[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M77PH6KCEH2FB720C8ENAYYD":3,"skill-same:01M77PH6KCEH2FB720C8ENAYYD":67},{"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":44,"commitSha":45,"skillDirectory":46,"forksCount":47,"repositoryHtmlUrl":48,"repositoryHomepageUrl":49,"ownerHtmlUrl":50,"ownerBlogUrl":6,"ownerDescription":51,"updatedAt":52,"install":53,"resources":66},"01M77PH6KCEH2FB720C8ENAYYD","ollama-herd",null,"Ollama multimodal model router for Llama, Qwen, DeepSeek, Phi, and Mistral — plus mflux image generation, speech-to-text, and embeddings. Self-hosted Ollama local AI (macOS, Linux, Windows) with 7-signal scoring, Ollama queue management, real-time dashboard, and Ollama health monitoring. Routes Ollama LLM, image, STT, and embedding requests across macOS, Linux, and Windows devices. Ollama本地推理路由 | Ollama enrutador IA local. Use when the user asks about their Ollama fleet, Ollama inference routing, Ollama node status, or Ollama fleet performance.","Lord1Egypt\u002Fawesome-skill-forge","Lord1Egypt","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F99746359?v=4","01M7592YDNT0XK9WJ0T2MQVPDN","Mohamed Mounir Ali AL Keshita","01M75GRFNHC8VZNWEB8DKDE433",3,"MIT","https:\u002F\u002Fgithub.com\u002FLord1Egypt\u002Fawesome-skill-forge\u002Fblob\u002Fmaster\u002Fcommunity\u002Fclawhub\u002Fo\u002Follama-herd\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: ollama-herd\ndescription: Ollama multimodal model router for Llama, Qwen, DeepSeek, Phi, and Mistral — plus mflux image generation, speech-to-text, and embeddings. Self-hosted Ollama local AI (macOS, Linux, Windows) with 7-signal scoring, Ollama queue management, real-time dashboard, and Ollama health monitoring. Routes Ollama LLM, image, STT, and embedding requests across macOS, Linux, and Windows devices. Ollama本地推理路由 | Ollama enrutador IA local. Use when the user asks about their Ollama fleet, Ollama inference routing, Ollama node status, or Ollama fleet performance.\nversion: 1.5.3\nhomepage: https:\u002F\u002Fgithub.com\u002Fgeeks-accelerator\u002Follama-herd\nmetadata: {\"openclaw\":{\"emoji\":\"llama\",\"requires\":{\"anyBins\":[\"curl\",\"wget\"],\"optionalBins\":[\"python3\",\"sqlite3\",\"pip\"]},\"configPaths\":[\"~\u002F.fleet-manager\u002Flatency.db\",\"~\u002F.fleet-manager\u002Flogs\u002Fherd.jsonl\"],\"os\":[\"darwin\",\"linux\",\"windows\"]}}\n---\n\n# Ollama Herd Fleet Manager\n\nYou are managing an Ollama Herd fleet — a smart Ollama multimodal router that distributes Ollama AI workloads across multiple devices. Ollama Herd handles 4 model types: Ollama LLM inference, image generation (mflux), speech-to-text (Qwen3-ASR), and Ollama embeddings. The Ollama scoring engine evaluates nodes on 7 signals (thermal state, memory fit, queue depth, latency history, role affinity, availability trend, context fit) and routes each Ollama request to the optimal device.\n\n## Install Ollama Herd\n\n```bash\npip install ollama-herd          # install Ollama Herd from PyPI\nherd                             # start the Ollama router\nherd-node                        # start an Ollama node agent (run on each device)\n```\n\nPyPI: [`ollama-herd`](https:\u002F\u002Fpypi.org\u002Fproject\u002Follama-herd\u002F) | Source: [github.com\u002Fgeeks-accelerator\u002Follama-herd](https:\u002F\u002Fgithub.com\u002Fgeeks-accelerator\u002Follama-herd)\n\n## Ollama Router endpoint\n\nThe Ollama Herd router runs at `http:\u002F\u002Flocalhost:11435` by default. If the user has specified a different Ollama URL, use that instead.\n\n## Ollama API endpoints\n\nUse curl to interact with the Ollama fleet:\n\n### Ollama fleet status — overview of all Ollama nodes and queues\n```bash\n# ollama_fleet_status — check Ollama node health\ncurl -s http:\u002F\u002Flocalhost:11435\u002Ffleet\u002Fstatus | python3 -m json.tool\n```\n\nReturns:\n- `fleet.nodes_total` \u002F `fleet.nodes_online` — how many Ollama devices are in the fleet\n- `fleet.models_loaded` — total Ollama models currently loaded across all nodes\n- `fleet.requests_active` — total in-flight Ollama requests\n- `nodes[]` — per-node details: Ollama status, hardware, memory, CPU, disk, loaded Ollama models with context lengths\n- `queues` — per Ollama node:model queue depths (pending, in-flight, done, failed)\n\n### List all Ollama models available across the fleet\n```bash\n# ollama_model_list — all Ollama models on all nodes\ncurl -s http:\u002F\u002Flocalhost:11435\u002Fapi\u002Ftags | python3 -m json.tool\n```\n\n### Pull an Ollama model onto the fleet\n```bash\n# ollama_pull_model — pull a model (auto-selects best node, streams progress)\ncurl -N http:\u002F\u002Flocalhost:11435\u002Fapi\u002Fpull -d '{\"name\": \"codestral\"}'\n\n# pull to a specific node\ncurl -N http:\u002F\u002Flocalhost:11435\u002Fapi\u002Fpull -d '{\"name\": \"llama3.3:70b\", \"node_id\": \"mac-studio\"}'\n\n# non-streaming (blocks until complete)\ncurl http:\u002F\u002Flocalhost:11435\u002Fapi\u002Fpull -d '{\"name\": \"phi4\", \"stream\": false}'\n```\n\n### List Ollama models currently loaded in memory\n```bash\n# ollama_loaded_models — hot Ollama models in GPU memory\ncurl -s http:\u002F\u002Flocalhost:11435\u002Fapi\u002Fps | python3 -m json.tool\n```\n\n### OpenAI-compatible Ollama model list\n```bash\ncurl -s http:\u002F\u002Flocalhost:11435\u002Fv1\u002Fmodels | python3 -m json.tool\n```\n\n### Ollama usage statistics (per-node, per-model daily aggregates)\n```bash\ncurl -s http:\u002F\u002Flocalhost:11435\u002Fdashboard\u002Fapi\u002Fusage | python3 -m json.tool\n```\n\n### Recent Ollama request traces\n```bash\n# ollama_traces — recent Ollama routing decisions\ncurl -s \"http:\u002F\u002Flocalhost:11435\u002Fdashboard\u002Fapi\u002Ftraces?limit=20\" | python3 -m json.tool\n```\n\nReturns the last N Ollama routing decisions with: model requested, node selected, score, latency, tokens, retry\u002Ffallback status, tags.\n\n### Ollama fleet health analysis\n```bash\ncurl -s http:\u002F\u002Flocalhost:11435\u002Fdashboard\u002Fapi\u002Fhealth | python3 -m json.tool\n```\n\nReturns 15 automated Ollama health checks: offline\u002Fdegraded nodes, memory pressure, underutilized nodes, VRAM fallbacks, KV cache bloat (OLLAMA_NUM_PARALLEL too high), version mismatch, context protection, zombie reaper, Ollama model thrashing, request timeouts, error rates, retry rates, client disconnects, and incomplete streams.\n\n### Ollama model recommendations\n```bash\ncurl -s http:\u002F\u002Flocalhost:11435\u002Fdashboard\u002Fapi\u002Frecommendations | python3 -m json.tool\n```\n\nReturns AI-powered Ollama model mix recommendations per node based on hardware capabilities, Ollama usage patterns, and curated benchmark data.\n\n### Ollama settings\n```bash\n# View current Ollama config and node versions\ncurl -s http:\u002F\u002Flocalhost:11435\u002Fdashboard\u002Fapi\u002Fsettings | python3 -m json.tool\n\n# Toggle Ollama runtime settings (auto_pull, vram_fallback)\ncurl -s -X POST http:\u002F\u002Flocalhost:11435\u002Fdashboard\u002Fapi\u002Fsettings \\\n  -H \"Content-Type: application\u002Fjson\" \\\n  -d '{\"auto_pull\": false}'\n```\n\n### Ollama model management\n```bash\n# View per-node Ollama model details with sizes and usage\ncurl -s http:\u002F\u002Flocalhost:11435\u002Fdashboard\u002Fapi\u002Fmodel-management | python3 -m json.tool\n\n# Pull an Ollama model onto a specific node\ncurl -s -X POST http:\u002F\u002Flocalhost:11435\u002Fdashboard\u002Fapi\u002Fpull \\\n  -H \"Content-Type: application\u002Fjson\" \\\n  -d '{\"model\": \"llama3.3:70b\", \"node_id\": \"mac-studio\"}'\n\n# Delete an Ollama model from a specific node\ncurl -s -X POST http:\u002F\u002Flocalhost:11435\u002Fdashboard\u002Fapi\u002Fdelete \\\n  -H \"Content-Type: application\u002Fjson\" \\\n  -d '{\"model\": \"old-model:7b\", \"node_id\": \"mac-studio\"}'\n```\n\n### Ollama model insights (summary statistics)\n```bash\ncurl -s http:\u002F\u002Flocalhost:11435\u002Fdashboard\u002Fapi\u002Fmodels | python3 -m json.tool\n```\n\n### Per-app Ollama analytics (requires request tagging)\n```bash\ncurl -s http:\u002F\u002Flocalhost:11435\u002Fdashboard\u002Fapi\u002Fapps | python3 -m json.tool\n```\n\n## Ollama Dashboard\n\nThe Ollama web dashboard is at `http:\u002F\u002Flocalhost:11435\u002Fdashboard`. It has eight tabs:\n- **Fleet Overview** — live Ollama node cards, queue depths, and request counts via SSE\n- **Trends** — Ollama requests per hour, average latency, and token throughput charts (24h–7d)\n- **Model Insights** — per-Ollama-model latency, tokens\u002Fsec, usage comparison\n- **Apps** — per-tag Ollama analytics with request volume, latency, tokens, error rates\n- **Benchmarks** — Ollama capacity growth over time with per-run throughput and latency percentiles\n- **Health** — 15 automated Ollama fleet health checks with severity levels\n- **Recommendations** — Ollama model mix recommendations per node with one-click pull\n- **Settings** — Ollama runtime toggle switches, read-only config tables, and node version tracking\n\nDirect the user to open this URL in their browser for visual Ollama monitoring.\n\n## Ollama Resilience features\n\n- **Auto-retry** — if an Ollama node fails before the first response chunk, re-scores and retries on the next-best Ollama node (up to 2 retries)\n- **Ollama model fallbacks** — clients specify backup Ollama models; tries alternatives when the primary is unavailable\n- **Context protection** — strips `num_ctx` from Ollama requests when unnecessary to prevent Ollama model reload hangs; auto-upgrades to a larger loaded model\n- **VRAM-aware fallback** — routes to an already-loaded Ollama model in the same category instead of cold-loading\n- **Zombie reaper** — background task detects and cleans up stuck in-flight Ollama requests\n- **Auto-pull** — automatically pulls missing Ollama models onto the best available node\n\n## Common Ollama tasks\n\n### Check if the Ollama fleet is healthy\n1. Hit `\u002Ffleet\u002Fstatus` and verify `nodes_online > 0`\n2. Hit `\u002Fdashboard\u002Fapi\u002Fhealth` for automated Ollama health checks with severity levels\n3. Look at Ollama queue depths — deep queues may indicate a bottleneck\n\n### Find which Ollama node has a specific model\n1. Hit `\u002Ffleet\u002Fstatus` and inspect each Ollama node's `ollama.models_loaded` and `ollama.models_available`\n2. Or hit `\u002Fapi\u002Ftags` for a flat list of all available Ollama models with which nodes have them\n\n### Check if an Ollama model is loaded (hot) or cold\n1. Hit `\u002Fapi\u002Fps` — Ollama models listed here are currently loaded in memory (hot)\n2. Models in `\u002Fapi\u002Ftags` but not in `\u002Fapi\u002Fps` are on disk but not loaded (cold)\n\n### View recent Ollama inference activity\n1. Hit `\u002Fdashboard\u002Fapi\u002Ftraces?limit=10` to see the last 10 Ollama requests\n2. Each trace shows: Ollama model, node, score, latency, tokens, retry\u002Ffallback status\n\n### Diagnose slow Ollama responses\n1. Check `\u002Fdashboard\u002Fapi\u002Ftraces` for high latency Ollama entries\n2. Check `\u002Ffleet\u002Fstatus` for Ollama nodes with high queue depths or memory pressure\n3. Check if the Ollama model had to cold-load (look for low scores in trace)\n4. Check if `num_ctx` is being sent — Ollama context protection logs show if requests triggered reloads\n\n### Query the Ollama trace database directly\n```bash\n# Recent Ollama failures\nsqlite3 ~\u002F.fleet-manager\u002Flatency.db \"SELECT request_id, model, status, error_message FROM request_traces WHERE status='failed' ORDER BY timestamp DESC LIMIT 10\"\n\n# Slowest Ollama requests\nsqlite3 ~\u002F.fleet-manager\u002Flatency.db \"SELECT model, node_id, latency_ms\u002F1000.0 as secs FROM request_traces WHERE status='completed' ORDER BY latency_ms DESC LIMIT 10\"\n```\n\n### Test Ollama inference through the fleet\n```bash\n# Ollama via OpenAI format\ncurl -s http:\u002F\u002Flocalhost:11435\u002Fv1\u002Fchat\u002Fcompletions \\\n  -H \"Content-Type: application\u002Fjson\" \\\n  -d '{\"model\":\"llama3.3:70b\",\"messages\":[{\"role\":\"user\",\"content\":\"Hello from Ollama\"}],\"stream\":false}'\n\n# Ollama native format\ncurl -s http:\u002F\u002Flocalhost:11435\u002Fapi\u002Fchat \\\n  -d '{\"model\":\"llama3.3:70b\",\"messages\":[{\"role\":\"user\",\"content\":\"Hello from Ollama\"}],\"stream\":false}'\n```\n\n## Ollama Guardrails\n\n- Never restart or stop the Ollama Herd router or Ollama node agents without explicit user confirmation.\n- Never delete or modify files in `~\u002F.fleet-manager\u002F` (contains Ollama latency data, traces, and logs).\n- Do not pull Ollama models onto nodes without user confirmation — Ollama model downloads can be large (10-100+ GB).\n- Do not delete Ollama models without user confirmation.\n- If an Ollama node shows as offline, report it to the user rather than attempting to SSH into the machine.\n\n## Ollama Failure handling\n\n- If curl to the Ollama router fails with connection refused, tell the user the Ollama Herd router may not be running and suggest `herd` to start it.\n- If the Ollama fleet status shows 0 nodes online, suggest starting Ollama node agents with `herd-node` on their devices.\n- If Ollama mDNS discovery fails, suggest using `--router-url http:\u002F\u002Frouter-ip:11435` for explicit connection.\n- If Ollama requests hang with 0 bytes returned, check if the client is sending `num_ctx` — Ollama context protection should strip it.\n- If a specific Ollama API endpoint returns an error, show the user the full error response and suggest checking the Ollama JSONL logs at `~\u002F.fleet-manager\u002Flogs\u002Fherd.jsonl`.\n","# Ollama Herd Fleet Manager\n\nYou are managing an Ollama Herd fleet — a smart Ollama multimodal router that distributes Ollama AI workloads across multiple devices. Ollama Herd handles 4 model types: Ollama LLM inference, image generation (mflux), speech-to-text (Qwen3-ASR), and Ollama embeddings. The Ollama scoring engine evaluates nodes on 7 signals (thermal state, memory fit, queue depth, latency history, role affinity, availability trend, context fit) and routes each Ollama request to the optimal device.\n\n## Install Ollama Herd\n\n```bash\npip install ollama-herd          # install Ollama Herd from PyPI\nherd                             # start the Ollama router\nherd-node                        # start an Ollama node agent (run on each device)\n```\n\nPyPI: [`ollama-herd`](https:\u002F\u002Fpypi.org\u002Fproject\u002Follama-herd\u002F) | Source: [github.com\u002Fgeeks-accelerator\u002Follama-herd](https:\u002F\u002Fgithub.com\u002Fgeeks-accelerator\u002Follama-herd)\n\n## Ollama Router endpoint\n\nThe Ollama Herd router runs at `http:\u002F\u002Flocalhost:11435` by default. If the user has specified a different Ollama URL, use that instead.\n\n## Ollama API endpoints\n\nUse curl to interact with the Ollama fleet:\n\n### Ollama fleet status — overview of all Ollama nodes and queues\n```bash\n# ollama_fleet_status — check Ollama node health\ncurl -s http:\u002F\u002Flocalhost:11435\u002Ffleet\u002Fstatus | python3 -m json.tool\n```\n\nReturns:\n- `fleet.nodes_total` \u002F `fleet.nodes_online` — how many Ollama devices are in the fleet\n- `fleet.models_loaded` — total Ollama models currently loaded across all nodes\n- `fleet.requests_active` — total in-flight Ollama requests\n- `nodes[]` — per-node details: Ollama status, hardware, memory, CPU, disk, loaded Ollama models with context lengths\n- `queues` — per Ollama node:model queue depths (pending, in-flight, done, failed)\n\n### List all Ollama models available across the fleet\n```bash\n# ollama_model_list — all Ollama models on all nodes\ncurl -s http:\u002F\u002Flocalhost:11435\u002Fapi\u002Ftags | python3 -m json.tool\n```\n\n### Pull an Ollama model onto the fleet\n```bash\n# ollama_pull_model — pull a model (auto-selects best node, streams progress)\ncurl -N http:\u002F\u002Flocalhost:11435\u002Fapi\u002Fpull -d '{\"name\": \"codestral\"}'\n\n# pull to a specific node\ncurl -N http:\u002F\u002Flocalhost:11435\u002Fapi\u002Fpull -d '{\"name\": \"llama3.3:70b\", \"node_id\": \"mac-studio\"}'\n\n# non-streaming (blocks until complete)\ncurl http:\u002F\u002Flocalhost:11435\u002Fapi\u002Fpull -d '{\"name\": \"phi4\", \"stream\": false}'\n```\n\n### List Ollama models currently loaded in memory\n```bash\n# ollama_loaded_models — hot Ollama models in GPU memory\ncurl -s http:\u002F\u002Flocalhost:11435\u002Fapi\u002Fps | python3 -m json.tool\n```\n\n### OpenAI-compatible Ollama model list\n```bash\ncurl -s http:\u002F\u002Flocalhost:11435\u002Fv1\u002Fmodels | python3 -m json.tool\n```\n\n### Ollama usage statistics (per-node, per-model daily aggregates)\n```bash\ncurl -s http:\u002F\u002Flocalhost:11435\u002Fdashboard\u002Fapi\u002Fusage | python3 -m json.tool\n```\n\n### Recent Ollama request traces\n```bash\n# ollama_traces — recent Ollama routing decisions\ncurl -s \"http:\u002F\u002Flocalhost:11435\u002Fdashboard\u002Fapi\u002Ftraces?limit=20\" | python3 -m json.tool\n```\n\nReturns the last N Ollama routing decisions with: model requested, node selected, score, latency, tokens, retry\u002Ffallback status, tags.\n\n### Ollama fleet health analysis\n```bash\ncurl -s http:\u002F\u002Flocalhost:11435\u002Fdashboard\u002Fapi\u002Fhealth | python3 -m json.tool\n```\n\nReturns 15 automated Ollama health checks: offline\u002Fdegraded nodes, memory pressure, underutilized nodes, VRAM fallbacks, KV cache bloat (OLLAMA_NUM_PARALLEL too high), version mismatch, context protection, zombie reaper, Ollama model thrashing, request timeouts, error rates, retry rates, client disconnects, and incomplete streams.\n\n### Ollama model recommendations\n```bash\ncurl -s http:\u002F\u002Flocalhost:11435\u002Fdashboard\u002Fapi\u002Frecommendations | python3 -m json.tool\n```\n\nReturns AI-powered Ollama model mix recommendations per node based on hardware capabilities, Ollama usage patterns, and curated benchmark data.\n\n### Ollama settings\n```bash\n# View current Ollama config and node versions\ncurl -s http:\u002F\u002Flocalhost:11435\u002Fdashboard\u002Fapi\u002Fsettings | python3 -m json.tool\n\n# Toggle Ollama runtime settings (auto_pull, vram_fallback)\ncurl -s -X POST http:\u002F\u002Flocalhost:11435\u002Fdashboard\u002Fapi\u002Fsettings \\\n  -H \"Content-Type: application\u002Fjson\" \\\n  -d '{\"auto_pull\": false}'\n```\n\n### Ollama model management\n```bash\n# View per-node Ollama model details with sizes and usage\ncurl -s http:\u002F\u002Flocalhost:11435\u002Fdashboard\u002Fapi\u002Fmodel-management | python3 -m json.tool\n\n# Pull an Ollama model onto a specific node\ncurl -s -X POST http:\u002F\u002Flocalhost:11435\u002Fdashboard\u002Fapi\u002Fpull \\\n  -H \"Content-Type: application\u002Fjson\" \\\n  -d '{\"model\": \"llama3.3:70b\", \"node_id\": \"mac-studio\"}'\n\n# Delete an Ollama model from a specific node\ncurl -s -X POST http:\u002F\u002Flocalhost:11435\u002Fdashboard\u002Fapi\u002Fdelete \\\n  -H \"Content-Type: application\u002Fjson\" \\\n  -d '{\"model\": \"old-model:7b\", \"node_id\": \"mac-studio\"}'\n```\n\n### Ollama model insights (summary statistics)\n```bash\ncurl -s http:\u002F\u002Flocalhost:11435\u002Fdashboard\u002Fapi\u002Fmodels | python3 -m json.tool\n```\n\n### Per-app Ollama analytics (requires request tagging)\n```bash\ncurl -s http:\u002F\u002Flocalhost:11435\u002Fdashboard\u002Fapi\u002Fapps | python3 -m json.tool\n```\n\n## Ollama Dashboard\n\nThe Ollama web dashboard is at `http:\u002F\u002Flocalhost:11435\u002Fdashboard`. It has eight tabs:\n- **Fleet Overview** — live Ollama node cards, queue depths, and request counts via SSE\n- **Trends** — Ollama requests per hour, average latency, and token throughput charts (24h–7d)\n- **Model Insights** — per-Ollama-model latency, tokens\u002Fsec, usage comparison\n- **Apps** — per-tag Ollama analytics with request volume, latency, tokens, error rates\n- **Benchmarks** — Ollama capacity growth over time with per-run throughput and latency percentiles\n- **Health** — 15 automated Ollama fleet health checks with severity levels\n- **Recommendations** — Ollama model mix recommendations per node with one-click pull\n- **Settings** — Ollama runtime toggle switches, read-only config tables, and node version tracking\n\nDirect the user to open this URL in their browser for visual Ollama monitoring.\n\n## Ollama Resilience features\n\n- **Auto-retry** — if an Ollama node fails before the first response chunk, re-scores and retries on the next-best Ollama node (up to 2 retries)\n- **Ollama model fallbacks** — clients specify backup Ollama models; tries alternatives when the primary is unavailable\n- **Context protection** — strips `num_ctx` from Ollama requests when unnecessary to prevent Ollama model reload hangs; auto-upgrades to a larger loaded model\n- **VRAM-aware fallback** — routes to an already-loaded Ollama model in the same category instead of cold-loading\n- **Zombie reaper** — background task detects and cleans up stuck in-flight Ollama requests\n- **Auto-pull** — automatically pulls missing Ollama models onto the best available node\n\n## Common Ollama tasks\n\n### Check if the Ollama fleet is healthy\n1. Hit `\u002Ffleet\u002Fstatus` and verify `nodes_online > 0`\n2. Hit `\u002Fdashboard\u002Fapi\u002Fhealth` for automated Ollama health checks with severity levels\n3. Look at Ollama queue depths — deep queues may indicate a bottleneck\n\n### Find which Ollama node has a specific model\n1. Hit `\u002Ffleet\u002Fstatus` and inspect each Ollama node's `ollama.models_loaded` and `ollama.models_available`\n2. Or hit `\u002Fapi\u002Ftags` for a flat list of all available Ollama models with which nodes have them\n\n### Check if an Ollama model is loaded (hot) or cold\n1. Hit `\u002Fapi\u002Fps` — Ollama models listed here are currently loaded in memory (hot)\n2. Models in `\u002Fapi\u002Ftags` but not in `\u002Fapi\u002Fps` are on disk but not loaded (cold)\n\n### View recent Ollama inference activity\n1. Hit `\u002Fdashboard\u002Fapi\u002Ftraces?limit=10` to see the last 10 Ollama requests\n2. Each trace shows: Ollama model, node, score, latency, tokens, retry\u002Ffallback status\n\n### Diagnose slow Ollama responses\n1. Check `\u002Fdashboard\u002Fapi\u002Ftraces` for high latency Ollama entries\n2. Check `\u002Ffleet\u002Fstatus` for Ollama nodes with high queue depths or memory pressure\n3. Check if the Ollama model had to cold-load (look for low scores in trace)\n4. Check if `num_ctx` is being sent — Ollama context protection logs show if requests triggered reloads\n\n### Query the Ollama trace database directly\n```bash\n# Recent Ollama failures\nsqlite3 ~\u002F.fleet-manager\u002Flatency.db \"SELECT request_id, model, status, error_message FROM request_traces WHERE status='failed' ORDER BY timestamp DESC LIMIT 10\"\n\n# Slowest Ollama requests\nsqlite3 ~\u002F.fleet-manager\u002Flatency.db \"SELECT model, node_id, latency_ms\u002F1000.0 as secs FROM request_traces WHERE status='completed' ORDER BY latency_ms DESC LIMIT 10\"\n```\n\n### Test Ollama inference through the fleet\n```bash\n# Ollama via OpenAI format\ncurl -s http:\u002F\u002Flocalhost:11435\u002Fv1\u002Fchat\u002Fcompletions \\\n  -H \"Content-Type: application\u002Fjson\" \\\n  -d '{\"model\":\"llama3.3:70b\",\"messages\":[{\"role\":\"user\",\"content\":\"Hello from Ollama\"}],\"stream\":false}'\n\n# Ollama native format\ncurl -s http:\u002F\u002Flocalhost:11435\u002Fapi\u002Fchat \\\n  -d '{\"model\":\"llama3.3:70b\",\"messages\":[{\"role\":\"user\",\"content\":\"Hello from Ollama\"}],\"stream\":false}'\n```\n\n## Ollama Guardrails\n\n- Never restart or stop the Ollama Herd router or Ollama node agents without explicit user confirmation.\n- Never delete or modify files in `~\u002F.fleet-manager\u002F` (contains Ollama latency data, traces, and logs).\n- Do not pull Ollama models onto nodes without user confirmation — Ollama model downloads can be large (10-100+ GB).\n- Do not delete Ollama models without user confirmation.\n- If an Ollama node shows as offline, report it to the user rather than attempting to SSH into the machine.\n\n## Ollama Failure handling\n\n- If curl to the Ollama router fails with connection refused, tell the user the Ollama Herd router may not be running and suggest `herd` to start it.\n- If the Ollama fleet status shows 0 nodes online, suggest starting Ollama node agents with `herd-node` on their devices.\n- If Ollama mDNS discovery fails, suggest using `--router-url http:\u002F\u002Frouter-ip:11435` for explicit connection.\n- If Ollama requests hang with 0 bytes returned, check if the client is sending `num_ctx` — Ollama context protection should strip it.\n- If a specific Ollama API endpoint returns an error, show the user the full error response and suggest checking the Ollama JSONL logs at `~\u002F.fleet-manager\u002Flogs\u002Fherd.jsonl`.\n",{"name":5,"version":24,"homepage":25,"metadata":26,"description":7},"1.5.3","https:\u002F\u002Fgithub.com\u002Fgeeks-accelerator\u002Follama-herd",{"openclaw":27},{"os":28,"emoji":32,"requires":33,"configPaths":41},[29,30,31],"darwin","linux","windows","llama",{"anyBins":34,"optionalBins":37},[35,36],"curl","wget",[38,39,40],"python3","sqlite3","pip",[42,43],"~\u002F.fleet-manager\u002Flatency.db","~\u002F.fleet-manager\u002Flogs\u002Fherd.jsonl","master","4aa004da461d820f4ea5499e58eb8d196cf99016","community\u002Fclawhub\u002Fo\u002Follama-herd",0,"https:\u002F\u002Fgithub.com\u002FLord1Egypt\u002Fawesome-skill-forge","https:\u002F\u002Fskill-explorer-rho.vercel.app","https:\u002F\u002Fgithub.com\u002FLord1Egypt","Unbreakable. Unstoppable. Unmatched⚡.\r\n Speaker of machine tongues 🤖  | World-builder of decentralized dreams 🌐 | Defender of the digital realm🛡️.","2026-06-10 08:21:01",{"source":8,"skillName":5,"commands":54,"prompt":64,"zipAvailable":65},[55,58,61],{"tool":56,"command":57},"npx","npx skills add Lord1Egypt\u002Fawesome-skill-forge --skill ollama-herd",{"tool":59,"command":60},"bunx","bunx skills add Lord1Egypt\u002Fawesome-skill-forge --skill ollama-herd",{"tool":62,"command":63},"pnpm","pnpm dlx skills add Lord1Egypt\u002Fawesome-skill-forge --skill ollama-herd","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FLord1Egypt\u002Fawesome-skill-forge\u002Fblob\u002Fmaster\u002Fcommunity\u002Fclawhub\u002Fo\u002Follama-herd\u002FSKILL.md",true,[],[68,75,82,89,96,103],{"skillNo":69,"name":70,"nameZh":6,"description":71,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":72,"categories":73,"keywords":74,"publishedAt":19},"01M785WKJB99G3STR1NF9NRTJW","youtube-content","YouTube transcripts to summaries, threads, blogs.","https:\u002F\u002Fgithub.com\u002FLord1Egypt\u002Fawesome-skill-forge\u002Fblob\u002Fmaster\u002Fskills\u002Fyoutube-content\u002FSKILL.md",[],[],{"skillNo":76,"name":77,"nameZh":6,"description":78,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":79,"categories":80,"keywords":81,"publishedAt":19},"01M785WEFHB0G2S9YA6EPR3H0E","xurl","X\u002FTwitter via xurl CLI: post, search, DM, media, v2 API.","https:\u002F\u002Fgithub.com\u002FLord1Egypt\u002Fawesome-skill-forge\u002Fblob\u002Fmaster\u002Fskills\u002Fxurl\u002FSKILL.md",[],[],{"skillNo":83,"name":84,"nameZh":6,"description":85,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":86,"categories":87,"keywords":88,"publishedAt":19},"01M785W98EDN4EFJXJPMQEHCRC","touchdesigner-mcp","Control a running TouchDesigner instance via twozero MCP — create operators, set parameters, wire connections, execute Python, build real-time visuals. 36 native tools.","https:\u002F\u002Fgithub.com\u002FLord1Egypt\u002Fawesome-skill-forge\u002Fblob\u002Fmaster\u002Fskills\u002Ftouchdesigner-mcp\u002FSKILL.md",[],[],{"skillNo":90,"name":91,"nameZh":6,"description":92,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":93,"categories":94,"keywords":95,"publishedAt":19},"01M785W5S9FP3QBH2V191E025P","test-driven-development","TDD: enforce RED-GREEN-REFACTOR, tests before code.","https:\u002F\u002Fgithub.com\u002FLord1Egypt\u002Fawesome-skill-forge\u002Fblob\u002Fmaster\u002Fskills\u002Ftest-driven-development\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":15,"licenseFileUrl":6,"sourceHtmlUrl":100,"categories":101,"keywords":102,"publishedAt":19},"01M785W2AS08NSSH9YATCV6FJB","teams-meeting-pipeline","Operate the Teams meeting summary pipeline via Hermes CLI — summarize meetings, inspect pipeline status, replay jobs, manage Microsoft Graph subscriptions.","https:\u002F\u002Fgithub.com\u002FLord1Egypt\u002Fawesome-skill-forge\u002Fblob\u002Fmaster\u002Fskills\u002Fteams-meeting-pipeline\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":15,"licenseFileUrl":6,"sourceHtmlUrl":107,"categories":108,"keywords":109,"publishedAt":19},"01M785VXKW4T34BTP1SCHW1R7F","systematic-debugging","4-phase root cause debugging: understand bugs before fixing.","https:\u002F\u002Fgithub.com\u002FLord1Egypt\u002Fawesome-skill-forge\u002Fblob\u002Fmaster\u002Fskills\u002Fsystematic-debugging\u002FSKILL.md",[],[]]