API: Management /api/v1

Native /api/v1/* routes that power Bedrova’s own UI: models, downloads, loaded models, memory, scheduler, settings, diagnostics, hardware, and capabilities. Responses follow the shared contracts.

All examples use the local base URL http://127.0.0.1:1337. This page is generated from the api-endpoints.v1 contract — see Server overview for auth, CORS, and exposure, and Settings reference for the controls.

Endpoints

GET /api/v1/models

Full model catalog with Bedrova metadata.

Auth: None on localhost; required when exposed · Streaming: no

Response. Catalog entries: modality, quant, estimated memory, compatible runtime packs, last health. Shape: model-catalog.v1.

Example request:

curl http://127.0.0.1:1337/api/v1/models

GET /api/v1/loaded-models

Models currently resident in memory.

Auth: None on localhost; required when exposed · Streaming: no

Response. Loaded/resident model registry with per-model live memory. Shape: loaded-model-registry.v1.

Example request:

curl http://127.0.0.1:1337/api/v1/loaded-models

GET /api/v1/memory

Live unified-memory accounting.

Auth: None on localhost; required when exposed · Streaming: no

Response. Resident model pool and the combined memory budget. Shape: resident-model-pool.v1.

Example request:

curl http://127.0.0.1:1337/api/v1/memory

POST /api/v1/models/fit-check

Pre-load fit-check for a model.

Returns allow / warn / block plus safer alternatives, given what is already loaded.

Auth: None on localhost; required when exposed · Streaming: no

Field	Type	Required	Notes
model_id	string	yes	Catalog model id to check.

Response. Fit decision, reasons, and safer alternatives. Shape: model-fit-check.v1.

Example request:

curl http://127.0.0.1:1337/api/v1/models/fit-check \
  -H 'Content-Type: application/json' -d '{"model_id": "dsv4-flash-q9"}'

GET /api/v1/downloads

Active and recent download jobs.

Auth: None on localhost; required when exposed · Streaming: no

Response. Download jobs with per-file and aggregate progress, pause/resume state. Shape: model-download-jobs.v1.

Example request:

curl http://127.0.0.1:1337/api/v1/downloads

GET /api/v1/hardware

Detected hardware and capability bands.

Auth: None on localhost; required when exposed · Streaming: no

Response. SoC, unified memory, and what-can-run capability bands. Shape: hardware-model-capability.v1.

Example request:

curl http://127.0.0.1:1337/api/v1/hardware

GET /api/v1/capabilities

Runtime capability report.

Auth: None on localhost; required when exposed · Streaming: no

Response. What the current runtime packs support, including gated features. Shape: runtime-capability-report.v1.

Example request:

curl http://127.0.0.1:1337/api/v1/capabilities

GET /api/v1/settings

Effective settings (secrets redacted).

Auth: None on localhost; required when exposed · Streaming: no

Response. Runtime settings and effective configuration; secrets are never echoed. Shape: runtime-settings.v1.

Example request:

curl http://127.0.0.1:1337/api/v1/settings

PATCH /api/v1/settings

Patch runtime settings.

Auth: None on localhost; required when exposed · Streaming: no

Field	Type	Required	Notes
server	object	no	Partial server settings.
loading	object	no	Partial loading/JIT/guardrail settings.
generation	object	no	Partial generation defaults.
engine_selection	object	no	Partial per-workload engine policy settings.

Response. Updated runtime settings with secrets redacted. Shape: runtime-settings.v1.

Example request:

curl -X PATCH http://127.0.0.1:1337/api/v1/settings -H 'Content-Type: application/json' -d '{"loading":{"jit_loading_enabled":true}}'

POST /api/v1/settings/export

Export diagnostics-safe settings.

Auth: None on localhost; required when exposed · Streaming: no

Response. Versioned settings export wrapper with redaction metadata.

Example request:

curl -X POST http://127.0.0.1:1337/api/v1/settings/export

POST /api/v1/settings/import

Import runtime settings.

Auth: None on localhost; required when exposed · Streaming: no

Field	Type	Required	Notes
settings	object	no	Settings object. If omitted, the request root is treated as settings.

Response. Imported and repaired runtime settings. Shape: runtime-settings.v1.

Example request:

curl -X POST http://127.0.0.1:1337/api/v1/settings/import -H 'Content-Type: application/json' -d '{"settings":{"loading":{"jit_loading_enabled":false}}}'

POST /api/v1/settings/reset

Reset runtime settings to defaults.

Auth: None on localhost; required when exposed · Streaming: no

Response. Default runtime settings after reset. Shape: runtime-settings.v1.

Example request:

curl -X POST http://127.0.0.1:1337/api/v1/settings/reset

GET /api/v1/runtime/extensions

List installed runtime extensions.

Auth: None on localhost; required when exposed · Streaming: no

Response. Runtime pack inventory. Shape: runtime-pack-inventory.v1.

Example request:

curl http://127.0.0.1:1337/api/v1/runtime/extensions

POST /api/v1/runtime/extensions/check

Check runtime extensions for updates.

Auth: None on localhost; required when exposed · Streaming: no

Response. Current extension inventory and available update records.

Example request:

curl -X POST http://127.0.0.1:1337/api/v1/runtime/extensions/check

POST /api/v1/runtime/extensions/update

Install or no-op update a runtime extension.

Auth: None on localhost; required when exposed · Streaming: no

Field	Type	Required	Notes
manifest	object	no	Signed runtime-pack manifest to install.
pack	object	no	Runtime pack installation request.

Response. Updated runtime-pack inventory or a no-op update response.

Example request:

curl -X POST http://127.0.0.1:1337/api/v1/runtime/extensions/update

GET /api/v1/logs

Read recent app-private server logs.

Auth: None on localhost; required when exposed · Streaming: no

Response. Recent structured server log entries.

Example request:

curl http://127.0.0.1:1337/api/v1/logs

GET /api/v1/diagnostics

Build a redacted diagnostics bundle.

Auth: None on localhost; required when exposed · Streaming: no

Response. Diagnostics bundle metadata: build id, git SHA, logs, versions. Prompts, files, and secrets are redacted. Shape: diagnostics-bundle.v1.

Example request:

curl http://127.0.0.1:1337/api/v1/diagnostics