool-calling-reliability.md)** — Explains Ollama tool-call leak symptoms, when vLLM with a tool-call parser is recommended, and how to repoint NemoClaw to a parser-aware local endpoint.\n- **Load [references\u002Finference-options.md](references\u002Finference-options.md)** when explaining which providers are available, what the onboard wizard presents, or how inference routing works. Lists all inference providers offered during NemoClaw onboarding.\n- **Load [references\u002Fuse-local-inference-details.md](references\u002Fuse-local-inference-details.md)** when you need detailed steps for Non-Interactive Setup, Selecting the API Path, Anthropic-Compatible Server, and related details.\n\n## Related Skills\n\n- [Inference Options](references\u002Finference-options.md) for the full list of providers available during onboarding.\n- [Tool-Calling Reliability](references\u002Ftool-calling-reliability.md) for diagnosing raw JSON tool-call output with local models.\n- [Switch Inference Models](references\u002Fswitch-inference-providers.md) for runtime model switching.\n- `nemoclaw-user-get-started` — Quickstart (use the `nemoclaw-user-get-started` skill) for first-time installation\n","\u003C!-- SPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved. -->\n\u003C!-- SPDX-License-Identifier: Apache-2.0 -->\n\n# Use a Local Inference Server\n\n## Gotchas\n\n- Ollama is convenient for local chat, but some model\u002Ftemplate combinations can return tool calls as plain text under realistic agent load.\n\n## Prerequisites\n\n- NemoClaw installed.\n- A local model server running, or a supported Ollama, vLLM, or NIM setup that the NemoClaw onboard wizard can use, start, or install.\n\nNemoClaw can route inference to a model server running on your machine instead of a cloud API.\nThis page covers Ollama, compatible-endpoint paths for other servers, and experimental managed options for vLLM and NVIDIA NIM.\n\nAll approaches use the same `inference.local` routing model.\nThe agent inside the sandbox never connects to your model server directly.\nOpenShell intercepts inference traffic and forwards it to the local endpoint you configure.\n\n## Ollama\n\nOllama is the default local inference option.\nThe onboard wizard detects Ollama automatically when it is installed or running on the host.\n\nIf Ollama is installed but not running, NemoClaw starts it for you.\nOn macOS and Linux, the wizard can also offer to install Ollama when it is not present.\nWhen the host Ollama is below the minimum version NemoClaw expects for its starter models (currently `0.7.0`), the wizard surfaces an explicit **Upgrade Ollama** entry in the provider menu instead of silently reusing the older daemon, and the express setup path resolves to that entry.\nThe wizard inspects both the CLI binary (`ollama --version`) and the locally running daemon (`\u002Fapi\u002Fversion` on `:11434`) so the upgrade entry still appears when only one side is stale, for example a fresh user-local binary paired with the original system daemon.\nThe gate skips Windows-host Ollama reached from WSL via `host.docker.internal`; the separate **Use \u002F Start \u002F Install Ollama on Windows host** entries handle that case and run their own actions on the Windows side.\nOn macOS, the wizard runs the platform install or upgrade path with `brew upgrade ollama`.\nOn Linux, the wizard runs the official `https:\u002F\u002Follama.com\u002Finstall.sh` path.\nUpgrades on Linux always take the sudo-driven system path because the sudo-free user-local fallback would leave the existing system daemon on `:11434` serving the stale binary.\nIf sudo is not available in a non-interactive run, NemoClaw refuses to silently downgrade the path and asks you to rerun interactively or upgrade Ollama manually.\nAfter an upgrade finishes, NemoClaw re-probes the running daemon's `\u002Fapi\u002Fversion` and fails the run if the daemon still reports below the minimum.\nFresh installs skip this re-probe because the bundled installers ship a daemon at or above the minimum.\nOn WSL, the wizard can use, start, restart, or install Ollama on the Windows host through PowerShell interop.\n\n### Linux Install Modes\n\nOn native Linux, the install path picks between a system install (under `\u002Fusr\u002Flocal`, via the official `https:\u002F\u002Follama.com\u002Finstall.sh`) and a sudo-free user-local install (under `${HOME}\u002F.local`).\nNemoClaw selects the mode automatically:\n\n- Running as root or with passwordless sudo (`sudo -n true` returns 0) selects the system install.\n- A non-interactive run (`NEMOCLAW_NON_INTERACTIVE=1` or no TTY on stdin) without passwordless sudo selects the user-local install.\n  This is the path that lets headless hosts complete onboarding without prompting for a sudo password.\n- An interactive shell without passwordless sudo selects the system install and lets the official installer prompt for the password as usual.\n\nOverride the detection with `NEMOCLAW_OLLAMA_INSTALL_MODE=system` or `NEMOCLAW_OLLAMA_INSTALL_MODE=user`.\n\nThe user-local install replicates only the binary extraction step of the official installer.\nIt downloads the release tarball, extracts it to `${HOME}\u002F.local`, and launches `${HOME}\u002F.local\u002Fbin\u002Follama serve` once.\nIt does not configure a systemd service, does not create the `ollama` system user, and does not install CUDA drivers, so the daemon must be relaunched manually after a reboot.\nNemoClaw also prints a one-line `PATH` hint if `${HOME}\u002F.local\u002Fbin` is not already on your `PATH`; you can add `export PATH=\"${HOME}\u002F.local\u002Fbin:$PATH\"` to your shell profile to invoke `ollama` directly.\n\nBoth modes rely on `zstd` for archive extraction. On Debian and Ubuntu, the system path uses `sudo apt-get` to install `zstd` automatically and explains the prompt before continuing.\nThe user-local path cannot bootstrap system packages without elevation, so if `zstd` is missing it prints per-distro install hints and exits — install `zstd` manually, then rerun onboarding.\n\nRun the onboard wizard.\n\n```console\n$ nemoclaw onboard\n```\n\nSelect **Local Ollama** from the provider list.\nNemoClaw lists installed models or offers starter models if none are installed.\nOn hosts where the larger starter models fit the currently available GPU memory, the starter list includes `qwen3.6:35b` and selects it by default.\nWhen another GPU workload is using most of the memory at onboard time, NemoClaw downgrades the menu to the largest model that still fits.\nIt pulls the selected model, loads it into memory, and validates it before continuing.\nWhen Ollama reports a loaded-model context length, NemoClaw uses that value for the `contextWindow` baked into `openclaw.json` unless you set `NEMOCLAW_CONTEXT_WINDOW` yourself.\nIf the selected model declares that it does not support tool calling, onboarding stops with guidance to choose a model whose `ollama show \u003Cmodel>` capabilities include `tools`.\nThe validation also requires structured chat-completions tool calls.\nIf the model leaks tool-call JSON as plain message text, onboarding stops so you can choose a model that returns tool calls in the expected response field.\nOn WSL, if you choose the Windows-host Ollama path, NemoClaw uses `host.docker.internal:11434` and pulls missing models through the Ollama HTTP API instead of requiring the `ollama` CLI inside WSL.\n\n### WSL with Windows-Host Ollama\n\nWhen NemoClaw runs inside WSL, the provider menu can include Windows-host Ollama actions:\n\n- Use Ollama on Windows host when the Windows daemon is already reachable.\n- Restart Ollama on Windows host when the daemon is installed but only bound to Windows loopback.\n- Start Ollama on Windows host when Ollama is installed but not running.\n- Install Ollama on Windows host when Windows does not have Ollama installed.\n\nThe install and restart paths set `OLLAMA_HOST=0.0.0.0:11434` on the Windows side so Docker and WSL can reach the daemon through `host.docker.internal`.\nAfter an install or restart action, NemoClaw relaunches Ollama from the detected Windows tray app or verified `ollama.exe` path and waits until `host.docker.internal:11434` responds.\n\nIf the HTTP endpoint is not reachable yet, NemoClaw also checks for the Windows `ollama.exe` process through PowerShell interop so it can offer a start or restart action instead of hiding the Windows-host path.\nIf the daemon does not become reachable, onboarding prints PowerShell commands you can run to inspect the Windows-side process and port state. Use one Ollama instance on port `11434` at a time.\nIf both WSL and Windows-host Ollama are running, pick the intended menu entry during onboarding so NemoClaw validates and pulls models against the right daemon.\n\n**Warning:**\n\nOllama is convenient for local chat, but some model\u002Ftemplate combinations can\nreturn tool calls as plain text under realistic agent load. If the TUI shows raw\nJSON such as `{\"name\":\"memory_search\",\"arguments\":{...}}` instead of running a\ntool, switch to vLLM with `--enable-auto-tool-choice` and the correct\n`--tool-call-parser`. See [Tool-Calling Reliability](references\u002Ftool-calling-reliability.md).\n\n### Authenticated Reverse Proxy\n\nOn non-WSL hosts, NemoClaw keeps Ollama bound to `127.0.0.1:11434` and starts a token-gated reverse proxy on `0.0.0.0:11435`.\nThe native install\u002Fstart paths also reset NemoClaw-managed systemd launches to the loopback binding.\nContainers and other hosts on the local network reach Ollama only through the\nproxy, which validates a Bearer token before forwarding requests.\nOn that native path, NemoClaw never exposes Ollama without authentication.\n\nWSL Ollama paths do not use this proxy.\nWindows-host Ollama uses the Windows daemon through `host.docker.internal`.\n\nFor non-WSL Ollama setups, the onboard wizard manages the proxy automatically:\n\n- Generates a random 24-byte token on first run and stores it in\n  `~\u002F.nemoclaw\u002Follama-proxy-token` with `0600` permissions.\n- Starts the proxy after Ollama and verifies it before continuing.\n- Cleans up stale proxy processes from previous runs.\n- Probes the sandbox Docker network path to the proxy before committing the inference route.\n- Stops matching proxy processes during uninstall before deleting NemoClaw state.\n- Reuses the persisted token after a host reboot so you do not need to re-run\n  onboard.\n\nOn native Linux hosts, a firewall can allow the host proxy health check while still blocking sandbox containers on the OpenShell Docker bridge.\nWhen the sandbox-side proxy probe fails with a TCP error, onboarding exits before it saves the inference route and prints a command like:\n\n```console\n$ sudo ufw allow from \u003Copenshell-docker-subnet> to any port 11435 proto tcp\n$ nemoclaw onboard\n```\n\nIf the probe cannot run, for example because Docker Desktop or WSL uses a different host routing model, onboarding continues and relies on the regular proxy health check.\n\nThe sandbox provider is configured to use proxy port `11435` with the generated\ntoken as its `OPENAI_API_KEY` credential.\nOpenShell's L7 proxy injects the token at egress, so the agent inside the\nsandbox never sees the token directly.\n\nAll proxy endpoints require the Bearer token, including `GET \u002Fapi\u002Ftags`.\nInternal health and reachability checks run via the proxy treat any HTTP\nresponse (including `401`) as proof the proxy is alive — they only fail\nwhen nothing answers at all.\n\nIf Ollama is already running on a non-loopback address when you start onboard,\nthe wizard restarts it on `127.0.0.1:11434` so the proxy is the only network\npath to the model server.\n\n### GPU Memory Cleanup\n\nWhen you switch away from Ollama, stop host services, or destroy an Ollama-backed sandbox, NemoClaw asks Ollama to unload currently loaded models from GPU memory.\nThe cleanup sends `keep_alive: 0` for each model reported by Ollama and runs on a best-effort basis, so shutdown continues if Ollama is already stopped.\nThis does not delete downloaded model files.\n\nLoad [references\u002Fuse-local-inference-details.md](references\u002Fuse-local-inference-details.md) for detailed steps on Non-Interactive Setup.\n\n## OpenAI-Compatible Server\n\nThis option works with any server that implements `\u002Fv1\u002Fchat\u002Fcompletions`, including vLLM, TensorRT-LLM, llama.cpp, LocalAI, and others.\nFor compatible endpoints, NemoClaw uses `\u002Fv1\u002Fchat\u002Fcompletions` by default.\nThis avoids a class of failures where local backends accept `\u002Fv1\u002Fresponses` requests but silently drop the system prompt and tool definitions.\nTo opt in to `\u002Fv1\u002Fresponses`, set `NEMOCLAW_PREFERRED_API=openai-responses` before running onboard.\n\nStart your model server.\nThe examples below use vLLM, but any OpenAI-compatible server works.\n\n```console\n$ vllm serve meta-llama\u002FLlama-3.1-8B-Instruct --port 8000\n```\n\nRun the onboard wizard.\n\n```console\n$ nemoclaw onboard\n```\n\nWhen the wizard asks you to choose an inference provider, select **Other OpenAI-compatible endpoint**.\nEnter the base URL of your local server, for example `http:\u002F\u002Flocalhost:8000\u002Fv1`.\n\nThe wizard prompts for an API key.\nIf your server does not require authentication, enter any non-empty string (for example, `dummy`).\n\nNemoClaw validates the endpoint by sending a test inference request before continuing.\nThe wizard probes `\u002Fv1\u002Fchat\u002Fcompletions` by default for the compatible-endpoint provider.\nIf you set `NEMOCLAW_PREFERRED_API=openai-responses`, NemoClaw probes `\u002Fv1\u002Fresponses` instead and only selects it when the response includes the streaming events OpenClaw requires.\nIf a reasoning model returns only reasoning content before producing a final answer, NemoClaw retries the smoke request with a larger response budget.\nRoute, configuration, and authentication failures still fail immediately.\n\nLoad [references\u002Fuse-local-inference-details.md](references\u002Fuse-local-inference-details.md) for detailed steps on Non-Interactive Setup, Selecting the API Path.\n\n## Anthropic-Compatible Server\n\nLoad [references\u002Fuse-local-inference-details.md](references\u002Fuse-local-inference-details.md) for detailed steps.\n\n## vLLM\n\nWhen vLLM is already running on `localhost:8000`, NemoClaw can detect it automatically and query the `\u002Fv1\u002Fmodels` endpoint to determine the loaded model.\nOn supported Linux hosts with NVIDIA GPUs, the onboard wizard can also install or start a managed vLLM container for you.\n\nFor an already-running vLLM server, run `nemoclaw onboard` and select **Local vLLM [experimental]** from the provider list.\n\n```console\n$ nemoclaw onboard\n```\n\nIf vLLM is already running, NemoClaw detects the running model and validates the endpoint.\nIf vLLM is not running and your host matches a DGX Spark or DGX Station managed profile, NemoClaw shows the **Install vLLM** or **Start vLLM** entry by default.\nGeneric Linux NVIDIA GPU hosts still require `NEMOCLAW_EXPERIMENTAL=1` or `NEMOCLAW_PROVIDER=install-vllm` before the managed entry appears.\nNemoClaw pulls the vLLM image, downloads model weights into `~\u002F.cache\u002Fhuggingface`, starts the `nemoclaw-vllm` container on `localhost:8000`, and prints progress markers while the model loads.\nThe first run can take 10 to 30 minutes.\nLater runs reuse the cached image and model weights.\n\nManaged vLLM uses these profiles:\n\n| Host profile | Default model |\n|---|---|\n| DGX Spark | `Qwen\u002FQwen3.6-27B-FP8` |\n| DGX Station | `Qwen\u002FQwen3.6-27B-FP8` |\n| Linux with an NVIDIA GPU | `nvidia\u002FNVIDIA-Nemotron-3-Nano-4B-FP8` |\n\n**Note:**\n\nNemoClaw forces the `chat\u002Fcompletions` API path for vLLM.\nThe vLLM `\u002Fv1\u002Fresponses` endpoint does not run the `--tool-call-parser`, so tool calls arrive as raw text.\n\nLoad [references\u002Fuse-local-inference-details.md](references\u002Fuse-local-inference-details.md) for detailed steps on Non-Interactive Setup, Override the Managed-vLLM Model.\n\n## NVIDIA NIM (Experimental)\n\nNemoClaw can pull, start, and manage a NIM container on hosts with a NIM-capable NVIDIA GPU.\n\nSet the experimental flag and run onboard.\n\n```console\n$ NEMOCLAW_EXPERIMENTAL=1 nemoclaw onboard\n```\n\nSelect **Local NVIDIA NIM [experimental]** from the provider list.\nNemoClaw filters available models by GPU VRAM, pulls the NIM container image, starts it, and waits for it to become healthy before continuing.\nOn hosts with mixed NVIDIA GPU models, the preflight summary shows each detected GPU model and the total VRAM so you can confirm which device class the model selection used.\n\nNIM container images are hosted on `nvcr.io` and require NGC registry authentication before `docker pull` succeeds.\nIf Docker is not already logged in to `nvcr.io`, onboard prompts for an [NGC API key](https:\u002F\u002Forg.ngc.nvidia.com\u002Fsetup\u002Fapi-key) and runs `docker login nvcr.io` over `--password-stdin` so the key is never written to disk or shell history.\nThe prompt masks the key during input and retries once on a bad key before failing.\nIn non-interactive mode, onboard exits with login instructions if Docker is not already authenticated; run `docker login nvcr.io` yourself, then re-run `nemoclaw onboard --non-interactive`.\nIf `NGC_API_KEY` or `NVIDIA_API_KEY` is already exported, NemoClaw passes it into the managed NIM container through the process environment instead of command-line arguments.\nIf the NIM container exits before the health endpoint becomes ready, onboarding stops early and prints the last container log lines.\n\n**Note:**\n\nNIM uses vLLM internally.\nThe same `chat\u002Fcompletions` API path restriction applies.\n\nLoad [references\u002Fuse-local-inference-details.md](references\u002Fuse-local-inference-details.md) for detailed steps on Non-Interactive Setup.\n\n## Timeout Configuration\n\nLoad [references\u002Fuse-local-inference-details.md](references\u002Fuse-local-inference-details.md) for detailed steps.\n\n## Verify the Configuration\n\nLoad [references\u002Fuse-local-inference-details.md](references\u002Fuse-local-inference-details.md) for detailed steps.\n\n## Switch Models at Runtime\n\nLoad [references\u002Fuse-local-inference-details.md](references\u002Fuse-local-inference-details.md) for detailed steps.\n\n## References\n\n- **Load [references\u002Fswitch-inference-providers.md](references\u002Fswitch-inference-providers.md)** when switching inference providers, changing the model runtime, or reconfiguring inference routing. Changes the active inference model without restarting the sandbox.\n- **Load [references\u002Fset-up-sub-agent.md](references\u002Fset-up-sub-agent.md)** when users ask how to add a second model, configure a sub-agent model, use Omni for vision tasks, configure agents.list, or use sessions_spawn in NemoClaw. Shows the NemoClaw-specific file paths and update flow for adding an auxiliary OpenClaw sub-agent model.\n- **[references\u002Ftool-calling-reliability.md](references\u002Ftool-calling-reliability.md)** — Explains Ollama tool-call leak symptoms, when vLLM with a tool-call parser is recommended, and how to repoint NemoClaw to a parser-aware local endpoint.\n- **Load [references\u002Finference-options.md](references\u002Finference-options.md)** when explaining which providers are available, what the onboard wizard presents, or how inference routing works. Lists all inference providers offered during NemoClaw onboarding.\n- **Load [references\u002Fuse-local-inference-details.md](references\u002Fuse-local-inference-details.md)** when you need detailed steps for Non-Interactive Setup, Selecting the API Path, Anthropic-Compatible Server, and related details.\n\n## Related Skills\n\n- [Inference Options](references\u002Finference-options.md) for the full list of providers available during onboarding.\n- [Tool-Calling Reliability](references\u002Ftool-calling-reliability.md) for diagnosing raw JSON tool-call output with local models.\n- [Switch Inference Models](references\u002Fswitch-inference-providers.md) for runtime model switching.\n- `nemoclaw-user-get-started` — Quickstart (use the `nemoclaw-user-get-started` skill) for first-time installation\n",{"name":5,"license":15,"description":7},"main","87afe6ce96220accf50a345dea46784c8c9f210e","YYC3-Skills-技能包\u002FYYC3-Skills-技能包-skills\u002Fskills\u002Fnemoclaw-user-configure-inference",0,"https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base","https:\u002F\u002Fkb.yyc3.top\u002F","https:\u002F\u002Fgithub.com\u002FYYC-Cube","万象归元于云枢 | 深栈智启新纪元","2026-11-10 10:37:32",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add YYC-Cube\u002FYYC3-Knowledge-Base --skill nemoclaw-user-configure-inference",{"tool":39,"command":40},"bunx","bunx skills add YYC-Cube\u002FYYC3-Knowledge-Base --skill nemoclaw-user-configure-inference",{"tool":42,"command":43},"pnpm","pnpm dlx skills add YYC-Cube\u002FYYC3-Knowledge-Base --skill nemoclaw-user-configure-inference","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-Skills-技能包\u002FYYC3-Skills-技能包-skills\u002Fskills\u002Fnemoclaw-user-configure-inference\u002FSKILL.md",true,[47,53,57,61,65,69,73,77,81],{"path":48,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":51,"sourceHtmlUrl":52},"BENCHMARK.md","REFERENCE","参考文档",4034,"https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-Skills-技能包\u002FYYC3-Skills-技能包-skills\u002Fskills\u002Fnemoclaw-user-configure-inference\u002FBENCHMARK.md",{"path":54,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":55,"sourceHtmlUrl":56},"evals\u002Fevals.json",8858,"https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-Skills-技能包\u002FYYC3-Skills-技能包-skills\u002Fskills\u002Fnemoclaw-user-configure-inference\u002Fevals\u002Fevals.json",{"path":58,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":59,"sourceHtmlUrl":60},"references\u002Finference-options.md",12416,"https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-Skills-技能包\u002FYYC3-Skills-技能包-skills\u002Fskills\u002Fnemoclaw-user-configure-inference\u002Freferences\u002Finference-options.md",{"path":62,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":63,"sourceHtmlUrl":64},"references\u002Fset-up-sub-agent.md",6707,"https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-Skills-技能包\u002FYYC3-Skills-技能包-skills\u002Fskills\u002Fnemoclaw-user-configure-inference\u002Freferences\u002Fset-up-sub-agent.md",{"path":66,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":67,"sourceHtmlUrl":68},"references\u002Fswitch-inference-providers.md",8431,"https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-Skills-技能包\u002FYYC3-Skills-技能包-skills\u002Fskills\u002Fnemoclaw-user-configure-inference\u002Freferences\u002Fswitch-inference-providers.md",{"path":70,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":71,"sourceHtmlUrl":72},"references\u002Ftool-calling-reliability.md",5122,"https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-Skills-技能包\u002FYYC3-Skills-技能包-skills\u002Fskills\u002Fnemoclaw-user-configure-inference\u002Freferences\u002Ftool-calling-reliability.md",{"path":74,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":75,"sourceHtmlUrl":76},"references\u002Fuse-local-inference-details.md",6612,"https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-Skills-技能包\u002FYYC3-Skills-技能包-skills\u002Fskills\u002Fnemoclaw-user-configure-inference\u002Freferences\u002Fuse-local-inference-details.md",{"path":78,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":79,"sourceHtmlUrl":80},"skill-card.md",2652,"https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-Skills-技能包\u002FYYC3-Skills-技能包-skills\u002Fskills\u002Fnemoclaw-user-configure-inference\u002Fskill-card.md",{"path":82,"resourceType":83,"resourceTypeDesc":84,"sizeBytes":85,"sourceHtmlUrl":86},"skill.oms.sig","OTHER","其他",5873,"https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-Skills-技能包\u002FYYC3-Skills-技能包-skills\u002Fskills\u002Fnemoclaw-user-configure-inference\u002Fskill.oms.sig",[88,96,103,110,117,124],{"skillNo":89,"name":90,"nameZh":6,"description":91,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":92,"licenseFileUrl":6,"sourceHtmlUrl":93,"categories":94,"keywords":95,"publishedAt":19},"01M7CTDWHERCXPQCP0K03A86AS","vercel-sandbox","Run agent-browser + Chrome inside Vercel Sandbox microVMs for browser automation from any Vercel-deployed app. Use when the user needs browser automation in a Vercel app (Next.js, SvelteKit, Nuxt, Remix, Astro, etc.), wants to run headless Chrome without binary size limits, needs persistent browser sessions across commands, or wants ephemeral isolated browser environments. Triggers include \"Vercel Sandbox browser\", \"microVM Chrome\", \"agent-browser in sandbox\", \"browser automation on Vercel\", or any task requiring Chrome in a Vercel Sandbox.","MIT","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-WebSpec-Web标准\u002FYYC3-WebSpec-Web标准-standards\u002Fbrowser-agent\u002Fskills\u002Fvercel-sandbox\u002FSKILL.md",[],[],{"skillNo":97,"name":98,"nameZh":6,"description":99,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":92,"licenseFileUrl":6,"sourceHtmlUrl":100,"categories":101,"keywords":102,"publishedAt":19},"01M7CTD8D2BNJY54S65M47HEJA","slack","Interact with Slack workspaces using browser automation. Use when the user needs to check unread channels, navigate Slack, send messages, extract data, find information, search conversations, or automate any Slack task. Triggers include \"check my Slack\", \"what channels have unreads\", \"send a message to\", \"search Slack for\", \"extract from Slack\", \"find who said\", or any task requiring programmatic Slack interaction.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-WebSpec-Web标准\u002FYYC3-WebSpec-Web标准-standards\u002Fbrowser-agent\u002Fskills\u002Fslack\u002FSKILL.md",[],[],{"skillNo":104,"name":105,"nameZh":6,"description":106,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":92,"licenseFileUrl":6,"sourceHtmlUrl":107,"categories":108,"keywords":109,"publishedAt":19},"01M7CTCSSB8ZP3Q58C3YZDZK1W","electron","Automate Electron desktop apps (VS Code, Slack, Discord, Figma, Notion, Spotify, etc.) using agent-browser via Chrome DevTools Protocol. Use when the user needs to interact with an Electron app, automate a desktop app, connect to a running app, control a native app, or test an Electron application. Triggers include \"automate Slack app\", \"control VS Code\", \"interact with Discord app\", \"test this Electron app\", \"connect to desktop app\", or any task requiring automation of a native Electron application.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-WebSpec-Web标准\u002FYYC3-WebSpec-Web标准-standards\u002Fbrowser-agent\u002Fskills\u002Felectron\u002FSKILL.md",[],[],{"skillNo":111,"name":112,"nameZh":6,"description":113,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":92,"licenseFileUrl":6,"sourceHtmlUrl":114,"categories":115,"keywords":116,"publishedAt":19},"01M7CTCBJY76688V4DCZGWFHGQ","dogfood","Systematically explore and test a web application to find bugs, UX issues, and other problems. Use when asked to \"dogfood\", \"QA\", \"exploratory test\", \"find issues\", \"bug hunt\", \"test this app\u002Fsite\u002Fplatform\", or review the quality of a web application. Produces a structured report with full reproduction evidence -- step-by-step screenshots, repro videos, and detailed repro steps for every issue -- so findings can be handed directly to the responsible teams.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-WebSpec-Web标准\u002FYYC3-WebSpec-Web标准-standards\u002Fbrowser-agent\u002Fskills\u002Fdogfood\u002FSKILL.md",[],[],{"skillNo":118,"name":119,"nameZh":6,"description":120,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":92,"licenseFileUrl":6,"sourceHtmlUrl":121,"categories":122,"keywords":123,"publishedAt":19},"01M7CTC4H0FFGVEAQR9PA4GA1A","agent-browser","Browser automation CLI for AI agents. Use when the user needs to interact with websites, including navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps, or automating any browser task. Triggers include requests to \"open a website\", \"fill out a form\", \"click a button\", \"take a screenshot\", \"scrape data from a page\", \"test this web app\", \"login to a site\", \"automate browser actions\", or any task requiring programmatic web interaction.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-WebSpec-Web标准\u002FYYC3-WebSpec-Web标准-standards\u002Fbrowser-agent\u002Fskills\u002Fagent-browser\u002FSKILL.md",[],[],{"skillNo":125,"name":126,"nameZh":6,"description":127,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":92,"licenseFileUrl":6,"sourceHtmlUrl":128,"categories":129,"keywords":130,"publishedAt":19},"01M7CTBKVFYBG78FDSABSDJTTQ","shadcn","Manages shadcn components and projects — adding, searching, fixing, debugging, styling, and composing UI, including chat interfaces. Provides project context, component docs, and usage examples. Applies when working with shadcn\u002Fui, component registries, presets, --preset codes, or any project with a components.json file. Also triggers for \"shadcn init\", \"create an app with --preset\", \"switch to --preset\", and for building a design system or applying a DESIGN.md or brand spec to shadcn\u002Fui.","https:\u002F\u002Fgithub.com\u002FYYC-Cube\u002FYYC3-Knowledge-Base\u002Fblob\u002Fmain\u002FYYC3-UI-界面组件\u002FYYC3-UI-界面组件-ui\u002Fskills\u002Fshadcn\u002FSKILL.md",[],[]]