sdk-clients
Use RocketRide Python and TypeScript/Node SDKs and CLI safely for auth, task tokens, data transfer, events, file store, and API troubleshooting.
- License 许可
- MIT
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- sdk-clients
- license
- MIT
- metadata
- { "disco-role": "operating" }
- description
- Use RocketRide Python and TypeScript/Node SDKs and CLI safely for auth, task tokens, data transfer, events, file store, and API troubleshooting.
- disable-model-invocation
- true
02 资源文件 5
- 参考文档references/cli-reference.md6.4 KB
- 参考文档references/python-sdk-api.md11.4 KB
- 参考文档references/troubleshooting.md9.0 KB
- 参考文档references/typescript-sdk-api.md13.8 KB
- 脚本scripts/sdk_import_smoke.py5.3 KB
03 SKILL.md 正文
SDK Clients
Use this sub-skill when a task is about calling RocketRide from Python,
TypeScript/Node, or the rocketride CLI: connecting, authenticating, starting a
pipeline, using the returned task token, sending data/files, streaming through a
pipe, subscribing to events, reading status, using the account file store, or
checking that the SDK imports cleanly.
Use this for
- Python
RocketRideClientscripts and async usage - TypeScript/Node
RocketRideClientscripts, browser/client usage, andDataPipe - CLI command selection and Python-vs-TypeScript CLI flag differences
- URI/auth environment handling, especially
ROCKETRIDE_URI,ROCKETRIDE_APIKEY,ROCKETRIDE_AUTH,ROCKETRIDE_PIPELINE, andROCKETRIDE_TOKEN - Task token lifecycle:
use()/start-> data operations/status ->terminate()/stop - Upload progress and server event callbacks
- Account file-store methods (
fs_*/fs*) and direct URLs - Safe import/signature smoke checks without starting services or connecting
Do not use this for
- Designing
.pipeJSON shape, lanes, sources, and recipes ->../pipeline-authoring/ - Starting engines, Docker, Helm, Cloud/self-hosting operations, or DAP protocol depth ->
../runtime-deployment/ - Node provider/service schemas and generated node documentation ->
../nodes-catalog/ - MCP server or n8n integration workflows ->
../mcp-and-integrations/ - IDE/VS Code extension behavior ->
../ide-and-apps/
First decisions
- Identify the client surface: Python SDK, TypeScript/Node SDK, browser SDK, or CLI. The method names and file-upload shapes differ.
- Identify endpoint class: local/self-hosted (
ws://...orhttp://...) versus Cloud (https://...orwss://...). For any Cloud host, never usehttp://,ws://, or a bare hostname because normalization can produce an unencryptedws://connection. - Identify authentication source. Python/TypeScript SDKs and CLIs read
ROCKETRIDE_APIKEY; MCP and some Cloud docs also useROCKETRIDE_AUTH. For SDK/CLI tasks, set or passROCKETRIDE_APIKEYunless the code explicitly copiesROCKETRIDE_AUTHinto the client auth value. - Identify whether a task token already exists. If not, start with
use()orrocketride startand capture the returned token beforesend,pipe,chat,status,stop, or file-store task-scoped operations.
Reference map
- Python SDK API — inspected Python signatures, async usage, token lifecycle, file store, events, and common pitfalls.
- TypeScript SDK API —
RocketRideClient,DataPipe, browser/Node differences, env handling, namespaces, and examples. - CLI reference — command matrix, Python vs TypeScript CLI flags, environment variables, and token use.
- Troubleshooting — auth/URI mistakes, token errors, upload/SSE issues, CLI path conflicts, file-store path rules, and safe import checks.
Safe smoke check
After installing the Python SDK in the environment that will run client code, run the bundled import/signature check. It performs no network connection:
python scripts/sdk_import_smoke.py
Use --json when another tool needs structured output.
Good output for SDK/CLI tasks
- Names the SDK/CLI surface and uses its exact method/flag names
- Uses secure Cloud URI schemes and avoids logging API keys
- Starts or reuses a task token before data operations
- Chooses
sendfor one-shot payloads,send_files/sendFilesfor files, andpipe/DataPipefor chunked or SSE-scoped streaming - Terminates tasks when done unless the workflow is intentionally long-lived
- Keeps
.pipeschema design and engine deployment details routed to the sibling sub-skills instead of expanding them here
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).