deepxiv-sdk
Use the DeepXiv SDK for citation-aware academic and web research through its Python Reader, deepxiv CLI, progressive paper APIs, optional local LangGraph agent, PMC, bioRxiv/medRxiv, and trending workflows.
- License 许可
- MIT
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- deepxiv-sdk
- license
- MIT
- metadata
- { "disco-role": "operating" }
- description
- Use the DeepXiv SDK for citation-aware academic and web research through its Python Reader, deepxiv CLI, progressive paper APIs, optional local LangGraph agent, PMC, bioRxiv/medRxiv, and trending workflows.
- disable-model-invocation
- true
02 资源文件 4
- 参考文档references/repo-provenance.md1.7 KB
- 参考文档references/repo-routing-metadata.json769 B
- 参考文档references/troubleshooting.md4.0 KB
- 脚本scripts/check_install.py2.9 KB
03 SKILL.md 正文
DeepXiv SDK
Use this repo skill when a task needs DeepXiv's hosted literature service or its optional local research agent. It teaches a later Researcher how to choose between the Python SDK and CLI, preserve citations and evidence strength, and avoid credential, quota, truncation, and stale-API mistakes.
Start here
- Confirm the installed package and CLI are the expected 1.0.x surface:
python -c "import deepxiv_sdk; print(deepxiv_sdk.__version__)"anddeepxiv --help. If the executable lists an older command set, repair the installation before using route-specific flags. - Install the base package with
python -m pip install deepxiv-sdk. For the optional local OpenAI-compatible LangGraph agent usepython -m pip install "deepxiv-sdk[agent]"(or[all]) and installtiktokenseparately if the optional-import probe reports it missing. - Keep DeepXiv and model-provider credentials outside prompts, source files,
logs, and commits. Ordinary retrieval uses a DeepXiv token; hosted
askrequires a registered account key and has a separate agentic quota. - Run the no-network installation check before diagnosing package or optional-dependency problems.
Route by task
- Python literature research: Read
reader-and-paper-research
for
Reader, hosted agentic search, source/citation handling, progressive reading, biomedical sources, PMC, and trending signals. - Shell commands and operations: Read cli-and-operations for command selection, flag partitions, JSON/text output, stdout/stderr, token setup, health/debug, and CLI recovery.
- Local model-driven research loop: Read
optional-local-agent for the
optional
Agent, LangGraph tools, provider options, persistent paper context, budgets, and circuit-breaker behavior.
Shared operating rules
- Prefer the narrowest evidence operation: search or
brieffirst, thenhead, one or two sections, orpreview; avoidraw/jsonfor routine screening. - Hosted
agent_searchandagent_search_streamdistinguish arXiv and web flags. Preserve arXiv IDs or URLs, separate retrieved sources from sources cited in the answer, and rejectanswer_truncatedor streamederroroutput as complete evidence. - The SDK retries ordinary GET timeouts/connections with exponential backoff, but hosted agentic calls deliberately do not auto-retry because each call can spend quota. Correct the query or wait rather than hammering the endpoint.
- Network, registered credentials, remote LLM keys, and live service results are runtime prerequisites; this skill does not claim them from import or help checks. Read cross-cutting troubleshooting and the provenance baseline when a mismatch or stale checkout is suspected.
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).