cli-and-service
Use DeepSearcher through the console command or bundled FastAPI service helper, including flags, endpoints, startup checks, and deployment troubleshooting.
- License 许可
- Apache-2.0
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- cli-and-service
- license
- Apache 2.0
- metadata
- { "disco-role": "operating" }
- description
- Use DeepSearcher through the console command or bundled FastAPI service helper, including flags, endpoints, startup checks, and deployment troubleshooting.
- disable-model-invocation
- true
02 资源文件 6
- 参考文档references/cli-reference.md1.7 KB
- 参考文档references/service-reference.md2.7 KB
- 参考文档references/troubleshooting.md2.3 KB
- 脚本scripts/check_cli_help.py8.5 KB
- 脚本scripts/check_service_routes.py2.7 KB
- 脚本scripts/serve_deepsearcher_api.py7.6 KB
03 SKILL.md 正文
CLI and Service
Use this sub-skill when the task is about the deepsearcher console command, command-line help, the HTTP service surface, or deployment-oriented checks. It covers the public command parser and a source-free FastAPI helper that mirrors the important endpoint shapes without depending on the original checkout's main.py.
Route here for
deepsearcher query ...anddeepsearcher load ...flag usage.- Deprecated
--query/--loadbehavior. - CLI help failures caused by early provider initialization.
- A bundled HTTP service helper that exposes
/set-provider-config/,/load-files/,/load-website/, and/query/. - Service startup checks, route checks, and lightweight deployment validation.
Route elsewhere
- Provider selection, credentials, and optional packages:
provider-configuration. - Data loading, crawling, chunking, and collection creation:
data-ingestion. - Query/retrieve internals and RAG agent behavior:
rag-query. - Retrieval benchmarks and 2WikiMultiHopQA evaluation:
evaluation.
Reference map
- CLI reference: exact command names, positional arguments, and flags.
- Service reference: endpoint payloads, response shapes, and startup options.
- Troubleshooting: help-time configuration failures, missing credentials, Milvus Lite locks, and source-free deployment checks.
Safe bundled helpers
- scripts/check_cli_help.py: run CLI help in a temp working directory with optional dummy env injection and classify common initialization failures.
- scripts/serve_deepsearcher_api.py: run a source-free FastAPI helper that uses installed DeepSearcher APIs and mirrors the service surface.
- scripts/check_service_routes.py: verify the bundled service helper exposes the expected routes.
Operating rules
- Do not tell future agents to run the original repository's
main.py; use the bundled service helper instead. - Treat CLI help as potentially stateful. The inspected checkout initializes configuration before argparse finishes, so
--helpcan fail if the default provider stack is not ready. - Use a temp working directory for help probes so the local Milvus default does not reuse an existing
./milvus.dblock. - Keep provider setup, ingestion, query, and evaluation routed to the sibling sub-skills when the task is not specifically about the console command or HTTP endpoints.
Minimal examples
python scripts/check_cli_help.py --command all
python scripts/check_service_routes.py --json
python scripts/serve_deepsearcher_api.py --host 127.0.0.1 --port 8000 --enable-cors
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).