render-capture
Use PixelRAG pixelshot and pixelrag_render to capture URLs, PDFs, HTML, and images as screenshot tile directories.
- License 许可
- Apache-2.0
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- render-capture
- license
- Apache 2.0
- metadata
- { "disco-role": "operating" }
- description
- Use PixelRAG pixelshot and pixelrag_render to capture URLs, PDFs, HTML, and images as screenshot tile directories.
- disable-model-invocation
- true
02 资源文件 4
- 参考文档references/api-and-cli.md3.3 KB
- 参考文档references/tile-format.md2.0 KB
- 参考文档references/troubleshooting.md2.8 KB
- 脚本scripts/pixelrag_render_smoke.py2.6 KB
03 SKILL.md 正文
PixelRAG Render Capture
Use this sub-skill when the task is to turn a document or page into screenshot tiles with pixelshot or the pixelrag_render Python API.
Start Here
- Confirm PixelRAG is installed:
python -c "import pixelrag_render"andpixelshot --help. - For a browser render, check Chrome resolution with the root
scripts/pixelrag_doctor.pyorpixelshot which-chrome. - Choose the input route:
- URL or URL list:
pixelshot https://example.org -o ./tilesor a.txtfile with one URL per line. - Local HTML:
pixelshot page.html -o ./tiles. - PDF: install the PDF extra and run
pixelshot paper.pdf -o ./tiles --dpi 200. - Existing image:
pixelshot figure.png -o ./tilescopies it into the output area.
- URL or URL list:
- Inspect the output directory before downstream indexing: each rendered document should create
*.png.tiles/withtiles.jsonand one or moretile_*.jpgor chunk files.
Read or Run
- Read api-and-cli.md for CLI flags and Python signatures.
- Read tile-format.md before handing tiles to
index-buildor debugging manifests. - Read troubleshooting.md for Chrome, CDP attach, network-idle, PDF, and incomplete-tile failures.
- Run pixelrag_render_smoke.py for a local-only dry run or optional tiny HTML render smoke.
Common Routes
| Request | Action |
|---|---|
| "Screenshot this URL" | Use pixelshot URL -o ./tiles; add --wait-network-idle for JS-heavy pages. |
| "Capture an authenticated page" | Attach to an already-running browser with --cdp-url or PIXELSHOT_CDP_URL; PixelRAG creates and closes only its own tab. |
| "Render a PDF" | Use the pdf extra and --dpi; if Poppler/pdf2image is unavailable, resolve that first. |
| "Why is only one viewport captured?" | Read the network/page-height guidance in troubleshooting; confirm current PixelRAG includes full-page measurement fixes. |
| "Now make it searchable" | Route to ../index-build/SKILL.md after tiles are present. |
Validation Checklist
pixelshot --helplists--cdp-url,--wait-network-idle,--tile-height,--quality, and--viewport-width.pixelshot which-chromeprints an executable Chrome/Chromium path or an actionable error.- Output includes at least one tile directory,
tiles.json, and image files. - For URL batches, the number of returned tile directories matches rendered inputs; failed inputs should be visible in logs.
- Do not depend on original repo demos at runtime; use this bundled smoke script or write a small local fixture.
04 同仓库更多 Skills
passnet-triton-opt
Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block/grid/warp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.
passnet-solve
END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.
passnet-skill
PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under ./pass_dir/, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.
passnet-pattern-fusion
Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()/replacement_args()/replacement_func() files, when a pass "failed to match", or when deciding how to fuse multiple kernels/ops.
passnet-orchestrate
Round/batch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep/revert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.
passnet-feedback
Fast, deterministic feedback for PassNet work: pre-flight pattern/match verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).