Skip to content

docs: add Higress gateway external API reference - #1125

Open
LUOSENGWA wants to merge 3 commits into
agentscope-ai:mainfrom
LUOSENGWA:docs/higress-gateway-api
Open

docs: add Higress gateway external API reference#1125
LUOSENGWA wants to merge 3 commits into
agentscope-ai:mainfrom
LUOSENGWA:docs/higress-gateway-api

Conversation

@LUOSENGWA

Copy link
Copy Markdown
Contributor

Add a dedicated reference documenting the interfaces Higress exposes to the outside world (data plane): LLM OpenAI-compatible endpoints, MCP server endpoints, exposed Worker ports, and bundled service routes; plus the Higress Console API surface used by the controller (control plane).

Also fix endpoint examples in k8s-native-agent-orch.md to use the real in-container host:port (aigw-local.agentteams.io:8080) and the actual MCP server name (mcp-github), and link the new reference from the architecture overview and the README documentation tables (EN + zh-CN).

Add a dedicated reference documenting the interfaces Higress exposes to the
outside world (data plane): LLM OpenAI-compatible endpoints, MCP server
endpoints, exposed Worker ports, and bundled service routes; plus the
Higress Console API surface used by the controller (control plane).

Also fix endpoint examples in k8s-native-agent-orch.md to use the real
in-container host:port (aigw-local.agentteams.io:8080) and the actual MCP
server name (mcp-github), and link the new reference from the architecture
overview and the README documentation tables (EN + zh-CN).
@gemini-code-assist

Copy link
Copy Markdown

Caution

The consumer version of Gemini Code Assist on GitHub has been sunset. All code review activity has officially ceased.

The S3/MinIO API listens on the object store port (9000), never on the
Higress HTTP gateway (8080); the controller rewrites misconfigured
:8080 endpoints to :9000 (see normalizeMinIOS3Endpoint, agentscope-ai#718).
- Correct /v1/models status: Higress ai-proxy matches /v1/chat/completions
  and /v1/embeddings only; /v1/models is an auth/connectivity probe, not a
  full OpenAI models-list endpoint (align with Worker Guide troubleshooting).
- Add POST /v1/embeddings (used by memorySearch when configured) and note
  /v1/chat/completions as the controller readiness probe
  (IsManagerLLMAuthReady).
- Add authentication summary table (key-auth / basic-auth / session cookie /
  no auth) with exact env var names.
- Add /v1/routes/{name}/plugin-instances/{plugin} control-plane endpoint
  (used by setup-higress.sh for basic-auth).
- Clarify exposed-port URLs carry the gateway port
  (worker-{name}-{port}-local.agentteams.io:8080 in-container / :18080 host).
- Align the LLM auth-probe example with the controller's
  llmAuthProbePromptTemplate body. EN + zh-CN.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant