[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7DH83S2SC2W3762C66QTFFZ":3,"skill-same:01M7DH83S2SC2W3762C66QTFFZ":102},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":17,"publishedAt":18,"linkOnly":19,"markdownContent":20,"markdownBody":21,"frontMatter":22,"refName":23,"commitSha":24,"skillDirectory":25,"forksCount":13,"repositoryHtmlUrl":26,"repositoryHomepageUrl":27,"ownerHtmlUrl":28,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":29,"install":30,"resources":43},"01M7DH83S2SC2W3762C66QTFFZ","huggingface-spaces",null,"Build, deploy, and maintain applications on Hugging Face Spaces — Gradio \u002F Docker \u002F Static SDKs, ZeroGPU and dedicated hardware, model loading, debugging, buckets, inference providers, community grants. Use whenever the user asks to create or host an app on Hugging Face, port code onto ZeroGPU, fix a Space that won't build or run, or otherwise work with `hf spaces …`, `@spaces.GPU`, Space README frontmatter, or the `spaces` Python package.","ak2dev\u002Fagentguard","ak2dev","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F233005497?v=4","01M7DGWK3E3XCC276C6YRFSN05","01M7DGWK4J2A9WQ44CS5CRP4R6",0,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fak2dev\u002Fagentguard\u002Fblob\u002Fmain\u002Fbench\u002Fcorpus\u002Fbenign\u002Freal-hf-huggingface-spaces\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: huggingface-spaces\ndescription: Build, deploy, and maintain applications on Hugging Face Spaces — Gradio \u002F Docker \u002F Static SDKs, ZeroGPU and dedicated hardware, model loading, debugging, buckets, inference providers, community grants. Use whenever the user asks to create or host an app on Hugging Face, port code onto ZeroGPU, fix a Space that won't build or run, or otherwise work with `hf spaces …`, `@spaces.GPU`, Space README frontmatter, or the `spaces` Python package.\n---\n\n# Hugging Face Spaces\n\nHugging Face Spaces host machine-learning applications. There are 1M+ today; each Space is a git repo. This skill covers creating, building, debugging, and maintaining them.\n\n## 0. Getting ready\n\nBefore anything else:\n\n1. Check the `hf` CLI is installed: `which hf`. If not, `pip install -U huggingface_hub`.\n2. Check the user is logged in: `hf auth whoami`. If not, run `hf auth login` — it prints a URL and a one-time code; ask the user to open the URL and enter the code, then login completes automatically (OAuth, no token needed). Alternatively, pass a write-scoped token from https:\u002F\u002Fhuggingface.co\u002Fsettings\u002Ftokens with `--token`.\n3. Note `whoami`'s `canPay` and `isPro` flags — they gate hardware choices below. A free (`isPro=False`) account can only host Static Spaces and up to 2 ZeroGPU Spaces.\n\nThe `hf-cli` skill teaches an agent every `hf` command and is the recommended companion to this one. Install it with `hf skills add hf-cli` (add `--claude --global` to install for Claude Code as well, user-level).\n\n## 1. What a Space is\n\nA Space is a git repo with three possible SDKs:\n\n- **Gradio** — most Spaces. Python, fast iteration, supports ZeroGPU.\n- **Docker** — arbitrary container. Use when you need a non-Python stack or a pre-built template (Streamlit, Argilla, Shiny, etc. — full list at https:\u002F\u002Fhuggingface.co\u002Fdocs\u002Fhub\u002Fspaces-sdks-docker). Does **not** support ZeroGPU.\n- **Static** — plain HTML, or a React\u002FSvelte\u002FVue project built at deploy time. Use for in-browser ML (transformers.js \u002F WebGPU \u002F WebAssembly \u002F onnxruntime-web), project pages, interactive reports, or Spaces that orchestrate other Spaces. No hardware needed.\n\n### Hardware tiers\n\nStatic Spaces are free for everyone and need no hardware. **Gradio and Docker Spaces run on compute and require a paid plan to create** — PRO for personal accounts, Team or Enterprise for organizations — with one exception: **free personal accounts in good standing (verified email, account older than 30 days) can host up to 2 ZeroGPU Spaces.**\n\nSo on a free account ZeroGPU is the *only* way to host a Gradio Space. `cpu-basic` is not the safe fallback it used to be — it is gated too.\n\n**ZeroGPU (`zero-a10g`)** — dynamic, per-request GPU allocation on NVIDIA RTX PRO 6000 Blackwell (sm_120). Two sizes: `large` (half MIG, 48 GB, 1× quota) and `xlarge` (full, 96 GB, 2× quota). Free for the Space creator; Space visitors consume their own daily quota (~5 min free \u002F 40 min Pro \u002F 60 min Enterprise). **Gradio-only**, **PyTorch-first**. Hosting caps per account: **2** free personal, **10** PRO, **50** Team \u002F Enterprise org.\n\n**`cpu-basic`** — 2 vCPU \u002F 16 GB, no hourly cost but needs a paid plan. For data viz, API-proxy Spaces, small CPU-bound models.\n\n**Dedicated GPU** (T4, L4, A10G, L40S, A100, H200) — billed to the Space creator by the hour. List + pricing: `hf spaces hardware`. Only the creator can attach these, and only if `canPay=True`. Use when ZeroGPU genuinely doesn't fit — non-PyTorch main model with heavy init, very-large-model long-context inference, etc.\n\nIf the user needs hardware they can't pay for — a dedicated GPU, or a Gradio Space beyond the free 2-ZeroGPU cap — they can still create a **Static** Space (free for everyone), push the app there, and request a community grant. See [`references\u002Fgrants.md`](references\u002Fgrants.md).\n\nFor the authoritative reference: https:\u002F\u002Fhuggingface.co\u002Fdocs\u002Fhub\u002Fspaces-overview\n\n## 2. Look for an existing demo first\n\nBefore deciding how to build anything, search for prior art:\n\n```bash\nhf spaces search \"\u003Cmodel name or task>\" --sdk gradio --limit 10\n```\n\nIf someone has built a similar Space, read its `app.py` and `requirements.txt` — that gives you the working pattern. Saves a lot of blind iteration. Mention to the user what you found before committing to an approach.\n\n## 3. Decide SDK and hardware\n\nFollow the user's explicit request first. If they were vague:\n\n- **Default for a public ML demo**: Gradio + ZeroGPU. Use this unless something below applies.\n- **The model's only inference path is non-PyTorch** (ONNX \u002F TF \u002F JAX \u002F vLLM as the MAIN model, with heavy init): dedicated GPU.\n  - But: marginal non-torch tools (a small ONNX preprocessor, a TF utility) inside a torch-main pipeline are fine on ZeroGPU. The hijack only patches torch; init the non-torch lib inside `@spaces.GPU` and pay the short per-call init cost.\n- **Tiny \u002F CPU-bound model, or API-proxy Space**: `cpu-basic` — but it needs a paid plan. On a free account, put it on `zero-a10g` with a no-op decorated function (ZeroGPU requires at least one) and keep the real work outside it — nothing ever requests a GPU, so no quota is burned. See [`references\u002Finference-providers.md`](references\u002Finference-providers.md).\n- **Browser-side ML or project page**: Static.\n- **Container with non-Python stack**: Docker.\n\n### Sourcing the model\n\n- **GitHub repo** — clone locally to read structure. If it already has a Gradio demo, the minimal viable path is to adapt it onto ZeroGPU (see [`references\u002Fzerogpu.md`](references\u002Fzerogpu.md)). Otherwise: read the README + inference code, prefer the PyTorch path, estimate VRAM (bf16 ≈ `params_B × 2` GB; 48 GB fits ≤24B params at bf16, or much larger with quantization — see [`references\u002Fzerogpu.md`](references\u002Fzerogpu.md) for quantization on ZeroGPU).\n- **HF model repo** — read its README, follow any linked GitHub.\n- **Paper \u002F blog post** — look for an official or unofficial implementation. Don't reimplement unless trivial or the user explicitly asks.\n- **Vague request** — search Spaces first; surface results.\n\nIf the model genuinely won't fit, check **Inference Providers** as an alternative: see [`references\u002Finference-providers.md`](references\u002Finference-providers.md). This avoids hosting the model at all.\n\n## 4. Create the Space\n\n```bash\nhf repos create \u003Cnamespace>\u002F\u003Cname> --type space --space-sdk \u003Cgradio|docker|static> \\\n    [--flavor zero-a10g|cpu-basic|\u003Cpaid-flavor>] \\\n    [--secrets KEY=val] [--env KEY=val] \\\n    --public|--private|--protected \\\n    --exist-ok\n```\n\n- `--space-sdk` is required.\n- `--flavor` selects hardware. `zero-a10g` is the (legacy) identifier for ZeroGPU. Omitting it gives `cpu-basic` — which is itself gated behind a paid plan, so on a free account pass `--flavor zero-a10g` explicitly. Run `hf spaces hardware` for the full paid list and pricing.\n- Visibility: `--public` (anyone can view), `--private` (only you), `--protected` (app is reachable but git repo \u002F Files tab is private).\n- `--secrets KEY=val` becomes an environment variable inside the Space and is **not** visible to visitors. Use for API keys, gated-repo tokens (`HF_TOKEN=hf_…`), etc. Can also be set later via `hf spaces secrets set \u003Cid> KEY=val`.\n- `--env KEY=val` is **visible to visitors** — use only for non-sensitive config (`GRADIO_SSR_MODE=false`, `PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True`, etc.).\n\n> Note: `hardware:` in the README YAML is silently ignored — hardware is only set via `--flavor` at creation, or later via `hf spaces settings \u003Cid> --hardware \u003Cname>`.\n\n## 5. Build the app\n\nThe Space now exists at `https:\u002F\u002Fhuggingface.co\u002Fspaces\u002F\u003Cnamespace>\u002F\u003Cname>` but is empty.\n\n### README.md frontmatter\n\nAlways required:\n\n```yaml\n---\ntitle: ...\nemoji: 🚀                # pick something representative\ncolorFrom: blue          # red|yellow|green|blue|indigo|purple|pink|gray (only these)\ncolorTo: indigo\nsdk: gradio              # gradio | docker | static\nsdk_version: 6.15.1      # latest stable unless you have a reason*\napp_file: app.py         # gradio only (docker \u002F static use Dockerfile \u002F index.html)\nshort_description: ...   # ≤ 60 chars (server rejects longer)\npython_version: \"3.12\"   # ZeroGPU officially supports 3.10.13 and 3.12.12\nstartup_duration_timeout: 30m   # default; bump to 1h for big LLMs \u002F heavy downloads\n---\n```\n\n\\* Default to the current latest stable, and **look up what that is** (`pip index versions gradio`, or the version a freshly-created Space defaults to) — the number above is a placeholder that goes stale, don't reuse it. Only pin older when the latest genuinely doesn't work for this Space: a custom component pins it, or you're adapting an existing demo and don't want to rewrite for 5.x→6.x breaking changes. If you need a 5.x, pick `5.50.0` (latest of the series; still supports custom components).\n\nAll frontmatter options: https:\u002F\u002Fhuggingface.co\u002Fdocs\u002Fhub\u002Fspaces-config-reference\n\n### Minimal ZeroGPU Gradio app\n\n```python\nimport spaces           # MUST come before torch \u002F diffusers \u002F transformers\nimport torch\nimport gradio as gr\nfrom diffusers import DiffusionPipeline\n\npipe = DiffusionPipeline.from_pretrained(\"\u003Crepo>\", torch_dtype=torch.bfloat16).to(\"cuda\")\n\n@spaces.GPU(duration=60)\ndef generate(prompt: str):\n    \"\"\"Generate an image from a text prompt.\"\"\"   # docstring → API \u002F MCP tool description\n    return pipe(prompt).images[0]\n\ngr.Interface(fn=generate, inputs=gr.Text(), outputs=gr.Image()).launch(mcp_server=True)\n```\n\nThree rules — full treatment in [`references\u002Fzerogpu.md`](references\u002Fzerogpu.md):\n\n1. **`import spaces` before torch \u002F any CUDA-touching import.** It monkey-patches `torch.cuda.*`; once CUDA is initialized in the main process, it's too late.\n2. **Load the model at module scope, `.to(\"cuda\")` eagerly.** ZeroGPU intercepts the call, packs weights to disk, and streams them into VRAM on the first `@spaces.GPU` entry. Lazy loading inside the decorator costs every user.\n3. **Decorate the function Gradio binds.** Estimate `duration` to the realistic worst case (smaller = higher queue priority and tighter quota check). For input-dependent runtime, pass a callable.\n\n### Examples, docstrings, and MCP\n\n- **Add `gr.Examples` whenever it makes sense** (the app takes input and representative inputs exist) — prefer the model\u002Frepo's own official examples. Keep example rows to the few inputs a user actually varies (prompt, image) and give the handler defaults for the rest (steps, seed, guidance) so a row is `[\"a prompt\"]`, not a wall of knobs. Use `cache_examples=True, cache_mode=\"lazy\"`. See [`references\u002Fgradio.md`](references\u002Fgradio.md).\n- **Give every API-triggered function a docstring and type hints.** Each Gradio event handler is exposed over the API; the docstring + signature are what a caller — and the MCP tool schema — sees.\n- **Launch with `demo.launch(mcp_server=True)`** (Gradio 5+) so the Space doubles as an MCP server: each API function becomes an MCP tool described by its docstring and hints.\n\n### requirements.txt\n\nShort version:\n\n- **Do NOT list**: `gradio`, `spaces`, `huggingface_hub` (preinstalled and platform-managed; pinning them causes resolution failures or silently breaks the ZeroGPU runtime).\n- **Do list if you use them**: `torchvision`, `torchaudio` (not preinstalled), plus everything else (`diffusers`, `transformers`, `accelerate`, `sentencepiece`, …).\n- ZeroGPU only accepts torch `2.8.0`, `2.9.1`, `2.10.0`, `2.11.0`. Default to leaving torch unpinned (the runtime preinstalls the latest). Only pin when a dep forces it.\n- For prebuilt CUDA-extension wheels (`flash_attn`, `xformers`, `pytorch3d`, `nvdiffrast`, `diff_gaussian_rasterization`, `torchmcubes`): use the prebuilt Blackwell wheels at `https:\u002F\u002Fhuggingface.co\u002Fdatasets\u002Fmultimodalart\u002Fzerogpu-blackwell-wheels\u002Ftree\u002Fmain\u002Fwheels`. Full mapping + caveats in [`references\u002Frequirements.md`](references\u002Frequirements.md).\n\n### Per-SDK depth\n\n- **Gradio patterns** (themes, `gr.Examples`, streaming, custom HTML components, `gr.Server`): [`references\u002Fgradio.md`](references\u002Fgradio.md).\n- **Docker**: https:\u002F\u002Fhuggingface.co\u002Fdocs\u002Fhub\u002Fspaces-sdks-docker. Examples: `hf spaces list --filter docker`.\n- **Static**: https:\u002F\u002Fhuggingface.co\u002Fdocs\u002Fhub\u002Fspaces-sdks-static. For built SPAs, set `app_build_command: npm run build` and `app_file: dist\u002Findex.html` in frontmatter.\n- **ZeroGPU specifics** (decorator semantics, sizing, AoTI, generators, concurrency, pickle \u002F `gr.State` across the worker boundary): [`references\u002Fzerogpu.md`](references\u002Fzerogpu.md) — read this whenever the Space targets ZeroGPU.\n\n\n## 6. Iterate on the Space, not locally\n\nTry to build a release candidate from the user quest locally and push it — then use the live URL as your test loop. The Space environment is the only one that matters; do not try to test locally. `python3 -m py_compile app.py` is the maximum local check worth doing before pushing.\n\nPush files with `hf upload \u003Cnamespace>\u002F\u003Cname> . --repo-type space`. **`--repo-type space` is required** — `hf upload` defaults to a *model* repo and will otherwise upload to (and silently create) a model repo of the same name. Add `--exclude \"**\u002F__pycache__\u002F**\"` so local bytecode caches aren't committed into the Space.\n\nOnce pushed, pick the cheapest update mechanism for each change — hot-reload for pure Python edits, `hf upload` for code-only files hot-reload can't touch, full rebuild only when `requirements.txt` \u002F `Dockerfile` \u002F README frontmatter actually changed. Full ladder + footguns (hot-reload poisoning factory reboot, runtime.sha lag, etc.) in [`references\u002Fdebugging.md`](references\u002Fdebugging.md).\n\n## 7. Verify\n\nDon't trust `RUNNING` alone — the app can be running but broken. Four steps, in order:\n\n**A. Alive?** Stage + hardware:\n```bash\nhf spaces info \u003Cns>\u002F\u003Cname> --expand runtime\n```\n\n**B. Logs clean post-boot?** Read the run log to confirm startup finished without warnings or silent fallbacks:\n```bash\nhf spaces logs \u003Cns>\u002F\u003Cname> --tail 200\n```\nLook for model-load completion, no import warnings, no \"falling back to CPU\" \u002F dtype downgrade messages, no `RUNNING` masking a half-broken app.\n\n**C. API actually responds.** With logs still tailing in another terminal (`hf spaces logs \u003Cns>\u002F\u003Cname> --follow`), call the endpoint:\n```python\nfrom gradio_client import Client, handle_file\nimport os\nc = Client(\"\u003Cns>\u002F\u003Cname>\", token=os.environ[\"HF_TOKEN\"], httpx_kwargs={\"timeout\": 600})\nprint(c.view_api())                    # discover endpoints — don't guess\nresult = c.predict(..., api_name=\"\u002Fgenerate\")\n```\n\n**D. Sniff output AND logs.** HTTP 200 ≠ correct output. Check both:\n```python\nhead = open(result, \"rb\").read(16)\n# glTF \u002F \\x89PNG \u002F RIFF…WEBP \u002F RIFF…WAVE \u002F [4:8]==b\"ftyp\" → png\u002Fjpg\u002Fwebp\u002Fwav\u002Fmp4\n```\nAnd look at the run log emitted during the call — silent fallbacks (model snapping to a different size, missing optional dep, dtype downgrade) only show up there.\n\nFull smoke-test patterns (streaming endpoints, OAuth-gated Spaces, `gr.Server` custom routes): [`references\u002Fdebugging.md`](references\u002Fdebugging.md).\n\n## 8. Permanent storage (buckets)\n\nSpaces are stateless — `\u002Fdata` is wiped on restart. If the Space needs to persist user uploads, generations, logs, or interact with a long-lived store, mount a **bucket**:\n\n```bash\nhf buckets create \u003Cns>\u002F\u003Cbucket-name>                                          # --private optional\nhf spaces volumes set \u003Cns>\u002F\u003Cspace> -v hf:\u002F\u002Fbuckets\u002F\u003Cns>\u002F\u003Cbucket-name>:\u002Fdata   # read-write at \u002Fdata\n```\n\nBuckets are paid storage; check `canPay` and confirm with the user. Full patterns (read-fast \u002F write-durable, public bucket URLs, model-cache anti-pattern): [`references\u002Fbuckets.md`](references\u002Fbuckets.md).\n\n## 9. When things break\n\nOrder of operations:\n\n1. Read the logs: `hf spaces logs \u003Cid> --build --follow` (build error) or `hf spaces logs \u003Cid> --follow` (runtime error). Find the **first** error, not the last.\n2. Grep [`references\u002Fknown-errors.md`](references\u002Fknown-errors.md) for the error string. Check if this is a known issue before trying your own fix — most common ZeroGPU \u002F Gradio \u002F dependency errors have a 1–2 line fix there.\n3. Iterate using the cheapest rung from [`references\u002Fdebugging.md`](references\u002Fdebugging.md). The vast majority of issues resolve with log-reading + smoke-test loops; interactive dev mode + SSH is a heavy-hammer last resort.\n\nIf you solve an error that wasn't in the known-errors list, suggest the user PR it back to this skill so future runs benefit.\n\n---\n\n## Reference index\n\n| When to read | File |\n|---|---|\n| **How ZeroGPU works** + correct patterns (decorator, sizing, pickle, generators, real-time, AoTI) | [`references\u002Fzerogpu.md`](references\u002Fzerogpu.md) |\n| **Iterate + debug**: logs, rung ladder, smoke testing (and dev mode + SSH as a last resort) | [`references\u002Fdebugging.md`](references\u002Fdebugging.md) |\n| **Error-string lookup** — the single place for all error symptoms (Spaces, ZeroGPU, Gradio, deps) | [`references\u002Fknown-errors.md`](references\u002Fknown-errors.md) |\n| Pinning deps, picking wheels, torch-family alignment | [`references\u002Frequirements.md`](references\u002Frequirements.md) |\n| `gr.Examples` (add when it makes sense), themes, custom HTML components, `gr.Server`, MCP server (`mcp_server=True`) | [`references\u002Fgradio.md`](references\u002Fgradio.md) |\n| Persistent storage, public bucket URLs | [`references\u002Fbuckets.md`](references\u002Fbuckets.md) |\n| Community grant requests (hardware the user can't pay for) | [`references\u002Fgrants.md`](references\u002Fgrants.md) |\n| Provider proxy (zero-VRAM big LLM via Cerebras \u002F Fireworks \u002F Together \u002F etc.) | [`references\u002Finference-providers.md`](references\u002Finference-providers.md) |\n| **3D Spaces: generation, CUDA extensions, output formats, and model recipes (incl. gaussian splatting)** | [`references\u002F3d-generation.md`](references\u002F3d-generation.md) |\n","# Hugging Face Spaces\n\nHugging Face Spaces host machine-learning applications. There are 1M+ today; each Space is a git repo. This skill covers creating, building, debugging, and maintaining them.\n\n## 0. Getting ready\n\nBefore anything else:\n\n1. Check the `hf` CLI is installed: `which hf`. If not, `pip install -U huggingface_hub`.\n2. Check the user is logged in: `hf auth whoami`. If not, run `hf auth login` — it prints a URL and a one-time code; ask the user to open the URL and enter the code, then login completes automatically (OAuth, no token needed). Alternatively, pass a write-scoped token from https:\u002F\u002Fhuggingface.co\u002Fsettings\u002Ftokens with `--token`.\n3. Note `whoami`'s `canPay` and `isPro` flags — they gate hardware choices below. A free (`isPro=False`) account can only host Static Spaces and up to 2 ZeroGPU Spaces.\n\nThe `hf-cli` skill teaches an agent every `hf` command and is the recommended companion to this one. Install it with `hf skills add hf-cli` (add `--claude --global` to install for Claude Code as well, user-level).\n\n## 1. What a Space is\n\nA Space is a git repo with three possible SDKs:\n\n- **Gradio** — most Spaces. Python, fast iteration, supports ZeroGPU.\n- **Docker** — arbitrary container. Use when you need a non-Python stack or a pre-built template (Streamlit, Argilla, Shiny, etc. — full list at https:\u002F\u002Fhuggingface.co\u002Fdocs\u002Fhub\u002Fspaces-sdks-docker). Does **not** support ZeroGPU.\n- **Static** — plain HTML, or a React\u002FSvelte\u002FVue project built at deploy time. Use for in-browser ML (transformers.js \u002F WebGPU \u002F WebAssembly \u002F onnxruntime-web), project pages, interactive reports, or Spaces that orchestrate other Spaces. No hardware needed.\n\n### Hardware tiers\n\nStatic Spaces are free for everyone and need no hardware. **Gradio and Docker Spaces run on compute and require a paid plan to create** — PRO for personal accounts, Team or Enterprise for organizations — with one exception: **free personal accounts in good standing (verified email, account older than 30 days) can host up to 2 ZeroGPU Spaces.**\n\nSo on a free account ZeroGPU is the *only* way to host a Gradio Space. `cpu-basic` is not the safe fallback it used to be — it is gated too.\n\n**ZeroGPU (`zero-a10g`)** — dynamic, per-request GPU allocation on NVIDIA RTX PRO 6000 Blackwell (sm_120). Two sizes: `large` (half MIG, 48 GB, 1× quota) and `xlarge` (full, 96 GB, 2× quota). Free for the Space creator; Space visitors consume their own daily quota (~5 min free \u002F 40 min Pro \u002F 60 min Enterprise). **Gradio-only**, **PyTorch-first**. Hosting caps per account: **2** free personal, **10** PRO, **50** Team \u002F Enterprise org.\n\n**`cpu-basic`** — 2 vCPU \u002F 16 GB, no hourly cost but needs a paid plan. For data viz, API-proxy Spaces, small CPU-bound models.\n\n**Dedicated GPU** (T4, L4, A10G, L40S, A100, H200) — billed to the Space creator by the hour. List + pricing: `hf spaces hardware`. Only the creator can attach these, and only if `canPay=True`. Use when ZeroGPU genuinely doesn't fit — non-PyTorch main model with heavy init, very-large-model long-context inference, etc.\n\nIf the user needs hardware they can't pay for — a dedicated GPU, or a Gradio Space beyond the free 2-ZeroGPU cap — they can still create a **Static** Space (free for everyone), push the app there, and request a community grant. See [`references\u002Fgrants.md`](references\u002Fgrants.md).\n\nFor the authoritative reference: https:\u002F\u002Fhuggingface.co\u002Fdocs\u002Fhub\u002Fspaces-overview\n\n## 2. Look for an existing demo first\n\nBefore deciding how to build anything, search for prior art:\n\n```bash\nhf spaces search \"\u003Cmodel name or task>\" --sdk gradio --limit 10\n```\n\nIf someone has built a similar Space, read its `app.py` and `requirements.txt` — that gives you the working pattern. Saves a lot of blind iteration. Mention to the user what you found before committing to an approach.\n\n## 3. Decide SDK and hardware\n\nFollow the user's explicit request first. If they were vague:\n\n- **Default for a public ML demo**: Gradio + ZeroGPU. Use this unless something below applies.\n- **The model's only inference path is non-PyTorch** (ONNX \u002F TF \u002F JAX \u002F vLLM as the MAIN model, with heavy init): dedicated GPU.\n  - But: marginal non-torch tools (a small ONNX preprocessor, a TF utility) inside a torch-main pipeline are fine on ZeroGPU. The hijack only patches torch; init the non-torch lib inside `@spaces.GPU` and pay the short per-call init cost.\n- **Tiny \u002F CPU-bound model, or API-proxy Space**: `cpu-basic` — but it needs a paid plan. On a free account, put it on `zero-a10g` with a no-op decorated function (ZeroGPU requires at least one) and keep the real work outside it — nothing ever requests a GPU, so no quota is burned. See [`references\u002Finference-providers.md`](references\u002Finference-providers.md).\n- **Browser-side ML or project page**: Static.\n- **Container with non-Python stack**: Docker.\n\n### Sourcing the model\n\n- **GitHub repo** — clone locally to read structure. If it already has a Gradio demo, the minimal viable path is to adapt it onto ZeroGPU (see [`references\u002Fzerogpu.md`](references\u002Fzerogpu.md)). Otherwise: read the README + inference code, prefer the PyTorch path, estimate VRAM (bf16 ≈ `params_B × 2` GB; 48 GB fits ≤24B params at bf16, or much larger with quantization — see [`references\u002Fzerogpu.md`](references\u002Fzerogpu.md) for quantization on ZeroGPU).\n- **HF model repo** — read its README, follow any linked GitHub.\n- **Paper \u002F blog post** — look for an official or unofficial implementation. Don't reimplement unless trivial or the user explicitly asks.\n- **Vague request** — search Spaces first; surface results.\n\nIf the model genuinely won't fit, check **Inference Providers** as an alternative: see [`references\u002Finference-providers.md`](references\u002Finference-providers.md). This avoids hosting the model at all.\n\n## 4. Create the Space\n\n```bash\nhf repos create \u003Cnamespace>\u002F\u003Cname> --type space --space-sdk \u003Cgradio|docker|static> \\\n    [--flavor zero-a10g|cpu-basic|\u003Cpaid-flavor>] \\\n    [--secrets KEY=val] [--env KEY=val] \\\n    --public|--private|--protected \\\n    --exist-ok\n```\n\n- `--space-sdk` is required.\n- `--flavor` selects hardware. `zero-a10g` is the (legacy) identifier for ZeroGPU. Omitting it gives `cpu-basic` — which is itself gated behind a paid plan, so on a free account pass `--flavor zero-a10g` explicitly. Run `hf spaces hardware` for the full paid list and pricing.\n- Visibility: `--public` (anyone can view), `--private` (only you), `--protected` (app is reachable but git repo \u002F Files tab is private).\n- `--secrets KEY=val` becomes an environment variable inside the Space and is **not** visible to visitors. Use for API keys, gated-repo tokens (`HF_TOKEN=hf_…`), etc. Can also be set later via `hf spaces secrets set \u003Cid> KEY=val`.\n- `--env KEY=val` is **visible to visitors** — use only for non-sensitive config (`GRADIO_SSR_MODE=false`, `PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True`, etc.).\n\n> Note: `hardware:` in the README YAML is silently ignored — hardware is only set via `--flavor` at creation, or later via `hf spaces settings \u003Cid> --hardware \u003Cname>`.\n\n## 5. Build the app\n\nThe Space now exists at `https:\u002F\u002Fhuggingface.co\u002Fspaces\u002F\u003Cnamespace>\u002F\u003Cname>` but is empty.\n\n### README.md frontmatter\n\nAlways required:\n\n```yaml\n---\ntitle: ...\nemoji: 🚀                # pick something representative\ncolorFrom: blue          # red|yellow|green|blue|indigo|purple|pink|gray (only these)\ncolorTo: indigo\nsdk: gradio              # gradio | docker | static\nsdk_version: 6.15.1      # latest stable unless you have a reason*\napp_file: app.py         # gradio only (docker \u002F static use Dockerfile \u002F index.html)\nshort_description: ...   # ≤ 60 chars (server rejects longer)\npython_version: \"3.12\"   # ZeroGPU officially supports 3.10.13 and 3.12.12\nstartup_duration_timeout: 30m   # default; bump to 1h for big LLMs \u002F heavy downloads\n---\n```\n\n\\* Default to the current latest stable, and **look up what that is** (`pip index versions gradio`, or the version a freshly-created Space defaults to) — the number above is a placeholder that goes stale, don't reuse it. Only pin older when the latest genuinely doesn't work for this Space: a custom component pins it, or you're adapting an existing demo and don't want to rewrite for 5.x→6.x breaking changes. If you need a 5.x, pick `5.50.0` (latest of the series; still supports custom components).\n\nAll frontmatter options: https:\u002F\u002Fhuggingface.co\u002Fdocs\u002Fhub\u002Fspaces-config-reference\n\n### Minimal ZeroGPU Gradio app\n\n```python\nimport spaces           # MUST come before torch \u002F diffusers \u002F transformers\nimport torch\nimport gradio as gr\nfrom diffusers import DiffusionPipeline\n\npipe = DiffusionPipeline.from_pretrained(\"\u003Crepo>\", torch_dtype=torch.bfloat16).to(\"cuda\")\n\n@spaces.GPU(duration=60)\ndef generate(prompt: str):\n    \"\"\"Generate an image from a text prompt.\"\"\"   # docstring → API \u002F MCP tool description\n    return pipe(prompt).images[0]\n\ngr.Interface(fn=generate, inputs=gr.Text(), outputs=gr.Image()).launch(mcp_server=True)\n```\n\nThree rules — full treatment in [`references\u002Fzerogpu.md`](references\u002Fzerogpu.md):\n\n1. **`import spaces` before torch \u002F any CUDA-touching import.** It monkey-patches `torch.cuda.*`; once CUDA is initialized in the main process, it's too late.\n2. **Load the model at module scope, `.to(\"cuda\")` eagerly.** ZeroGPU intercepts the call, packs weights to disk, and streams them into VRAM on the first `@spaces.GPU` entry. Lazy loading inside the decorator costs every user.\n3. **Decorate the function Gradio binds.** Estimate `duration` to the realistic worst case (smaller = higher queue priority and tighter quota check). For input-dependent runtime, pass a callable.\n\n### Examples, docstrings, and MCP\n\n- **Add `gr.Examples` whenever it makes sense** (the app takes input and representative inputs exist) — prefer the model\u002Frepo's own official examples. Keep example rows to the few inputs a user actually varies (prompt, image) and give the handler defaults for the rest (steps, seed, guidance) so a row is `[\"a prompt\"]`, not a wall of knobs. Use `cache_examples=True, cache_mode=\"lazy\"`. See [`references\u002Fgradio.md`](references\u002Fgradio.md).\n- **Give every API-triggered function a docstring and type hints.** Each Gradio event handler is exposed over the API; the docstring + signature are what a caller — and the MCP tool schema — sees.\n- **Launch with `demo.launch(mcp_server=True)`** (Gradio 5+) so the Space doubles as an MCP server: each API function becomes an MCP tool described by its docstring and hints.\n\n### requirements.txt\n\nShort version:\n\n- **Do NOT list**: `gradio`, `spaces`, `huggingface_hub` (preinstalled and platform-managed; pinning them causes resolution failures or silently breaks the ZeroGPU runtime).\n- **Do list if you use them**: `torchvision`, `torchaudio` (not preinstalled), plus everything else (`diffusers`, `transformers`, `accelerate`, `sentencepiece`, …).\n- ZeroGPU only accepts torch `2.8.0`, `2.9.1`, `2.10.0`, `2.11.0`. Default to leaving torch unpinned (the runtime preinstalls the latest). Only pin when a dep forces it.\n- For prebuilt CUDA-extension wheels (`flash_attn`, `xformers`, `pytorch3d`, `nvdiffrast`, `diff_gaussian_rasterization`, `torchmcubes`): use the prebuilt Blackwell wheels at `https:\u002F\u002Fhuggingface.co\u002Fdatasets\u002Fmultimodalart\u002Fzerogpu-blackwell-wheels\u002Ftree\u002Fmain\u002Fwheels`. Full mapping + caveats in [`references\u002Frequirements.md`](references\u002Frequirements.md).\n\n### Per-SDK depth\n\n- **Gradio patterns** (themes, `gr.Examples`, streaming, custom HTML components, `gr.Server`): [`references\u002Fgradio.md`](references\u002Fgradio.md).\n- **Docker**: https:\u002F\u002Fhuggingface.co\u002Fdocs\u002Fhub\u002Fspaces-sdks-docker. Examples: `hf spaces list --filter docker`.\n- **Static**: https:\u002F\u002Fhuggingface.co\u002Fdocs\u002Fhub\u002Fspaces-sdks-static. For built SPAs, set `app_build_command: npm run build` and `app_file: dist\u002Findex.html` in frontmatter.\n- **ZeroGPU specifics** (decorator semantics, sizing, AoTI, generators, concurrency, pickle \u002F `gr.State` across the worker boundary): [`references\u002Fzerogpu.md`](references\u002Fzerogpu.md) — read this whenever the Space targets ZeroGPU.\n\n\n## 6. Iterate on the Space, not locally\n\nTry to build a release candidate from the user quest locally and push it — then use the live URL as your test loop. The Space environment is the only one that matters; do not try to test locally. `python3 -m py_compile app.py` is the maximum local check worth doing before pushing.\n\nPush files with `hf upload \u003Cnamespace>\u002F\u003Cname> . --repo-type space`. **`--repo-type space` is required** — `hf upload` defaults to a *model* repo and will otherwise upload to (and silently create) a model repo of the same name. Add `--exclude \"**\u002F__pycache__\u002F**\"` so local bytecode caches aren't committed into the Space.\n\nOnce pushed, pick the cheapest update mechanism for each change — hot-reload for pure Python edits, `hf upload` for code-only files hot-reload can't touch, full rebuild only when `requirements.txt` \u002F `Dockerfile` \u002F README frontmatter actually changed. Full ladder + footguns (hot-reload poisoning factory reboot, runtime.sha lag, etc.) in [`references\u002Fdebugging.md`](references\u002Fdebugging.md).\n\n## 7. Verify\n\nDon't trust `RUNNING` alone — the app can be running but broken. Four steps, in order:\n\n**A. Alive?** Stage + hardware:\n```bash\nhf spaces info \u003Cns>\u002F\u003Cname> --expand runtime\n```\n\n**B. Logs clean post-boot?** Read the run log to confirm startup finished without warnings or silent fallbacks:\n```bash\nhf spaces logs \u003Cns>\u002F\u003Cname> --tail 200\n```\nLook for model-load completion, no import warnings, no \"falling back to CPU\" \u002F dtype downgrade messages, no `RUNNING` masking a half-broken app.\n\n**C. API actually responds.** With logs still tailing in another terminal (`hf spaces logs \u003Cns>\u002F\u003Cname> --follow`), call the endpoint:\n```python\nfrom gradio_client import Client, handle_file\nimport os\nc = Client(\"\u003Cns>\u002F\u003Cname>\", token=os.environ[\"HF_TOKEN\"], httpx_kwargs={\"timeout\": 600})\nprint(c.view_api())                    # discover endpoints — don't guess\nresult = c.predict(..., api_name=\"\u002Fgenerate\")\n```\n\n**D. Sniff output AND logs.** HTTP 200 ≠ correct output. Check both:\n```python\nhead = open(result, \"rb\").read(16)\n# glTF \u002F \\x89PNG \u002F RIFF…WEBP \u002F RIFF…WAVE \u002F [4:8]==b\"ftyp\" → png\u002Fjpg\u002Fwebp\u002Fwav\u002Fmp4\n```\nAnd look at the run log emitted during the call — silent fallbacks (model snapping to a different size, missing optional dep, dtype downgrade) only show up there.\n\nFull smoke-test patterns (streaming endpoints, OAuth-gated Spaces, `gr.Server` custom routes): [`references\u002Fdebugging.md`](references\u002Fdebugging.md).\n\n## 8. Permanent storage (buckets)\n\nSpaces are stateless — `\u002Fdata` is wiped on restart. If the Space needs to persist user uploads, generations, logs, or interact with a long-lived store, mount a **bucket**:\n\n```bash\nhf buckets create \u003Cns>\u002F\u003Cbucket-name>                                          # --private optional\nhf spaces volumes set \u003Cns>\u002F\u003Cspace> -v hf:\u002F\u002Fbuckets\u002F\u003Cns>\u002F\u003Cbucket-name>:\u002Fdata   # read-write at \u002Fdata\n```\n\nBuckets are paid storage; check `canPay` and confirm with the user. Full patterns (read-fast \u002F write-durable, public bucket URLs, model-cache anti-pattern): [`references\u002Fbuckets.md`](references\u002Fbuckets.md).\n\n## 9. When things break\n\nOrder of operations:\n\n1. Read the logs: `hf spaces logs \u003Cid> --build --follow` (build error) or `hf spaces logs \u003Cid> --follow` (runtime error). Find the **first** error, not the last.\n2. Grep [`references\u002Fknown-errors.md`](references\u002Fknown-errors.md) for the error string. Check if this is a known issue before trying your own fix — most common ZeroGPU \u002F Gradio \u002F dependency errors have a 1–2 line fix there.\n3. Iterate using the cheapest rung from [`references\u002Fdebugging.md`](references\u002Fdebugging.md). The vast majority of issues resolve with log-reading + smoke-test loops; interactive dev mode + SSH is a heavy-hammer last resort.\n\nIf you solve an error that wasn't in the known-errors list, suggest the user PR it back to this skill so future runs benefit.\n\n---\n\n## Reference index\n\n| When to read | File |\n|---|---|\n| **How ZeroGPU works** + correct patterns (decorator, sizing, pickle, generators, real-time, AoTI) | [`references\u002Fzerogpu.md`](references\u002Fzerogpu.md) |\n| **Iterate + debug**: logs, rung ladder, smoke testing (and dev mode + SSH as a last resort) | [`references\u002Fdebugging.md`](references\u002Fdebugging.md) |\n| **Error-string lookup** — the single place for all error symptoms (Spaces, ZeroGPU, Gradio, deps) | [`references\u002Fknown-errors.md`](references\u002Fknown-errors.md) |\n| Pinning deps, picking wheels, torch-family alignment | [`references\u002Frequirements.md`](references\u002Frequirements.md) |\n| `gr.Examples` (add when it makes sense), themes, custom HTML components, `gr.Server`, MCP server (`mcp_server=True`) | [`references\u002Fgradio.md`](references\u002Fgradio.md) |\n| Persistent storage, public bucket URLs | [`references\u002Fbuckets.md`](references\u002Fbuckets.md) |\n| Community grant requests (hardware the user can't pay for) | [`references\u002Fgrants.md`](references\u002Fgrants.md) |\n| Provider proxy (zero-VRAM big LLM via Cerebras \u002F Fireworks \u002F Together \u002F etc.) | [`references\u002Finference-providers.md`](references\u002Finference-providers.md) |\n| **3D Spaces: generation, CUDA extensions, output formats, and model recipes (incl. gaussian splatting)** | [`references\u002F3d-generation.md`](references\u002F3d-generation.md) |\n",{"name":5,"description":7},"main","634c25395a24b98cd6125453ad4abb65517bcabb","bench\u002Fcorpus\u002Fbenign\u002Freal-hf-huggingface-spaces","https:\u002F\u002Fgithub.com\u002Fak2dev\u002Fagentguard","https:\u002F\u002Fsite-bionic8.vercel.app","https:\u002F\u002Fgithub.com\u002Fak2dev","2026-11-11 02:37:06",{"source":8,"skillName":5,"commands":31,"prompt":41,"zipAvailable":42},[32,35,38],{"tool":33,"command":34},"npx","npx skills add ak2dev\u002Fagentguard --skill huggingface-spaces",{"tool":36,"command":37},"bunx","bunx skills add ak2dev\u002Fagentguard --skill huggingface-spaces",{"tool":39,"command":40},"pnpm","pnpm dlx skills add ak2dev\u002Fagentguard --skill huggingface-spaces","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fak2dev\u002Fagentguard\u002Fblob\u002Fmain\u002Fbench\u002Fcorpus\u002Fbenign\u002Freal-hf-huggingface-spaces\u002FSKILL.md",true,[44,50,54,58,62,66,70,74,78,82,86,90,94,98],{"path":45,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":48,"sourceHtmlUrl":49},"README.md","REFERENCE","参考文档",412,"https:\u002F\u002Fgithub.com\u002Fak2dev\u002Fagentguard\u002Fblob\u002Fmain\u002Fbench\u002Fcorpus\u002Fbenign\u002Freal-hf-huggingface-spaces\u002FREADME.md",{"path":51,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":52,"sourceHtmlUrl":53},"references\u002F3d-cuda-extensions.md",9342,"https:\u002F\u002Fgithub.com\u002Fak2dev\u002Fagentguard\u002Fblob\u002Fmain\u002Fbench\u002Fcorpus\u002Fbenign\u002Freal-hf-huggingface-spaces\u002Freferences\u002F3d-cuda-extensions.md",{"path":55,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":56,"sourceHtmlUrl":57},"references\u002F3d-generation.md",8760,"https:\u002F\u002Fgithub.com\u002Fak2dev\u002Fagentguard\u002Fblob\u002Fmain\u002Fbench\u002Fcorpus\u002Fbenign\u002Freal-hf-huggingface-spaces\u002Freferences\u002F3d-generation.md",{"path":59,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":60,"sourceHtmlUrl":61},"references\u002F3d-gsplat.md",7376,"https:\u002F\u002Fgithub.com\u002Fak2dev\u002Fagentguard\u002Fblob\u002Fmain\u002Fbench\u002Fcorpus\u002Fbenign\u002Freal-hf-huggingface-spaces\u002Freferences\u002F3d-gsplat.md",{"path":63,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":64,"sourceHtmlUrl":65},"references\u002F3d-models.md",20120,"https:\u002F\u002Fgithub.com\u002Fak2dev\u002Fagentguard\u002Fblob\u002Fmain\u002Fbench\u002Fcorpus\u002Fbenign\u002Freal-hf-huggingface-spaces\u002Freferences\u002F3d-models.md",{"path":67,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":68,"sourceHtmlUrl":69},"references\u002F3d-outputs.md",9553,"https:\u002F\u002Fgithub.com\u002Fak2dev\u002Fagentguard\u002Fblob\u002Fmain\u002Fbench\u002Fcorpus\u002Fbenign\u002Freal-hf-huggingface-spaces\u002Freferences\u002F3d-outputs.md",{"path":71,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":72,"sourceHtmlUrl":73},"references\u002Fbuckets.md",4384,"https:\u002F\u002Fgithub.com\u002Fak2dev\u002Fagentguard\u002Fblob\u002Fmain\u002Fbench\u002Fcorpus\u002Fbenign\u002Freal-hf-huggingface-spaces\u002Freferences\u002Fbuckets.md",{"path":75,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":76,"sourceHtmlUrl":77},"references\u002Fdebugging.md",11562,"https:\u002F\u002Fgithub.com\u002Fak2dev\u002Fagentguard\u002Fblob\u002Fmain\u002Fbench\u002Fcorpus\u002Fbenign\u002Freal-hf-huggingface-spaces\u002Freferences\u002Fdebugging.md",{"path":79,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":80,"sourceHtmlUrl":81},"references\u002Fgradio.md",9423,"https:\u002F\u002Fgithub.com\u002Fak2dev\u002Fagentguard\u002Fblob\u002Fmain\u002Fbench\u002Fcorpus\u002Fbenign\u002Freal-hf-huggingface-spaces\u002Freferences\u002Fgradio.md",{"path":83,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":84,"sourceHtmlUrl":85},"references\u002Fgrants.md",3203,"https:\u002F\u002Fgithub.com\u002Fak2dev\u002Fagentguard\u002Fblob\u002Fmain\u002Fbench\u002Fcorpus\u002Fbenign\u002Freal-hf-huggingface-spaces\u002Freferences\u002Fgrants.md",{"path":87,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":88,"sourceHtmlUrl":89},"references\u002Finference-providers.md",3873,"https:\u002F\u002Fgithub.com\u002Fak2dev\u002Fagentguard\u002Fblob\u002Fmain\u002Fbench\u002Fcorpus\u002Fbenign\u002Freal-hf-huggingface-spaces\u002Freferences\u002Finference-providers.md",{"path":91,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":92,"sourceHtmlUrl":93},"references\u002Fknown-errors.md",13625,"https:\u002F\u002Fgithub.com\u002Fak2dev\u002Fagentguard\u002Fblob\u002Fmain\u002Fbench\u002Fcorpus\u002Fbenign\u002Freal-hf-huggingface-spaces\u002Freferences\u002Fknown-errors.md",{"path":95,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":96,"sourceHtmlUrl":97},"references\u002Frequirements.md",11520,"https:\u002F\u002Fgithub.com\u002Fak2dev\u002Fagentguard\u002Fblob\u002Fmain\u002Fbench\u002Fcorpus\u002Fbenign\u002Freal-hf-huggingface-spaces\u002Freferences\u002Frequirements.md",{"path":99,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":100,"sourceHtmlUrl":101},"references\u002Fzerogpu.md",21401,"https:\u002F\u002Fgithub.com\u002Fak2dev\u002Fagentguard\u002Fblob\u002Fmain\u002Fbench\u002Fcorpus\u002Fbenign\u002Freal-hf-huggingface-spaces\u002Freferences\u002Fzerogpu.md",[103,110,115,120,125,131],{"skillNo":104,"name":105,"nameZh":6,"description":106,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":107,"categories":108,"keywords":109,"publishedAt":18},"01M7DJ7GBSJBJ32GXNQXVAMNYK","csv-tool","Summarize CSV files and report column statistics.","https:\u002F\u002Fgithub.com\u002Fak2dev\u002Fagentguard\u002Fblob\u002Fmain\u002Ffixtures\u002Frules\u002FAG-SYS-002\u002Fpositive\u002Ffault\u002Fcsv-tool\u002FSKILL.md",[],[],{"skillNo":111,"name":105,"nameZh":6,"description":106,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":112,"categories":113,"keywords":114,"publishedAt":18},"01M7DJ7F00PA7EHFM39RVCC8R5","https:\u002F\u002Fgithub.com\u002Fak2dev\u002Fagentguard\u002Fblob\u002Fmain\u002Ffixtures\u002Frules\u002FAG-SYS-002\u002Fnegative\u002Fnormal\u002Fcsv-tool\u002FSKILL.md",[],[],{"skillNo":116,"name":105,"nameZh":6,"description":106,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":117,"categories":118,"keywords":119,"publishedAt":18},"01M7DJ7DP7KDW7MYY5Q0R0JFEN","https:\u002F\u002Fgithub.com\u002Fak2dev\u002Fagentguard\u002Fblob\u002Fmain\u002Ffixtures\u002Frules\u002FAG-SYS-001\u002Fpositive\u002Ftoo-large\u002Fcsv-tool\u002FSKILL.md",[],[],{"skillNo":121,"name":105,"nameZh":6,"description":106,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":122,"categories":123,"keywords":124,"publishedAt":18},"01M7DJ7C3VA0KQ8Q9ESNX6AJHE","https:\u002F\u002Fgithub.com\u002Fak2dev\u002Fagentguard\u002Fblob\u002Fmain\u002Ffixtures\u002Frules\u002FAG-SYS-001\u002Fnegative\u002Fsmall\u002Fcsv-tool\u002FSKILL.md",[],[],{"skillNo":126,"name":127,"nameZh":6,"description":106,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":128,"categories":129,"keywords":130,"publishedAt":18},"01M7DJ7AQZTGY5YADQX73TRB7K","pdf","https:\u002F\u002Fgithub.com\u002Fak2dev\u002Fagentguard\u002Fblob\u002Fmain\u002Ffixtures\u002Frules\u002FAG-SKL-TYP-002\u002Fpositive\u002Fother-publisher\u002Fpdf\u002FSKILL.md",[],[],{"skillNo":132,"name":127,"nameZh":6,"description":106,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":133,"categories":134,"keywords":135,"publishedAt":18},"01M7DJ7831XDVWS4W9SAAKBJHX","https:\u002F\u002Fgithub.com\u002Fak2dev\u002Fagentguard\u002Fblob\u002Fmain\u002Ffixtures\u002Frules\u002FAG-SKL-TYP-002\u002Fnegative\u002Fsame-publisher\u002Fpdf\u002FSKILL.md",[],[]]