Sources API

Manage document sources -- upload, process, tag, and monitor extraction.

All endpoints are prefixed with /api/v1/sources unless noted otherwise.

User guide: Sources — how to upload and manage sources in the UI, CLI, and Python SDK
Architecture: Extraction Pipeline — how the multi-stage pipeline works internally

Upload & Import

Upload Single File

POST /api/v1/sources

Upload a document via multipart form data. Returns 202 Accepted immediately while indexing and extraction run in the background.

curl -X POST http://localhost:8080/api/v1/sources \
  -F "file=@document.pdf"

Parameter	Type	Required	Default	Description
`file`	file	Yes	--	Document file to upload
`extract_entities`	bool	No	`true`	Run entity extraction after indexing
`analysis_depth`	string	No	`full`	Extraction depth: `full` or `quick`
`domain`	string	No	`null`	Force extraction domain (e.g. `technical`, `generic`). Auto-detected if omitted.
`enable_normalization`	bool	No	`true`	Normalize content on upload (encoding fixes, whitespace, OCR cleaning). Disable for code or structured data.
`enable_vision`	bool	No	auto-detect	Enable vision processing for images in PDFs and image files. Default: auto-detect based on vision model configuration.
`content_filtering`	bool	No	`true`	Filter non-essential content (TOC, legal, boilerplate) from entity extraction. Filtered content remains searchable via RAG.
`filtering_mode`	string	No	`balanced`	Strictness of post-extraction filters: `unfiltered`, `minimal`, `lenient`, `balanced`, `strict`, `maximum`. See Filtering Modes.
`skip_duplicates`	bool	No	`false`	Skip upload if identical content already exists (by SHA-256 hash)

Upload settings are persistent

auto_analyze, enable_normalization, enable_vision, content_filtering, and filtering_mode are persisted on the source row at upload time. Recovery, retry, and re-extract reuse the persisted values by default — clients only re-pass them when they want to override.

Response 202 Accepted -- SourceResponse

{
  "id": "src_abc123",
  "filename": "document.pdf",
  "file_type": "pdf",
  "file_size": 204800,
  "status": "pending",
  "enabled": true,
  "extraction_depth": "full",
  "created_at": "2026-03-09T12:00:00",
  "updated_at": "2026-03-09T12:00:00"
}

Key fields shown above. The full response includes lifecycle timestamps (indexing_*, extraction_*, commit_*), LLM metrics (llm_total_calls, llm_total_input_tokens, etc.), and progress fields (current_step, step_description) — all initially null or 0. See SourceResponse for the complete schema.

Polling for progress

Use GET /api/v1/sources/{id} to poll the source status as it transitions through pending -> indexing -> indexed -> extracting -> extracted -> committing -> committed.

Batch Upload

POST /api/v1/sources/batch

Upload multiple files simultaneously. Returns 202 Accepted.

curl -X POST http://localhost:8080/api/v1/sources/batch \
  -F "files=@doc1.pdf" \
  -F "files=@doc2.pdf"

Parameter	Type	Required	Default	Description
`files`	file[]	Yes	--	Multiple document files
`extract_entities`	bool	No	`true`	Run entity extraction after indexing
`analysis_depth`	string	No	`full`	Extraction depth: `full` or `quick`
`enable_normalization`	bool	No	`true`	Normalize content on upload
`enable_vision`	bool	No	auto-detect	Enable vision processing for images in PDFs and image files
`domain`	string	No	`null`	Force extraction domain
`content_filtering`	bool	No	`true`	Filter non-essential content (TOC, legal, boilerplate) from entity extraction. Filtered content remains searchable via RAG.
`filtering_mode`	string	No	`balanced`	Strictness of post-extraction filters. See Filtering Modes.
`skip_duplicates`	bool	No	`false`	Skip files whose content already exists

Response 202 Accepted

{
  "uploaded": 2,
  "failed": 0,
  "files": [
    { "id": "src_abc123", "filename": "doc1.pdf", "status": "pending", "..." : "..." },
    { "id": "src_def456", "filename": "doc2.pdf", "status": "pending", "..." : "..." }
  ],
  "errors": []
}

Each item in files is a full SourceResponse. When a file fails, it appears in errors instead:

{
  "uploaded": 1,
  "failed": 1,
  "files": [ { "..." : "..." } ],
  "errors": [
    { "filename": "bad.xyz", "error": "Unsupported file type" }
  ]
}

Batch size limit

Returns 400 if the number of files exceeds the configured max_upload_files limit.

Import from URL

POST /api/v1/sources/url

Fetch a web page, extract clean markdown content, and process it through the standard file pipeline.

curl -X POST http://localhost:8080/api/v1/sources/url \
  -H "Content-Type: application/json" \
  -d '{"url": "https://example.com/article"}'

Parameter	Type	Required	Default	Description
`url`	string	Yes	--	URL to import (must start with `http://` or `https://`)
`extract_entities`	bool	No	`true`	Run entity extraction after indexing
`analysis_depth`	string	No	`full`	Extraction depth: `full` or `quick`
`enable_normalization`	bool	No	`true`	Normalize content on upload
`enable_vision`	bool	No	auto-detect	Enable vision processing for images in fetched HTML / PDFs
`domain`	string	No	`null`	Force extraction domain
`content_filtering`	bool	No	`true`	Filter non-essential content from entity extraction. Filtered content remains searchable via RAG.
`filtering_mode`	string	No	`balanced`	Strictness of post-extraction filters. See Filtering Modes.
`skip_duplicates`	bool	No	`false`	Skip if identical content exists

URL fetcher Content-Type validation

The URL fetcher validates the upstream Content-Type against the same allowlist used for direct file uploads (batching.allowed_content_types). It honors any charset=… parameter in the response header and routes binary responses (PDF, ZIP, DOCX, etc.) to the binary loader path so application/pdf URLs are no longer mishandled as HTML.

Response 202 Accepted -- SourceResponse

The response is identical in shape to the single file upload response. The source_type will be webpage and origin_url will contain the imported URL.

Status	Description
`400`	Invalid URL format
`422`	Failed to fetch URL or content shorter than 50 characters

Source CRUD

List Sources

GET /api/v1/sources

Paginated list of sources with optional filters. Returns PaginatedSourcesResponse containing SourceSummaryResponse items.

curl "http://localhost:8080/api/v1/sources?status=committed"

Parameter	Type	Required	Default	Description
`page`	int	No	`1`	Page number (1-indexed)
`page_size`	int	No	Server default (50)	Items per page (capped at `max_page_size`)
`source_type`	string	No	`null`	Filter by source type (`pdf`, `text`, `csv`, `webpage`, etc.)
`status`	string	No	`null`	Filter by processing status (`pending`, `indexing`, `indexed`, `extracting`, `extracted`, `committing`, `committed`, `error`)
`enabled`	string	No	`null`	Filter by enabled state: `enabled` or `disabled`
`search`	string	No	`null`	Search in title and origin URL
`tag_id`	string	No	`null`	Filter by tag ID

Response 200 OK -- PaginatedSourcesResponse

{
  "data": [
    {
      "id": "src_abc123",
      "filename": "research-paper.pdf",
      "file_type": "pdf",
      "file_size": 204800,
      "title": "A Research Paper",
      "status": "committed",
      "chunk_count": 42,
      "extraction_entities_count": 85,
      "extraction_relationships_count": 120,
      "cached_quality_grade": "A",
      "tags": [
        { "id": "tag_001", "name": "Research", "color": "#4dabf5" }
      ],
      "created_at": "2026-03-09T10:00:00"
    }
  ],
  "pagination": {
    "total": 1,
    "page": 1,
    "page_size": 20,
    "total_pages": 1,
    "has_next": false,
    "has_prev": false
  }
}

Each item is a SourceSummaryResponse with additional fields including embedding info, LLM metrics, duration timings, and quality scores.

Get Source

GET /api/v1/sources/{source_id}

Returns the full source detail including all lifecycle fields, LLM metrics, and user metadata.

curl http://localhost:8080/api/v1/sources/src_abc123

Parameter	Type	Required	Description
`source_id`	string (path)	Yes	Source ID

Response 200 OK -- SourceResponse

See the Upload Single File section for a full response example.

Status	Description
`404`	Source not found

Update Source

PATCH /api/v1/sources/{source_id}

Update mutable source fields.

curl -X PATCH http://localhost:8080/api/v1/sources/src_abc123 \
  -H "Content-Type: application/json" \
  -d '{
    "title": "Updated Title",
    "enabled": true,
    "user_metadata": { "category": "research" }
  }'

Parameter	Type	Required	Description
`source_id`	string (path)	Yes	Source ID
`title`	string	No	New display title
`processing_status`	string	No	Override status (`ready` or `error`)
`enabled`	bool	No	Enable or disable the source
`user_metadata`	object	No	Arbitrary key-value metadata

Response 200 OK -- SourceResponse

Status	Description
`404`	Source not found

Delete Source

DELETE /api/v1/sources/{source_id}

Permanently deletes the source and cascades to all chunks, citations, graph nodes, edges, templates, and search index entries.

curl -X DELETE http://localhost:8080/api/v1/sources/src_abc123

Parameter	Type	Required	Description
`source_id`	string (path)	Yes	Source ID

Response 204 No Content

Status	Description
`404`	Source not found

Source Metadata

List Extraction Domains

GET /api/v1/sources/domains

Returns available extraction domains for dropdown selection. Includes built-in domains and any per-database custom domains.

curl http://localhost:8080/api/v1/sources/domains

Response 200 OK

{
  "domains": [
    {
      "name": "generic",
      "description": "General-purpose entity extraction",
      "builtin": true,
      "extraction_density": "medium",
      "prompt_tokens": 1200
    },
    {
      "name": "technical",
      "description": "Technical documentation and specifications",
      "builtin": true,
      "extraction_density": "high",
      "prompt_tokens": 1800
    }
  ]
}

Get Processing Stats

GET /api/v1/sources/stats

Aggregate processing statistics across all sources.

curl http://localhost:8080/api/v1/sources/stats

Response 200 OK

{
  "total_files": 25,
  "by_status": {
    "committed": 20,
    "indexed": 3,
    "error": 2
  },
  "total_chunks": 1042,
  "total_entities": 850,
  "total_relationships": 1200
}

Extraction Management

Trigger Extraction

POST /api/v1/sources/{source_id}/extraction

Trigger manual entity extraction for a source. The source must be in indexed or extracted status. Returns 202 Accepted while extraction runs in the background.

curl -X POST http://localhost:8080/api/v1/sources/src_abc123/extraction \
  -H "Content-Type: application/json" \
  -d '{
    "analysis_depth": "full",
    "domain": "technical",
    "force": false
  }'

Parameter	Type	Required	Default	Description
`source_id`	string (path)	Yes	--	Source ID
`analysis_depth`	string	No	`full`	Extraction depth: `full` or `quick`
`domain`	string	No	`null`	Force extraction domain. Auto-detected if omitted.
`filtering_mode`	string	No	persisted	Override the source's persisted `filtering_mode` for this run only.
`force`	bool	No	`false`	Re-extract even if extraction results already exist

The endpoint reuses the source's persisted upload settings (filtering_mode, enable_vision, content_filtering) by default. Pass them in the body to override per-call without changing the row.

Response 202 Accepted

{
  "source_id": "src_abc123",
  "job_id": "job_xyz789",
  "status": "queued",
  "message": "Extraction started"
}

Status	Description
`400`	Source is not in an extractable state
`404`	Source not found
`409`	Extraction already in progress (use `force=true` to re-extract)

Get Extraction Progress

GET /api/v1/sources/{source_id}/extraction

Returns detailed extraction progress including job status, chunk-level counts, and timing estimates.

curl http://localhost:8080/api/v1/sources/src_abc123/extraction

Parameter	Type	Required	Description
`source_id`	string (path)	Yes	Source ID

Response 200 OK

{
  "source_id": "src_abc123",
  "job_id": "job_xyz789",
  "status": "running",
  "has_extraction_job": true,
  "total_chunks": 10,
  "completed_chunks": 6,
  "failed_chunks": 0,
  "progress_percent": 60.0,
  "chunks_by_status": {
    "completed": 6,
    "running": 1,
    "queued": 3
  },
  "total_entities": 52,
  "total_relationships": 78,
  "extraction_depth": "full",
  "started_at": "2026-03-09T10:00:00",
  "completed_at": null,
  "timing": {
    "avg_duration_ms": 4200,
    "min_duration_ms": 2100,
    "max_duration_ms": 6800
  },
  "current_chunk": {
    "chunk_index": 6,
    "status": "running",
    "started_at": "2026-03-09T10:02:30"
  }
}

When no extraction job exists:

{
  "source_id": "src_abc123",
  "status": "indexed",
  "has_extraction_job": false,
  "message": "No active extraction job for this source"
}

Status	Description
`404`	Source not found

Cancel Extraction

DELETE /api/v1/sources/{source_id}/extraction

Cancels all pending and queued extraction chunks. Already running or completed chunks are not affected. Source status reverts to indexed (RAG search still works).

curl -X DELETE http://localhost:8080/api/v1/sources/src_abc123/extraction

Parameter	Type	Required	Description
`source_id`	string (path)	Yes	Source ID

Response 204 No Content

Status	Description
`404`	Source not found or no active extraction job

Reclassify Source Domain

POST /api/v1/sources/{source_id}/reclassify

Change the extraction domain for a source and queue a new extraction pass. Returns 202 Accepted while the new extraction runs in the background.

For sources that are already committed, this endpoint atomically resets prior graph artifacts (nodes, edges, templates) before dispatching so the new extraction starts clean.

When to use: When auto-detection chose the wrong domain, or when you want to re-run extraction under a different domain template after reviewing the initial results. Prefer this over setting domain at upload time — reclassify decouples domain selection from the upload flow.

curl -X POST http://localhost:8080/api/v1/sources/src_abc123/reclassify \
  -H "Content-Type: application/json" \
  -d '{"domain": "medical"}'

Parameter	Type	Required	Description
`source_id`	string (path)	Yes	Source ID
`domain`	string	Yes	Domain name to use (e.g. `medical`, `legal`). See `GET /sources/domains`.

Response 202 Accepted

{
  "source_id": "src_abc123",
  "status": "extracting"
}

Status	Description
`400`	Source is not in a reclassifiable state (`indexed` or `committed` required)
`404`	Source not found
`503`	No LLM provider configured

List Extraction Tasks

GET /api/v1/sources/{source_id}/extraction/tasks

Paginated list of individual chunk extraction tasks (LLM processing groups). Useful for debugging and analytics.

curl "http://localhost:8080/api/v1/sources/src_abc123/extraction/tasks?page=1&page_size=20"

Parameter	Type	Required	Default	Description
`source_id`	string (path)	Yes	--	Source ID
`page`	int	No	`1`	Page number (1-indexed)
`page_size`	int	No	Server default	Items per page
`include_content`	bool	No	`false`	Include full `input_text` and `llm_response_json` (large payloads)

Response 200 OK -- ExtractionTaskListResponse

{
  "tasks": [
    {
      "id": "task_001",
      "job_id": "job_xyz789",
      "chunk_index": 0,
      "hierarchical_group_id": "group_a",
      "small_chunk_ids": ["chunk_001", "chunk_002"],
      "status": "completed",
      "created_at": "2026-03-09T10:00:00",
      "queued_at": "2026-03-09T10:00:01",
      "started_at": "2026-03-09T10:00:05",
      "completed_at": "2026-03-09T10:00:09",
      "llm_duration_ms": 3800,
      "retry_count": 0,
      "entity_count": 8,
      "relationship_count": 12,
      "invalid_relationship_count": 1,
      "small_chunk_numbers": [1, 2],
      "input_text_length": 3200,
      "llm_response_length": 1800,
      "input_tokens": 1100,
      "output_tokens": 620,
      "context_window_available": 128000,
      "input_text": null,
      "llm_response_json": null,
      "filtering_log": null,
      "error_message": null,
      "error_type": null
    }
  ],
  "total": 10,
  "page": 1,
  "page_size": 20
}

Content fields

Set include_content=true to populate input_text and llm_response_json. By default only their lengths are returned for performance.

Get Extraction Task

GET /api/v1/sources/{source_id}/extraction/tasks/{task_id}

Returns a single extraction task with full details, including content fields.

curl http://localhost:8080/api/v1/sources/src_abc123/extraction/tasks/task_001

Parameter	Type	Required	Description
`source_id`	string (path)	Yes	Source ID
`task_id`	string (path)	Yes	Extraction task ID

Response 200 OK -- ExtractionTaskResponse

Same shape as items in the task list, but with input_text, llm_response_json, and filtering_log fully populated.

Status	Description
`404`	Extraction task not found

Get Extraction Task Stats

GET /api/v1/sources/{source_id}/extraction/stats

Aggregate statistics (min/avg/max) for extraction tasks, computed via SQL aggregates without loading every row.

curl http://localhost:8080/api/v1/sources/src_abc123/extraction/stats

Parameter	Type	Required	Description
`source_id`	string (path)	Yes	Source ID

Response 200 OK -- ExtractionTaskStatsResponse

{
  "total_tasks": 10,
  "context_window": 128000,
  "min_input_tokens": 800,
  "max_input_tokens": 2400,
  "avg_input_tokens": 1500,
  "min_output_tokens": 300,
  "max_output_tokens": 900,
  "avg_output_tokens": 600,
  "min_total_tokens": 1100,
  "max_total_tokens": 3300,
  "avg_total_tokens": 2100,
  "min_utilization": 0.86,
  "max_utilization": 2.58,
  "avg_utilization": 1.64,
  "min_duration_ms": 2100,
  "max_duration_ms": 6800,
  "avg_duration_ms": 4200,
  "total_entities": 85,
  "avg_entities_per_task": 8.5,
  "total_relationships": 120,
  "avg_relationships_per_task": 12.0,
  "total_retries": 2,
  "max_retries_single_task": 1,
  "total_invalid_relationships": 5,
  "avg_invalid_per_task": 0.5,
  "total_entities_filtered": 3,
  "total_relationships_filtered": 7,
  "filtering_stage_summary": [
    { "stage": "exact_dedup", "total_removed": 2, "chunk_count": 2 },
    { "stage": "relationship_dedup", "total_removed": 5, "chunk_count": 3 }
  ],
  "system_prompt": "You are an entity extraction assistant...",
  "extraction_rules_template": "...",
  "entity_templates": "...",
  "relationship_templates": "...",
  "domain_guidance": "...",
  "domain_examples": "..."
}

Status	Description
`404`	No extraction statistics available for this source

Get Extraction Chart Data

GET /api/v1/sources/{source_id}/extraction/charts

Returns all extraction tasks with minimal fields for UI chart rendering. No pagination -- returns all tasks at once for efficient charting.

curl http://localhost:8080/api/v1/sources/src_abc123/extraction/charts

Parameter	Type	Required	Description
`source_id`	string (path)	Yes	Source ID

Response 200 OK

[
  {
    "chunk_index": 0,
    "status": "completed",
    "retry_count": 0,
    "entity_count": 8,
    "relationship_count": 12,
    "input_text_length": 3200,
    "llm_duration_ms": 3800
  },
  {
    "chunk_index": 1,
    "status": "completed",
    "retry_count": 1,
    "entity_count": 6,
    "relationship_count": 9,
    "input_text_length": 2800,
    "llm_duration_ms": 5200
  }
]

Get Cross-Chunk Filtering Log

GET /api/v1/sources/{source_id}/extraction/filteringlog

Returns the cross-chunk deduplication filtering log from the post-extraction merging stage. Shows entities and relationships removed during structural filtering, exact/semantic deduplication, and relationship deduplication.

curl http://localhost:8080/api/v1/sources/src_abc123/extraction/filteringlog

Parameter	Type	Required	Description
`source_id`	string (path)	Yes	Source ID

Response 200 OK

{
  "stages": [
    {
      "stage": "exact_dedup",
      "removed_count": 3,
      "details": ["Entity 'Python' duplicate removed", "..."]
    }
  ],
  "total_removed": 5
}

Status	Description
`404`	Source not found or no filtering log available

Retry Errored Source

POST /api/v1/sources/{source_id}/retry

Manually retry a source that is in error status. The retry target is determined by error_stage on the source record — the service routes the source back to the appropriate pipeline stage (indexing, extraction, or commit).

curl -X POST http://localhost:8080/api/v1/sources/src_abc123/retry

Parameter	Type	Required	Description
`source_id`	string (path)	Yes	Source ID

Response 200 OK -- SourceResponse

Returns the updated source record after the retry has been queued.

Status	Description
`404`	Source not found
`409`	Source is not in `error` status or `error_stage` is unknown

Re-extract Source

POST /api/v1/sources/{source_id}/re_extract

Manually re-run entity extraction on a source (distinct from retry). The key difference:

Retry preserves the cached extraction payload and re-runs only the failed stage (cheap — no additional LLM tokens for commit-only retries).
Re-extract discards the cached payload and any previous extraction results, resets the source to indexed, and re-runs the full LLM extraction (expensive — costs LLM tokens).

Use this when you want to re-analyze a document after changing the extraction domain, fixing domain-specific rules, or correcting the initial extraction output.

curl -X POST http://localhost:8080/api/v1/sources/src_abc123/re_extract

Parameter	Type	Required	Description
`source_id`	string (path)	Yes	Source ID

Response 202 Accepted -- SourceResponse

Returns the updated source record after the re-extraction job has been queued.

Allowed source states:

committed — Atomically deletes graph artifacts (nodes, edges, templates), resets to indexed, and re-extracts.
error (after post-INDEXING stage) — Resets to indexed, clears cached payload, and re-extracts.
indexed / extracted / extracting / mcp_extracting / committing — Forcibly resets to indexed, clears payload, and re-extracts.

Rejected source states:

pending / indexing — Returns 422; the source has not yet produced extraction artifacts. Wait for indexing to complete, then retry.

Persisted settings carry over

Re-extract reuses the source's persisted upload settings (auto_analyze, enable_normalization, enable_vision, content_filtering, filtering_mode) by default — what you uploaded with is what you re-extract with. Clients can override any of these per call by passing them in the request body.

force_re_extract also resets every quality counter on the source row back to zero and clears vector_indexing_status to pending, so the new run starts with a clean counter set.

Status	Description
`404`	Source not found
`422`	Source is in `pending` or `indexing` state (not yet indexable)

List Source Recovery Events

GET /api/v1/sources/{source_id}/recovery_events

Returns the recovery audit trail for a source — every automatic recovery attempt, what was dispatched, and when. Backs the source detail page's recovery panel so operators can diagnose repeated failures without grepping container logs. Events are returned newest first.

curl "http://localhost:8080/api/v1/sources/src_abc123/recovery_events?limit=20"

Parameter	Type	Required	Default	Description
`source_id`	string (path)	Yes	--	Source ID
`limit`	int (query)	No	`50`	Maximum events to return (1--200)

note

Recovery events use a ?limit= cap (max 200) rather than the standard ?page=&page_size= pagination model — events are an audit trail, not a paged resource collection.

Response 200 OK

{
  "events": [
    {
      "id": "rev_001",
      "source_id": "src_abc123",
      "event_type": "recovery",
      "created_at": "2026-03-09T10:05:00",
      "reason": "Stalled extraction detected",
      "dispatched_operation": "OP_EXTRACT_SOURCE"
    }
  ]
}

Status	Description
`404`	Source not found

Cleanup Orphan Chunk Tasks

POST /api/v1/sources/cleanup/orphan_tasks

Triggers an immediate sweep of orphaned chunk tasks — tasks whose parent extraction job completed or failed but whose rows were not updated. Normally run automatically on a schedule; use this endpoint to trigger it on demand after bulk operations or during recovery.

curl -X POST http://localhost:8080/api/v1/sources/cleanup/orphan_tasks

Response 200 OK

{
  "deleted_count": 12,
  "retention_days": 7
}

Field	Type	Description
`deleted_count`	int	Number of orphaned task rows removed
`retention_days`	int	Configured orphan retention window (from `SourceRecoverySettings`)

Abort All Processing

DELETE /api/v1/sources/{source_id}/processing

Cancels all queued/running tasks (indexing or extraction) and resets the source status appropriately.

curl -X DELETE http://localhost:8080/api/v1/sources/src_abc123/processing

Parameter	Type	Required	Description
`source_id`	string (path)	Yes	Source ID

Response 204 No Content

Status	Description
`400`	Source is not in a processing state
`404`	Source not found

Status after abort

pending / indexing -> error (with message "Processing/Indexing aborted by user")
extracting -> indexed (RAG still usable)
committing -> extracted

Chunks

List Chunks

GET /api/v1/sources/{source_id}/chunks

Paginated list of document chunks for a source.

curl "http://localhost:8080/api/v1/sources/src_abc123/chunks?page=1&page_size=20"

Parameter	Type	Required	Default	Description
`source_id`	string (path)	Yes	--	Source ID
`page`	int	No	`1`	Page number (1-indexed)
`page_size`	int	No	Server default	Items per page
`status`	string	No	`null`	Filter by chunk status

Response 200 OK -- ChunkListResponse

{
  "chunks": [
    {
      "id": "chunk_001",
      "source_id": "src_abc123",
      "chunk_index": 0,
      "content": "This is the first chunk of the document...",
      "page_number": 1,
      "section": "Introduction",
      "group_index": 0,
      "status": "indexed",
      "created_at": "2026-03-09T10:00:05"
    }
  ],
  "total": 42,
  "page": 1,
  "page_size": 20
}

Get Chunk

GET /api/v1/sources/{source_id}/chunks/{chunk_id}

Returns a single chunk by ID.

curl http://localhost:8080/api/v1/sources/src_abc123/chunks/chunk_001

Parameter	Type	Required	Description
`source_id`	string (path)	Yes	Source ID
`chunk_id`	string (path)	Yes	Chunk ID

Response 200 OK -- ChunkResponse

{
  "id": "chunk_001",
  "source_id": "src_abc123",
  "chunk_index": 0,
  "content": "This is the first chunk of the document...",
  "page_number": 1,
  "section": "Introduction",
  "group_index": 0,
  "status": "indexed",
  "created_at": "2026-03-09T10:00:05"
}

Status	Description
`404`	Chunk not found or does not belong to this source

Citations

List Citations

GET /api/v1/sources/{source_id}/citations

Paginated list of entity citations (attributions) for a source. Each citation links an extracted entity back to the source chunk it was found in.

curl "http://localhost:8080/api/v1/sources/src_abc123/citations?page=1&page_size=20"

Parameter	Type	Required	Default	Description
`source_id`	string (path)	Yes	--	Source ID
`page`	int	No	`1`	Page number (1-indexed)
`page_size`	int	No	Server default	Items per page

Response 200 OK -- CitationListResponse

{
  "citations": [
    {
      "id": "cit_001",
      "entity_uri": "urn:chaoscypher:node:abc123",
      "entity_label": "Python",
      "entity_type": "Programming Language",
      "source_id": "src_abc123",
      "chunk_id": "chunk_001",
      "confidence": 0.95,
      "extraction_method": "llm",
      "context_snippet": "...Python is a versatile programming language...",
      "created_at": "2026-03-09T10:05:00"
    }
  ],
  "total": 85,
  "page": 1,
  "page_size": 20
}

Source Data Access

Get Source Stats

GET /api/v1/sources/{source_id}/stats

Returns computed statistics for a single source.

curl http://localhost:8080/api/v1/sources/src_abc123/stats

Parameter	Type	Required	Description
`source_id`	string (path)	Yes	Source ID

Response 200 OK

{
  "chunk_count": 42,
  "citation_count": 85,
  "entity_count": 65,
  "relationship_count": 120,
  "total_content_length": 52000,
  "avg_chunk_length": 1238
}

Status	Description
`404`	Source not found

Get Source Entities

GET /api/v1/sources/{source_id}/entities

Paginated list of entities extracted from the document. Each entity includes a computed quality_score (0-100).

curl "http://localhost:8080/api/v1/sources/src_abc123/entities?page=1&page_size=20&sort_by=quality&sort_order=desc"

Parameter	Type	Required	Default	Description
`source_id`	string (path)	Yes	--	Source ID
`page`	int	No	`1`	Page number (1-indexed)
`page_size`	int	No	Server default	Items per page
`sort_by`	string	No	`default`	Sort field: `default`, `quality`, `confidence`, `name`, `type`
`sort_order`	string	No	`desc`	Sort direction: `asc` or `desc`

Response 200 OK

{
  "entities": [
    {
      "name": "Python",
      "type": "ProgrammingLanguage",
      "confidence": 0.95,
      "description": "A versatile programming language",
      "source_chunks": [0, 3, 7],
      "quality_score": 92.5
    }
  ],
  "pagination": {
    "page": 1,
    "page_size": 20,
    "total": 85,
    "total_pages": 5,
    "has_next": true,
    "has_prev": false
  }
}

Status	Description
`404`	Source not found

Get Source Relationships

GET /api/v1/sources/{source_id}/relationships

Paginated list of relationships extracted from the document. Each relationship is enriched with human-readable from and to entity names.

curl "http://localhost:8080/api/v1/sources/src_abc123/relationships?page=1&page_size=20"

Parameter	Type	Required	Default	Description
`source_id`	string (path)	Yes	--	Source ID
`page`	int	No	`1`	Page number (1-indexed)
`page_size`	int	No	Server default	Items per page

Response 200 OK

{
  "relationships": [
    {
      "source": 0,
      "target": 5,
      "type": "USES",
      "confidence": 0.88,
      "from": "FastAPI",
      "to": "Python"
    }
  ],
  "pagination": {
    "page": 1,
    "page_size": 20,
    "total": 120,
    "total_pages": 6,
    "has_next": true,
    "has_prev": false
  }
}

Status	Description
`404`	Source not found

Get Source Templates

GET /api/v1/sources/{source_id}/templates

Paginated list of graph templates created from extraction of this source.

curl "http://localhost:8080/api/v1/sources/src_abc123/templates?page=1&page_size=20&template_type=node"

Parameter	Type	Required	Default	Description
`source_id`	string (path)	Yes	--	Source ID
`template_type`	string	No	`null`	Filter by type: `node` or `edge`
`page`	int	No	`1`	Page number (1-indexed)
`page_size`	int	No	Server default	Items per page

Response 200 OK

{
  "templates": [
    {
      "id": "template_abc123",
      "name": "ProgrammingLanguage",
      "type": "node",
      "source_id": "src_abc123",
      "properties": ["name", "paradigm", "version"]
    }
  ],
  "pagination": {
    "page": 1,
    "page_size": 20,
    "total": 12,
    "total_pages": 1,
    "has_next": false,
    "has_prev": false
  }
}

Status	Description
`404`	Source not found

Get Source LLM Metrics

GET /api/v1/sources/{source_id}/llm_metrics

Summary of LLM usage metrics for a source, including call counts, token consumption, cost estimates, and derived rates.

curl http://localhost:8080/api/v1/sources/src_abc123/llm_metrics

Parameter	Type	Required	Description
`source_id`	string (path)	Yes	Source ID

Response 200 OK

{
  "source_id": "src_abc123",
  "has_metrics": true,
  "summary": {
    "total_calls": 6,
    "successful_calls": 6,
    "failed_calls": 0,
    "retry_calls": 1,
    "first_try_successes": 5,
    "retry_successes": 1,
    "permanent_failures": 0,
    "total_input_tokens": 24000,
    "total_output_tokens": 8500,
    "wasted_tokens": 400,
    "avg_call_duration_ms": 4200,
    "total_duration_ms": 25200,
    "estimated_cost_usd": 0.0325,
    "error_counts": {},
    "model": "gpt-4o",
    "success_rate": 1.0,
    "retry_rate": 0.167,
    "waste_percentage": 0.012
  }
}

Status	Description
`404`	Source not found

List Source LLM Calls

GET /api/v1/sources/{source_id}/llm_metrics/calls

Paginated list of individual LLM API calls made during extraction of this source.

curl "http://localhost:8080/api/v1/sources/src_abc123/llm_metrics/calls?page=1&page_size=20&success=true"

Parameter	Type	Required	Default	Description
`source_id`	string (path)	Yes	--	Source ID
`page`	int	No	`1`	Page number (1-indexed)
`page_size`	int	No	Server default	Items per page
`success`	bool	No	`null`	Filter by success status

Response 200 OK

{
  "calls": [
    {
      "id": "call_001",
      "source_id": "src_abc123",
      "success": true,
      "input_tokens": 1100,
      "output_tokens": 620,
      "duration_ms": 3800,
      "model": "gpt-4o",
      "created_at": "2026-03-09T10:00:05"
    }
  ],
  "pagination": {
    "page": 1,
    "page_size": 20,
    "total": 6,
    "total_pages": 1,
    "has_next": false,
    "has_prev": false
  }
}

Status	Description
`404`	Source not found

Source Tag Assignment

GET /api/v1/sources/{source_id}/tags

Returns all tags assigned to a specific source.

curl http://localhost:8080/api/v1/sources/src_abc123/tags

Parameter	Type	Required	Description
`source_id`	string (path)	Yes	Source ID

Response 200 OK -- list[TagResponse]

[
  {
    "id": "tag_001",
    "database_name": "default",
    "name": "Research",
    "color": "#4dabf5",
    "description": "Research papers",
    "created_at": "2026-03-01T08:00:00"
  }
]

Assign Tag to Source

POST /api/v1/sources/{source_id}/tags/{tag_id}

Assign a tag to a source.

curl -X POST http://localhost:8080/api/v1/sources/src_abc123/tags/tag_001

Parameter	Type	Required	Description
`source_id`	string (path)	Yes	Source ID
`tag_id`	string (path)	Yes	Tag ID

Response 204 No Content

Status	Description
`404`	Source or tag not found

Remove Tag from Source

DELETE /api/v1/sources/{source_id}/tags/{tag_id}

Remove a tag from a source.

curl -X DELETE http://localhost:8080/api/v1/sources/src_abc123/tags/tag_001

Parameter	Type	Required	Description
`source_id`	string (path)	Yes	Source ID
`tag_id`	string (path)	Yes	Tag ID

Response 204 No Content

Status	Description
`404`	Tag assignment not found

Page Images

Rendered page images are generated for PDF sources when vision processing is enabled. Images are stored per-source under data/databases/{db_name}/images/{source_id}/ and served directly by the API.

List Source Images

GET /api/v1/sources/{source_id}/images

Returns a list of available rendered page images for a source document. Returns an empty list if no images have been generated.

curl http://localhost:8080/api/v1/sources/src_abc123/images

Parameter	Type	Required	Description
`source_id`	string (path)	Yes	Source ID

Response 200 OK -- list[object]

[
  {
    "filename": "page_1.png",
    "url": "/sources/src_abc123/images/page_1.png"
  },
  {
    "filename": "page_2.png",
    "url": "/sources/src_abc123/images/page_2.png"
  }
]

Images are sorted by filename. The url field is the path to pass to the Get Source Image endpoint.

Get Source Image

GET /api/v1/sources/{source_id}/images/{filename}

Serve a specific rendered page image. Returns the image as image/png. Path traversal is prevented -- the resolved path must remain within the source image directory.

curl http://localhost:8080/api/v1/sources/src_abc123/images/page_1.png \
  --output page_1.png

Parameter	Type	Required	Description
`source_id`	string (path)	Yes	Source ID
`filename`	string (path)	Yes	Image filename (e.g. `page_1.png`)

Response 200 OK -- PNG image binary (Content-Type: image/png)

Status	Description
`403`	Path traversal attempt detected
`404`	Image file not found

Response Models

SourceResponse

Full source detail model returned by get, create, and update endpoints. Contains all lifecycle fields across indexing, extraction, commit, and LLM metrics stages.

Field	Type	Description
`id`	string	Source ID
`database_name`	string	Database this source belongs to
`filename`	string	Original filename
`filepath`	string?	Storage path
`file_type`	string?	File type (e.g. `pdf`, `csv`)
`file_size`	int?	File size in bytes
`title`	string?	Display title
`source_type`	string?	Source type (e.g. `pdf`, `webpage`)
`origin_url`	string?	Original URL for web imports
`version`	int	Version number (default `1`)
`parent_id`	string?	Parent source ID
`status`	string	Lifecycle status: `pending` \| `indexing` \| `indexed` \| `extracting` \| `extracted` \| `committing` \| `committed` \| `error`
`enabled`	bool	Whether the source is active
`error_message`	string?	Error description if status is `error`
`error_stage`	string?	Stage where error occurred
`chunk_count`	int	Number of chunks created
`total_content_length`	int	Total character count of all chunks
`embedding_model`	string?	Embedding model used
`embedding_dimensions`	int?	Embedding vector dimensions
`indexing_started_at`	datetime?	When indexing started
`indexing_completed_at`	datetime?	When indexing finished
`indexing_duration_seconds`	float?	Calculated indexing duration
`extraction_depth`	string?	Extraction depth (`full` or `quick`)
`extraction_entities_count`	int	Entities extracted
`extraction_relationships_count`	int	Relationships extracted
`extraction_domain`	string?	Domain used for extraction
`extraction_domain_auto`	bool	Whether domain was auto-detected
`extraction_started_at`	datetime?	When extraction started
`extraction_completed_at`	datetime?	When extraction finished
`extraction_duration_seconds`	float?	Calculated extraction duration
`current_extraction_job_id`	string?	Active extraction job ID
`commit_started_at`	datetime?	When commit started
`commit_completed_at`	datetime?	When commit finished
`commit_duration_seconds`	float?	Calculated commit duration
`commit_nodes_created`	int	Graph nodes committed
`commit_edges_created`	int	Graph edges committed
`commit_templates_created`	int	Templates created
`current_step`	int?	Current processing step number
`total_steps`	int?	Total processing steps
`step_description`	string?	Current step label
`llm_total_calls`	int	Total LLM API calls
`llm_successful_calls`	int	Successful LLM calls
`llm_failed_calls`	int	Failed LLM calls
`llm_retry_calls`	int	Retried LLM calls
`llm_first_try_successes`	int	Calls that succeeded on first attempt
`llm_retry_successes`	int	Calls that succeeded after retry
`llm_permanent_failures`	int	Calls that permanently failed
`llm_total_input_tokens`	int	Total input tokens consumed
`llm_total_output_tokens`	int	Total output tokens generated
`llm_wasted_tokens`	int	Tokens wasted on failed calls
`llm_avg_call_duration_ms`	int?	Average call duration in ms
`llm_total_duration_ms`	int	Total LLM call duration in ms
`llm_estimated_cost_usd`	float?	Estimated cost in USD
`llm_error_counts`	object?	Error type breakdown
`llm_model`	string?	LLM model used
`created_at`	datetime	Creation timestamp
`updated_at`	datetime	Last update timestamp
`user_metadata`	object?	User-defined metadata
`upload_options`	object	Persisted upload settings — see UploadOptions
`quality_metrics`	object	Per-stage quality counters and loader/search status — see QualityMetrics. Full reference: Quality Metrics API.
`vector_indexing_status`	string	One of `pending`, `indexed`, `degraded`, `failed`. Mirrored at the top level for convenience; also lives inside `quality_metrics`. See Search Status.

UploadOptions

The settings the user (or default) supplied at upload time, persisted on the source row so recovery, retry, and re-extract honor them without the client having to re-pass them.

Field	Type	Description
`auto_analyze`	bool	Auto-queue extraction after indexing finishes
`enable_normalization`	bool?	`null` = use file-type default; `true`/`false` = user override
`enable_vision`	bool	Use the vision model on images and scanned PDFs
`content_filtering`	bool	Apply domain content-exclusion rules during extraction
`filtering_mode`	string	`unfiltered` / `minimal` / `lenient` / `balanced` / `strict` / `maximum`

QualityMetrics

The 15 quality counters plus three companion fields. Counters reset to zero on Re-extract (force_re_extract); the quality grade itself is not affected.

Field	Type	Description
`loader_encoding_used`	string?	Encoding the loader actually used (`utf-8`, `cp1252`, `latin-1-fallback`, etc.). `null` until the loader runs.
`loader_warnings_count`	int	Non-fatal loader hiccups (e.g. a single bad JSONL line)
`loader_files_skipped`	int	Archive entries skipped (unsupported / oversized / security violation)
`cleaner_lines_removed`	int	Lines dropped by the OCR cleaner (gibberish, page numbers, repeated headers)
`cleaner_paragraphs_deduplicated`	int	Duplicate paragraphs collapsed by the cleaner
`cleaner_chars_removed`	int	Net character delta from text cleaning (encoding fixes, control chars, whitespace)
`chunks_filtered_count`	int	Chunks the content-filter stripped to under 100 chars before extraction
`llm_chunks_truncated`	int	Chunks where the LLM hit its token cap
`llm_chunks_aborted_by_loop`	int	Chunks where the streaming loop detector aborted the LLM
`parser_lines_dropped`	int	Malformed `E\|`/`R\|`/`P\|` lines the parser couldn't make sense of
`dedup_entities_merged`	int	Entities collapsed into another entity by exact-name or semantic dedup
`structural_entities_filtered`	int	Entities representing document structure removed by the structural filter
`orphan_entities_filtered`	int	Entities with zero relationships dropped at commit time
`relationships_dropped_invalid`	int	Relationships pointing at non-existent entity indices
`relationships_dropped_capped`	int	Relationships dropped by the per-entity / same-source-type / total-ratio caps
`citations_skipped_no_chunk_index`	int	Citations skipped at commit because the underlying entity / relationship had no chunk index
`vector_indexed_at`	datetime?	When the vector indexing call succeeded; `null` if not yet
`vector_indexing_status`	string	`pending`, `indexed`, `degraded`, or `failed`

SourceSummaryResponse

Lightweight model used in list views. Excludes large payload fields like user_metadata, detailed timestamps, and full LLM metrics. Includes a tags array enriched at the API layer.

Refer to the List Sources response example for the full shape.

PaginatedSourcesResponse

Pagination wrapper for source list responses.

Field	Type	Description
`data`	SourceSummaryResponse[]	Page of source summaries
`pagination`	object	Pagination metadata (total, page, page_size, total_pages, has_next, has_prev)

ChunkResponse

Field	Type	Description
`id`	string	Chunk ID
`source_id`	string?	Parent source ID
`chunk_index`	int	Position in the document (0-indexed)
`content`	string	Chunk text content
`page_number`	int?	PDF page number
`section`	string?	Section heading
`group_index`	int?	Hierarchical group index
`status`	string	Chunk status
`created_at`	datetime	Creation timestamp

ChunkListResponse

Field	Type	Description
`chunks`	ChunkResponse[]	Page of chunks
`total`	int	Total chunks for this source
`page`	int	Current page number
`page_size`	int	Items per page

CitationResponse

Field	Type	Description
`id`	string	Citation ID
`entity_uri`	string	URI of the cited entity
`entity_label`	string	Human-readable entity name
`entity_type`	string?	Entity template name
`source_id`	string	Source this citation belongs to
`chunk_id`	string	Chunk where the entity was found
`confidence`	float	Extraction confidence (0.0--1.0)
`extraction_method`	string	How the entity was extracted (e.g. `llm`)
`context_snippet`	string?	Text snippet surrounding the entity mention
`created_at`	datetime	Creation timestamp

CitationListResponse

Field	Type	Description
`citations`	CitationResponse[]	Page of citations
`total`	int	Total citations for this source
`page`	int	Current page number
`page_size`	int	Items per page

ExtractionTaskResponse

Field	Type	Description
`id`	string	Task ID
`job_id`	string	Parent extraction job ID
`chunk_index`	int	Chunk group index
`hierarchical_group_id`	string?	Group identifier for hierarchical chunking
`small_chunk_ids`	string[]?	Individual chunk IDs in this group
`status`	string	`pending` \| `queued` \| `running` \| `completed` \| `failed`
`created_at`	datetime	Creation timestamp
`queued_at`	datetime?	When queued for processing
`started_at`	datetime?	When LLM processing started
`completed_at`	datetime?	When processing finished
`llm_duration_ms`	int?	LLM call duration in ms
`retry_count`	int	Number of retries
`entity_count`	int	Entities extracted by this task
`relationship_count`	int	Relationships extracted
`invalid_relationship_count`	int	Invalid relationships filtered out
`small_chunk_numbers`	int[]?	1-indexed chunk numbers for UI display
`input_text_length`	int?	Input text character count
`llm_response_length`	int?	LLM response character count
`input_tokens`	int?	Actual input token count from LLM API
`output_tokens`	int?	Actual output token count from LLM API
`context_window_available`	int?	Model context window size
`input_text`	string?	Full input text (only in detail view or when `include_content=true`)
`llm_response_json`	string?	Raw LLM JSON response (only in detail view or when `include_content=true`)
`filtering_log`	object?	Per-chunk filtering diagnostics (detail view only)
`finish_reason`	string?	Normalized provider finish reason: `stop`, `length`, `content_filter`, `tool_calls`, `error`, `unknown`. `null` for tasks that predate the field (migration 0022).
`aborted_by_loop`	bool?	`true` when the streaming loop detector aborted the LLM mid-response. Tasks predating migration 0022 carry `null`.
`error_message`	string?	Error message if failed
`error_type`	string?	Error classification

ExtractionTaskListResponse

Field	Type	Description
`tasks`	ExtractionTaskResponse[]	Page of extraction tasks
`total`	int	Total tasks for this source
`page`	int	Current page number
`page_size`	int	Items per page

ExtractionTaskStatsResponse

Aggregate statistics computed via SQL aggregates across all extraction tasks.

Field	Type	Description
`total_tasks`	int	Total extraction tasks
`context_window`	int?	LLM context window size
`min_input_tokens`	int?	Minimum input tokens across tasks
`max_input_tokens`	int?	Maximum input tokens
`avg_input_tokens`	int?	Average input tokens
`min_output_tokens`	int?	Minimum output tokens
`max_output_tokens`	int?	Maximum output tokens
`avg_output_tokens`	int?	Average output tokens
`min_total_tokens`	int?	Minimum total tokens (input + output)
`max_total_tokens`	int?	Maximum total tokens
`avg_total_tokens`	int?	Average total tokens
`min_utilization`	float?	Minimum context window utilization %
`max_utilization`	float?	Maximum utilization %
`avg_utilization`	float?	Average utilization %
`min_duration_ms`	int?	Minimum LLM call duration in ms
`max_duration_ms`	int?	Maximum duration
`avg_duration_ms`	int?	Average duration
`total_entities`	int	Total entities across all tasks
`avg_entities_per_task`	float	Average entities per task
`total_relationships`	int	Total relationships
`avg_relationships_per_task`	float	Average relationships per task
`total_retries`	int	Total retry attempts
`max_retries_single_task`	int	Most retries on a single task
`total_invalid_relationships`	int	Total invalid relationships filtered
`avg_invalid_per_task`	float	Average invalid relationships per task
`total_entities_filtered`	int	Entities removed by pipeline filtering
`total_relationships_filtered`	int	Relationships removed by filtering
`filtering_stage_summary`	object[]?	Per-stage filtering breakdown
`system_prompt`	string?	System prompt used for extraction
`extraction_rules_template`	string?	Extraction rules portion of the prompt
`entity_templates`	string?	Entity template portion
`relationship_templates`	string?	Relationship template portion
`domain_guidance`	string?	Domain-specific guidance
`domain_examples`	string?	Domain-specific examples

TagResponse

Field	Type	Description
`id`	string	Tag ID
`database_name`	string	Database this tag belongs to
`name`	string	Tag display name
`color`	string?	Hex color code
`description`	string?	Tag description
`created_at`	datetime	Creation timestamp

Status	Description
`400`	Duplicate tag name or validation error
`404`	Tag not found

Upload & Import​

Upload Single File​

Batch Upload​

Import from URL​

Source CRUD​

List Sources​

Get Source​

Update Source​

Delete Source​

Source Metadata​

List Extraction Domains​

Get Processing Stats​

Extraction Management​

Trigger Extraction​

Get Extraction Progress​

Cancel Extraction​

Reclassify Source Domain​

List Extraction Tasks​

Get Extraction Task​

Get Extraction Task Stats​

Get Extraction Chart Data​

Get Cross-Chunk Filtering Log​

Retry Errored Source​

Re-extract Source​

List Source Recovery Events​

Cleanup Orphan Chunk Tasks​

Abort All Processing​

Chunks​

List Chunks​

Get Chunk​

Citations​

List Citations​

Source Data Access​

Get Source Stats​

Get Source Entities​

Get Source Relationships​

Get Source Templates​

Get Source LLM Metrics​

List Source LLM Calls​

Tags​

List All Tags​

Get Tag​

Create Tag​

Update Tag​

Delete Tag​

Source Tag Assignment​

List Tags for Source​

Assign Tag to Source​

Remove Tag from Source​

Page Images​

List Source Images​

Get Source Image​

Response Models​

SourceResponse​

UploadOptions​

QualityMetrics​

SourceSummaryResponse​

PaginatedSourcesResponse​

ChunkResponse​

ChunkListResponse​

CitationResponse​

CitationListResponse​

ExtractionTaskResponse​

ExtractionTaskListResponse​

ExtractionTaskStatsResponse​

TagResponse​

Upload & Import

Upload Single File

Batch Upload

Import from URL

Source CRUD

List Sources

Get Source

Update Source

Delete Source

Source Metadata

List Extraction Domains

Get Processing Stats

Extraction Management

Trigger Extraction

Get Extraction Progress

Cancel Extraction

Reclassify Source Domain

List Extraction Tasks

Get Extraction Task

Get Extraction Task Stats

Get Extraction Chart Data

Get Cross-Chunk Filtering Log

Retry Errored Source

Re-extract Source

List Source Recovery Events

Cleanup Orphan Chunk Tasks

Abort All Processing

Chunks

List Chunks

Get Chunk

Citations

List Citations

Source Data Access

Get Source Stats

Get Source Entities

Get Source Relationships

Get Source Templates

Get Source LLM Metrics

List Source LLM Calls

Tags

List All Tags

Get Tag

Create Tag

Update Tag

Delete Tag

Source Tag Assignment

List Tags for Source

Assign Tag to Source

Remove Tag from Source

Page Images

List Source Images

Get Source Image

Response Models

SourceResponse

UploadOptions

QualityMetrics

SourceSummaryResponse

PaginatedSourcesResponse

ChunkResponse

ChunkListResponse

CitationResponse

CitationListResponse

ExtractionTaskResponse

ExtractionTaskListResponse

ExtractionTaskStatsResponse

TagResponse