Spaces:

akshayyy1
/

vector-auditor

Running

App Files Files Community

vector-auditor / src /api

Commit History

Oct 7 8 days ago

12:11
feat: 10 free upgrades — streaming upload, cache invalidation, hybrid search, ONNX reranker, SQLAlchemy pool tuning, batch embed cache, semantic chunking, Prometheus histograms, DLQ auto-retry, structured logging+OTel

AkshayM

3e047b8

Jun 16

10:56
Add max_tokens parameter to NexAGI endpoint

AkshayM

f89f0c8
08:44
Add privacy column to Document model + migration 005

AkshayM

262e6d5
08:22
Remove PII from research mode and response handlers

AkshayM

390b099
08:18
Tune Presidio: strict mode masks names+orgs for NDA, light mode skips them for research

AkshayM

073907c
08:15
Add privacy flag to upload — PII runs only when privacy=True (NDA mode)

AkshayM

82a91b9
07:39
Fix tuple unpacking error in gather result for PDF parsing

AkshayM

d3f3dec
07:31
Skip MarkItDown for PDFs — run only as fallback if pypdf returns empty

AkshayM

e2a2578
07:17
Swap pdfplumber → pypdf for upload parsing (3x faster text extraction)

AkshayM

a965d17

Jun 15

19:19
Fix NexAGI: reasoning_details is already a list of dicts

AkshayM

6102a51
19:11
Fix NexAGI: reasoning_details list model_dump, add reasoning_details to message schema

AkshayM

9cb8ece
18:42
Add /NexAGI endpoint — OpenRouter reasoning via nex-agi/nex-n2-pro:free

AkshayM

464d909

Jun 13

20:42
feat: provider-based LLM profiles + per-request model switching⁠

AkshayM

9a38dd4
18:16
fix: remove stale page_texts_pdf references from PII block

AkshayM

e8e7520
11:58
fix: use pdfplumber text for chunking to ensure accurate page numbers for citations

AkshayM

247a9a5
11:16
fix: keep MarkItDown for chunking, pass pdfplumber page texts separately so page numbers align for bbox extraction

AkshayM

e01df47
11:11
fix: use pdfplumber text for chunking so page numbers align with page ranges (fixes bbox extraction)

AkshayM

4f4595a
09:56
fix: actually anonymize PII in document text before indexing + sanitize all LLM outputs

AkshayM

548bb13

Jun 12

07:38
fix: populate username from GitHub login on OAuth signup, return email/username in OAuth login response

AkshayM

4753a4e
06:49
feat: add username field to User model, schemas, and API

AkshayM

c58901b
06:35
perf: move PII detection from upload HTTP handler to job worker (runs concurrent with parsing)

AkshayM

6e20180

Jun 11

18:03
feat: harden prompt injection guardrails with 11 detection layers + fix missing guardrails on /analyze

AkshayM

4eb6f45

Jun 10

16:14
fix: add PII scan result log to upload endpoint

AkshayM

df2fe46
15:23
perf: revert to all-MiniLM-L6-v2, query embed cache, fire-and-forget Cloudinary, pre-warm reranker, reduce top-k

AkshayM

1476572

Jun 8

20:36
fix: import parse_document outside .pdf scope; coerce datetime→str in all response schemas

AkshayM

51bad61
19:12
production hardening: circuit breakers, graceful degradation, health checks, parallel uploads, HF Spaces deploy config

AkshayM

fc2fe9f

Jun 7

09:06
fix: remove local file fallback, use Cloudinary only, add POST /documents/backfill endpoint for old docs

AkshayM

1fd184a
09:00
fix: add file_url to every citation so the frontend PDF viewer gets the URL directly without extra API call

AkshayM

7226ce1
08:58
fix: cloudinary_url now falls back to local file_url so frontend PDF viewer always gets a working link

AkshayM

7edd24f
08:52
fix: add file_url to DocumentResponse, pass request to _doc_response, handle old-path files in serve_file

AkshayM

aae70f2
08:51
fix: Cloudinary upload resource_type=auto for PDF preview instead of raw download

AkshayM

a7baf2b
08:48
fix: Cloudinary upload_bytes->upload method name bug, add GET /files/{doc_id} endpoint for PDF serving fallback

AkshayM

3953df3
07:03
fix: enable NeMo Guardrails with correct parameters.api_key/parameters.base_url, add PII detection to guardrails flow, set PII_ENABLED=true

AkshayM

6ca471b
06:46
fix: handle LLM connection errors gracefully (SSL/connect errors → 502); convert Analyze 500 → 502 with clear message

AkshayM

5d8e007
06:39
fix: auto-fallback to Qdrant :memory: when remote unavailable

AkshayM

2636655
06:25
fix: load .env at module level in main.py (removes run.py dependency); auto-start worker + register processor on startup⁠

AkshayM

84839f7
06:20
fix: upload pipeline completely dead - worker never started, processor never registered⁠

AkshayM

2942460
06:09
fix: 15 critical/high bugs across codebase⁠

AkshayM

da3be01
05:54
fix upload pipeline: await add_document, pre-load model eagerly, add verbose UPLOAD/VectorStore logging; remove unused deployment configs

AkshayM

90c440b

Jun 6

22:08
fix models.py: match Neon schema varchar sizes, add email unique constraint

AkshayM

167c070

Commit History

Oct 7 8 days ago

feat: 10 free upgrades — streaming upload, cache invalidation, hybrid search, ONNX reranker, SQLAlchemy pool tuning, batch embed cache, semantic chunking, Prometheus histograms, DLQ auto-retry, structured logging+OTel

Jun 16

Add max_tokens parameter to NexAGI endpoint

Add privacy column to Document model + migration 005

Remove PII from research mode and response handlers

Tune Presidio: strict mode masks names+orgs for NDA, light mode skips them for research

Add privacy flag to upload — PII runs only when privacy=True (NDA mode)

Fix tuple unpacking error in gather result for PDF parsing

Skip MarkItDown for PDFs — run only as fallback if pypdf returns empty

Swap pdfplumber → pypdf for upload parsing (3x faster text extraction)

Jun 15

Fix NexAGI: reasoning_details is already a list of dicts

Fix NexAGI: reasoning_details list model_dump, add reasoning_details to message schema

Add /NexAGI endpoint — OpenRouter reasoning via nex-agi/nex-n2-pro:free

Jun 13

feat: provider-based LLM profiles + per-request model switching⁠

fix: remove stale page_texts_pdf references from PII block

fix: use pdfplumber text for chunking to ensure accurate page numbers for citations

fix: keep MarkItDown for chunking, pass pdfplumber page texts separately so page numbers align for bbox extraction

fix: use pdfplumber text for chunking so page numbers align with page ranges (fixes bbox extraction)

fix: actually anonymize PII in document text before indexing + sanitize all LLM outputs

Jun 12

fix: populate username from GitHub login on OAuth signup, return email/username in OAuth login response

feat: add username field to User model, schemas, and API

perf: move PII detection from upload HTTP handler to job worker (runs concurrent with parsing)

Jun 11

feat: harden prompt injection guardrails with 11 detection layers + fix missing guardrails on /analyze

Jun 10

fix: add PII scan result log to upload endpoint

perf: revert to all-MiniLM-L6-v2, query embed cache, fire-and-forget Cloudinary, pre-warm reranker, reduce top-k

Jun 8

fix: import parse_document outside .pdf scope; coerce datetime→str in all response schemas

production hardening: circuit breakers, graceful degradation, health checks, parallel uploads, HF Spaces deploy config

Jun 7

fix: remove local file fallback, use Cloudinary only, add POST /documents/backfill endpoint for old docs

fix: add file_url to every citation so the frontend PDF viewer gets the URL directly without extra API call

fix: cloudinary_url now falls back to local file_url so frontend PDF viewer always gets a working link

fix: add file_url to DocumentResponse, pass request to _doc_response, handle old-path files in serve_file

fix: Cloudinary upload resource_type=auto for PDF preview instead of raw download

fix: Cloudinary upload_bytes->upload method name bug, add GET /files/{doc_id} endpoint for PDF serving fallback

fix: enable NeMo Guardrails with correct parameters.api_key/parameters.base_url, add PII detection to guardrails flow, set PII_ENABLED=true

fix: handle LLM connection errors gracefully (SSL/connect errors → 502); convert Analyze 500 → 502 with clear message

fix: auto-fallback to Qdrant :memory: when remote unavailable

fix: load .env at module level in main.py (removes run.py dependency); auto-start worker + register processor on startup⁠

fix: upload pipeline completely dead - worker never started, processor never registered⁠

fix: 15 critical/high bugs across codebase⁠

fix upload pipeline: await add_document, pre-load model eagerly, add verbose UPLOAD/VectorStore logging; remove unused deployment configs

Jun 6

fix models.py: match Neon schema varchar sizes, add email unique constraint