core-runtime
Start, configure, deploy, inspect, and troubleshoot LangBot core runtime boot, Application wiring, config.yaml, health checks, Docker or standalone Plugin Runtime and Box modes.
- License 许可
- Apache-2.0
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- core-runtime
- license
- Apache 2.0
- metadata
- { "disco-role": "operating" }
- description
- Start, configure, deploy, inspect, and troubleshoot LangBot core runtime boot, Application wiring, config.yaml, health checks, Docker or standalone Plugin Runtime and Box modes.
- disable-model-invocation
- true
02 资源文件 3
- 参考文档references/deployment-and-operations.md2.7 KB
- 参考文档references/startup-and-configuration.md3.1 KB
- 参考文档references/troubleshooting.md1.7 KB
03 SKILL.md 正文
Core Runtime
Use this sub-skill when the task is about getting LangBot running, reasoning about the boot graph, changing runtime configuration, deployment topology, startup flags, health/resource reporting, or operational probes.
Read First
- references/startup-and-configuration.md
for entry points,
config.yaml, boot stages, and local startup commands. - references/deployment-and-operations.md for Docker/Compose/Kubernetes, Plugin Runtime, Box Runtime, and resource probes.
- references/troubleshooting.md for startup, config, port, dependency, and runtime-service failures.
Core Facts
main.pydelegates tolangbot.__main__.main().- The
langbotconsole script accepts--standalone-runtime,--standalone-box,--debug, and an operatormigrate --cloudsubcommand. - First startup generates missing data/config files and initializes the database before serving HTTP.
BuildAppStagewires the long-livedApplicationobject: managers, services, runtime connectors, controllers, storage, telemetry, and task management.- The HTTP controller serves Quart/Hypercorn on
api.portand mounts MCP at/mcpafter route registration. - Default local data lives under the current working directory's
data/when running the package or source command.
Typical Workflows
Local source startup
uv sync --dev
uv run main.py
If you intentionally installed a local sibling SDK checkout, preserve it with:
uv run --no-sync main.py
Use --standalone-runtime when an external Plugin Runtime is already running,
and --standalone-box when an external Box Runtime is already running.
Package startup
uvx langbot
# or
pip install langbot
langbot --help
langbot
Health check
curl http://127.0.0.1:5300/healthz
Expect code: 0 plus resource counters. If the route is unavailable, diagnose
boot, port binding, config generation, and dependency installation before
looking at higher-level pipeline behavior.
Boundaries
- HTTP route auth and MCP tool design belong to
api-mcp-web. - Platform message flow, adapters, pipelines, providers, and tools belong to
platform-pipeline-provider. - Plugin Runtime and Box API details belong to
plugin-box-skills. - Database migrations, RAG, vector backends, storage, and tenancy belong to
persistence-rag-workspaces. - Verification selection belongs to
testing-qa.
Validation
Use startup-focused checks before broad suites:
python scripts/langbot_repo_doctor.py --repo-root /path/to/LangBot
uv run --python 3.12 pytest tests/e2e -q --tb=short
Only run long resource probes or cloud soak gates when the task is specifically about runtime retention, capacity, or production stability.
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).