How edvardgrishin27/frameproof works
frameproof turns a video or link into a searchable, time-coded index with explicit coverage gaps, then exposes separate commands for text search, frame retrieval, and arithmetic verification of citations. The main path is fetch/probe, transcript and screen analysis, coverage-aware frame selection, extraction/OCR, and index creation. The README also documents Kinescope support and token budgeting. Some subsystem links rely on documented workflow or the SOURCE INDEX; unsampled implementation details are not inferred.
Main entry points: __main__.py CLI commands · index.py Index and search · fetch.py Video fetch · extract.py Frame extraction
Components 15
Video ingestion
- Video fetch
frameproof/fetch.py - Kinescope loader
frameproof/kinescope.py - Video probe
frameproof/probe.py
Media analysis
- Speech transcript
frameproof/transcribe.py - Screen-change signal
frameproof/analyze.py - Coverage selection
frameproof/select.py - Frame extraction
frameproof/extract.py - Screen OCR
frameproof/ocr.py
Index and retrieval
- CLI commands
frameproof/__main__.py - Index and search
frameproof/index.py
Evidence and cost
- Coverage report
frameproof/report.py - Citation verification
frameproof/verify.py - Token budgeting
frameproof/budget.py
Other components
- User or agent
- Video hosts
Connections 23
- User or agent → CLI commands: runs commands
README.md - CLI commands → Video fetch: fetches media
frameproof/__main__.py - Video fetch → Kinescope loader: routes Kinescope
frameproof/fetch.py - Video fetch → Video hosts: downloads media
README.md - CLI commands → Video probe: probes video
frameproof/__main__.py - CLI commands → Speech transcript: loads subtitles
frameproof/__main__.py - CLI commands → Screen-change signal: analyzes screen
frameproof/__main__.py - Screen-change signal → Video probe: uses video metadata
frameproof/analyze.py - CLI commands → Coverage selection: selects frames
frameproof/__main__.py - Coverage selection → Screen-change signal: uses change signal
frameproof/select.py - CLI commands → Frame extraction: extracts selected frames
frameproof/__main__.py - Frame extraction → Video probe: uses video metadata
frameproof/extract.py - Frame extraction → Token budgeting: sizes frames
frameproof/extract.py - CLI commands → Screen OCR: recognizes screen text
frameproof/__main__.py - Screen OCR → Index and search: adds OCR text
frameproof/ocr.py - CLI commands → Index and search: writes or searches index
frameproof/__main__.py - CLI commands → Coverage report: reports coverage
frameproof/__main__.py - Coverage report → Coverage selection: reports selection
frameproof/report.py - Coverage report → Token budgeting: estimates tokens
frameproof/report.py - CLI commands → Citation verification: checks citations
frameproof/__main__.py - Citation verification → Index and search: reads indexed evidence
frameproof/verify.py - User or agent → Index and search: searches speech and OCR
README.md - User or agent → CLI commands: requests specific frames
README.md