runtime-deployment
Start, self-host, debug, and deploy RocketRide runtime and engine.
- License 许可
- MIT
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- runtime-deployment
- license
- MIT
- metadata
- { "disco-role": "operating" }
- description
- Start, self-host, debug, and deploy RocketRide runtime and engine.
- disable-model-invocation
- true
02 资源文件 3
- 参考文档references/docker-helm.md10.3 KB
- 参考文档references/self-hosting-and-protocol.md11.1 KB
- 参考文档references/troubleshooting.md11.6 KB
03 SKILL.md 正文
Runtime Deployment
Use this sub-skill when a task is about starting a RocketRide engine, choosing Cloud versus self-hosting, checking runtime connectivity, explaining the native WebSocket/DAP protocol, or deploying the engine with a release archive, source build, Docker Compose, or Helm.
Use this for
- Starting a local or on-prem RocketRide engine on port
5565 - Choosing the right client URI for Cloud, a local binary, Docker, or Kubernetes
- Debugging
/ping,/task/service, WebSocket upgrades, auth, and task events - Explaining which SDK calls map to which DAP-style runtime commands
- Planning release-archive, source-build, Docker Compose, or Helm deployments
- Reasoning about observability events, trace levels, and runtime metrics
- Placing provider API keys, database credentials, and vector-store settings in
the engine environment without hardcoding secrets in
.pipefiles
Do not use for
- Python or TypeScript SDK method details and code examples →
../sdk-clients/ .pipedesign, lane wiring, or pipeline recipes →../pipeline-authoring/- Node provider schemas and generated node docs →
../nodes-catalog/ - VS Code extension/runtime UX and visual app surfaces →
../ide-and-apps/ - MCP server, n8n, assistant-tool, or webhook integration details →
../mcp-and-integrations/
Route first
- Read self-hosting and protocol for
release archives, source builds, Linux runtime dependencies, the
enginecommand, Cloud/local URI rules,/ping,/task/service, DAP frames, SDK-to- command mapping, auth, and observability. - Read Docker and Helm for Compose services, host versus container networking, Kubernetes values, secrets, external databases, GPU scheduling, HA, probes, and deployment validation.
- Read troubleshooting when a runtime does not listen, a client cannot connect, auth fails, observability is silent, Docker networking is wrong, Helm rendering/deployment fails, or provider credentials are missing.
Safety and execution guardrails
- Do not start long-running engines, Docker stacks, Kubernetes deployments, or source builds unless the user explicitly asks for that action and accepts the local side effects.
- Prefer static review and command planning when the user only asks what to run.
- Never test real provider credentials or Cloud tokens unless the user supplies them for that purpose.
- Treat source builds, Docker, Helm, GPU, database, and external-provider paths as optional deployment workflows, not required smoke checks for ordinary skill use.
Quick triage order
- Identify target: Cloud, release archive, source-built runtime directory, Docker Compose, or Helm.
- Check the expected endpoint:
- Health:
http://<host>:5565/ping - Task socket:
ws://<host>:5565/task/service - Cloud base URI:
https://api.rocketride.aiwith token auth
- Health:
- Verify whether the problem is a listening/binding issue, URI/scheme issue, authentication issue, proxy/WebSocket-upgrade issue, or provider/runtime dependency issue.
- Use runtime events only for live observation: RocketRide observability is a WebSocket event stream, not a durable metrics database.
Good output for deployment tasks
- Names the target mode and exact endpoint scheme
- Shows a health check and expected signal
- Separates local auth assumptions from Cloud/production auth requirements
- Keeps provider secrets in environment variables or Kubernetes Secrets
- Notes when a command is heavy or starts services
- Includes rollback/stop or dry-run guidance for Docker and Helm
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).