AGENT SKILL

docs-and-parsing

Use PaperQA parsing, chunking, document-format, PDF-reader, and multimodal media workflows.

License 许可
Apache-2.0
最近更新
2026-09-03
Stars
★ 351
Forks
27

01 前言 Front Matter

name
docs-and-parsing
license
Apache 2.0
metadata
{ "disco-role": "operating" }
description
Use PaperQA parsing, chunking, document-format, PDF-reader, and multimodal media workflows.
disable-model-invocation
true

02 资源文件 5

03 SKILL.md 正文

docs-and-parsing

Use this sub-skill when the task is to parse local documents into PaperQA Text chunks, select a PDF reader, diagnose parser dependency failures, or reason about multimodal media extraction before retrieval/querying.

Route here for

  • paperqa.readers.read_doc, parse_text, parse_image, parse_office_doc, chunk_text, chunk_pdf, and code/Markdown line-based chunking.
  • Local PDFs, .txt, Markdown/source/config files, .html, Office files, standalone images, and parser kwargs such as page_range, page_size_limit, parse_media, full_page, dpi, and reader_config.
  • Choosing among paper-qa-pypdf, paper-qa-pymupdf, paper-qa-docling, and paper-qa-nemotron, including optional extras and service requirements.
  • Understanding media parsing versus media enrichment cost and retrieval effects.

Route elsewhere

  • RAG querying, Docs.aquery, ask, agent_query, answer/evidence objects, and callbacks: use ../agentic-rag/.
  • CLI index scanning, manifest columns, file filters, full-text index reuse, and pqa commands: use ../cli-and-indexing/.
  • LLM/provider selection, enrichment model credentials, named settings, prompts, embeddings, or vector stores: use ../settings-and-configuration/.

Operating entry points

  1. Start with references/parsers-and-formats.md to choose the file-format path and PDF reader.
  2. Use references/api-reference.md for verified signatures, return shapes, parser kwargs, and safe parse/chunk recipes.
  3. Use references/troubleshooting.md when imports, corrupt files, media parsing, reader services, or enrichment fail.
  4. Run scripts/inspect_parsers.py to inventory parser availability without requiring optional packages.
  5. Run scripts/parse_text_smoke.py for a no-network text/Markdown parse-and-chunk smoke before attempting PDFs or Office files.

Minimal safe pattern

For a local text or Markdown file, parse and chunk without LLMs or embeddings:

from paperqa.readers import parse_text, chunk_text
from paperqa.types import Doc

parsed = parse_text("notes.md")
texts = chunk_text(
    parsed,
    Doc(docname="notes", citation="Local notes", dockey="notes"),
    chunk_chars=1000,
    overlap=100,
)

For PDFs, construct Settings(parsing={...}) or pass parse_pdf= directly to read_doc; do not assume every optional reader or media backend is installed.

04 同仓库更多 Skills

VectorSpaceLab
VectorSpaceLab/AREX-Skill

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.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

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.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

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.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

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.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

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.

★ 351
VectorSpaceLab
VectorSpaceLab/AREX-Skill

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).

★ 351