Gen3 Gt Api Runbooks Anythingllm - GT-Edge-AI/GT-AI-OS GitHub Wiki
Start Here
Configure AnythingLLM Generic OpenAI LLM connection with the GT API base URL and key; align chat and optional embedding calls with the same published alias boundary. Recommended preset: Chat plus embeddings framework client.
Why this matters
This runbook maps AnythingLLM to GT API routes operators actually publish.
Details
Compatibility: Guided OpenAI-compatible setup ยท Category: Workflow and builders
Official documentation
- https://docs.anythingllm.com/setup/llm-configuration/custom/openai
- https://docs.anythingllm.com/features/chat-ui
- https://docs.anythingllm.com/setup/embedder-configuration/local
Configuration fields
- Generic OpenAI โ API Base URL:
https://<tenant-host>/api/tenant - Generic OpenAI โ API Key:
gtak_... - Generic OpenAI โ Model:
published-agent-alias
GT route mapping
| GT route | Verdict | Client integration |
|---|---|---|
GET /v1/models |
native | Model field set to published alias; optional admin test calls /v1/models |
POST /v1/chat/completions |
native | AnythingLLM Generic OpenAI chat โ /v1/chat/completions |
POST /v1/embeddings |
native | Generic OpenAI embedder pointed at GT when embed scope enabled |
POST /v1/audio/transcriptions |
not_supported | Not part of AnythingLLM Generic OpenAI chat setup |
POST /v1/audio/speech |
not_supported | Not part of AnythingLLM Generic OpenAI chat setup |
POST /v1/images/generations |
not_supported | Not supported in standard AnythingLLM OpenAI chat configuration |
POST /v1/conversations/files |
not_supported | External automation โ not default workspace upload path |
POST /v1/datasets/{id}/files |
not_supported | External automation to GT dataset upload route |
GET /v1/files/{id} |
not_supported | External polling after GT upload |
Not supported in this product
- AnythingLLM workspace document pipeline does not natively upload to GT dataset routes.
- Default embedder may bypass GT unless explicitly configured to use Generic OpenAI against GT base.
Prerequisites
- Decide whether embeddings run through GT or AnythingLLM-local models.
- Create inference key with chat-only or chat+embed scopes accordingly.
Setup steps
- Open AnythingLLM admin โ LLM Preference โ Generic OpenAI.
- Set API Base URL to
https://<tenant-host>/api/tenantand paste GT API key. - Enter published alias as the chat model name.
- Run a short workspace chat test before broader rollout.
GT extensions and caveats
- AnythingLLM document uploads do not replace GT dataset ingestion โ add external scripts if needed.
- Integrate GT conversations outside the default LLM connector when persisted GT threads are required.
Validation checklist
- Workspace chat completes with the GT alias.
- Embedding traffic through GT succeeds when embed scope is enabled.
- Operators understand published alias vs UI display labels.
Plain-text export
AnythingLLM runbook
Guided OpenAI-compatible setup ยท Workflow and builders
Recommended key preset: Chat plus embeddings framework client
Evidence: documented compatibility (vendor docs cross-check)
Configure AnythingLLM Generic OpenAI LLM connection with the GT API base URL and key; align chat and optional embedding calls with the same published alias boundary.
Official documentation:
- https://docs.anythingllm.com/setup/llm-configuration/custom/openai
- https://docs.anythingllm.com/features/chat-ui
- https://docs.anythingllm.com/setup/embedder-configuration/local
Configuration fields:
- Generic OpenAI โ API Base URL: https://<tenant-host>/api/tenant
- Generic OpenAI โ API Key: gtak_...
- Generic OpenAI โ Model: published-agent-alias
GT route mapping:
- GET /v1/models (native): Model field set to published alias; optional admin test calls `/v1/models`
- POST /v1/chat/completions (native): AnythingLLM Generic OpenAI chat โ `/v1/chat/completions`
- POST /v1/embeddings (native): Generic OpenAI embedder pointed at GT when embed scope enabled
- POST /v1/audio/transcriptions (not_supported): Not part of AnythingLLM Generic OpenAI chat setup
- POST /v1/audio/speech (not_supported): Not part of AnythingLLM Generic OpenAI chat setup
- POST /v1/images/generations (not_supported): Not supported in standard AnythingLLM OpenAI chat configuration
- POST /v1/conversations/files (not_supported): External automation โ not default workspace upload path
- POST /v1/datasets/{id}/files (not_supported): External automation to GT dataset upload route
- GET /v1/files/{id} (not_supported): External polling after GT upload
Not supported in this product:
- AnythingLLM workspace document pipeline does not natively upload to GT dataset routes.
- Default embedder may bypass GT unless explicitly configured to use Generic OpenAI against GT base.
Prerequisites:
- Decide whether embeddings run through GT or AnythingLLM-local models.
- Create inference key with chat-only or chat+embed scopes accordingly.
Setup steps:
1. Open AnythingLLM admin โ LLM Preference โ Generic OpenAI.
2. Set API Base URL to `https://<tenant-host>/api/tenant` and paste GT API key.
3. Enter published alias as the chat model name.
4. Run a short workspace chat test before broader rollout.
GT extensions and caveats:
- AnythingLLM document uploads do not replace GT dataset ingestion โ add external scripts if needed.
- Integrate GT conversations outside the default LLM connector when persisted GT threads are required.
Validation checklist:
- Workspace chat completes with the GT alias.
- Embedding traffic through GT succeeds when embed scope is enabled.
- Operators understand published alias vs UI display labels.