AGENT SKILL

agentic-rag

Use PaperQA's Python API for agentic and manual scientific RAG over Docs, Doc, Text, ask, agent_query, evidence retrieval, answers, callbacks, and async boundaries.

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

01 前言 Front Matter

name
agentic-rag
license
Apache 2.0
metadata
{ "disco-role": "operating" }
description
Use PaperQA's Python API for agentic and manual scientific RAG over Docs, Doc, Text, ask, agent_query, evidence retrieval, answers, callbacks, and async boundaries.
disable-model-invocation
true

02 资源文件 4

03 SKILL.md 正文

PaperQA Agentic RAG API

Use this sub-skill when the task is to run PaperQA from Python: build Docs, add Doc/Text objects or local files/texts, retrieve evidence with Docs.aget_evidence, generate answers with Docs.aquery, or use ask/agent_query with fake or ToolSelector agents.

Route here for

  • Manual Docs workflows: aadd, aadd_file, aadd_url, aadd_texts, aget_evidence, aquery.
  • Agent workflows: ask, agent_query, agent_type="fake", default ToolSelector, agent callbacks, answer/session/status outputs.
  • No-network object smokes and pre-chunked text ingestion with deferred embeddings.
  • Understanding PQASession, Context, AnswerResponse, citations, references, and source/context outputs.

Route elsewhere

  • CLI commands, index build/search/reuse, pqa ask/search/index, manifests: cli-and-indexing.
  • Parser choice, PDF/media/Office/code parsing details, impossible/corrupt documents: docs-and-parsing.
  • Provider/model/settings tuning, local embeddings, Qdrant, prompts beyond API usage: settings-and-configuration.
  • Metadata clients, DOI/title hydration, OpenReview/Zotero/ClinicalTrials sources: metadata-and-sources.

Start with the bundled references

  1. Check verified signatures and return shapes in references/api-reference.md.
  2. Use self-contained recipes in references/workflows.md.
  3. Diagnose common runtime failures with references/troubleshooting.md.
  4. For a safe no-network smoke, run scripts/smoke_docs_objects.py with --help, then smoke.

Operating cautions

  • Prefer async APIs (await docs.aadd_texts(...), await docs.aquery(...)) in notebooks, web servers, and other running event loops.
  • Passing a citation and disabling metadata hydration avoids citation-inference LLM calls and metadata-provider network calls during manual file addition.
  • Settings(parsing={"defer_embedding": True}) defers embeddings at ingest; retrieval/answering can still call the configured embedding and LLM providers.
  • The no-network smoke only proves installed object/API construction; it does not validate provider credentials, embeddings, parsing quality, or answer quality.

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