rest-service
Guides Graphiti FastAPI REST service routes, configuration, deployment, and smoke checks.
- License 许可
- Apache-2.0
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- rest-service
- license
- Apache 2.0
- metadata
- { "disco-role": "operating" }
- description
- Guides Graphiti FastAPI REST service routes, configuration, deployment, and smoke checks.
- disable-model-invocation
- true
02 资源文件 4
- 参考文档references/configuration.md2.3 KB
- 参考文档references/rest-api.md3.5 KB
- 参考文档references/troubleshooting.md2.2 KB
- 脚本scripts/graph_service_smoke.py4.0 KB
03 SKILL.md 正文
rest-service
Use this sub-skill for the Graphiti FastAPI service surface. It covers route behavior, request/response shapes, queue semantics, environment variables, Docker startup, health checks, and live REST smoke testing.
Read first
references/rest-api.mdfor endpoint paths, request models, response shapes, and queue semantics.references/configuration.mdfor service settings, backend variables, Docker and Uvicorn startup patterns, and test commands.references/troubleshooting.mdwhen startup fails,/messagesreturns before search is ready, or a backend/credential mismatch blocks the service.scripts/graph_service_smoke.pyto exercise/healthcheck,/messages,/episodes, and/searchagainst a running service.
What belongs here
Use this sub-skill when the task mentions:
graph_service.main:app,ZepGraphiti, or the serviceSettingsmodel/healthcheck,/messages,/entity-node,/search,/entity-edge,/episodes,/get-memory,/group, or/clearDB_BACKEND,NEO4J_*, orFALKORDB_*variables for the REST API- REST service Docker or Uvicorn deployment
- the service's async message ingestion queue
What does not belong here
Route these elsewhere:
- Direct Python SDK use ->
sub-skills/core-sdk/ - MCP tool names, transports, config YAML, or MCP Docker variants ->
sub-skills/mcp-server/ - General FastAPI theory that is not tied to Graphiti's routes or settings
REST workflow
- Start a graph backend.
- Set
OPENAI_API_KEYand backend-specific variables. - Start the FastAPI app.
- Call
/healthcheck. - POST messages or entity nodes.
- Poll
/episodes/{group_id}before assuming queued ingestion has completed. - Query
/searchor/get-memory. - Delete test groups or call
/clearonly when intentional.
Important route semantics
POST /messagesreturns202 Acceptedand queues work. It does not prove the episode has already been processed.POST /entity-nodecreates a node directly and returns it.POST /searchsearches fact edges and returns afactslist.GET /episodes/{group_id}is the most useful polling path after/messages.POST /clearandDELETE /group/{group_id}are destructive.
Validation path
For a running service:
python scripts/graph_service_smoke.py --base-url http://127.0.0.1:8000
Use --health-only when you only need startup validation, and use the full smoke
when you have a live backend plus model credentials.
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).