complydoc
Architecture
Files, or the output of another framework's loader, are passed to four
independent analyses. A run writes an HTML and JSON report; extract_text returns masked text.
complydoc architecture
Files are read by per-format readers and loader output is taken as it is; both feed four independent analyses (cost, readiness, identifiers, hidden content), configured by YAML, which produce an HTML and JSON report and masked text, all inside a network guard that blocks outbound connections.
NETWORK GUARD
ANALYSIS
INPUT
documents/
pdf · img · docx · xlsx
INPUT
loader output
langchain · llamaindex
READ
Ingest
text layer · ocr
LOADER
Inspect
one or several loaders
01
Cost
tokens · prices
02
Readiness
measured signals
03
Identifiers
checksums · masked
04
Hidden content
visibility · instructions
YAML
Config
prices · signals · patterns
OUT
Report
html · json
OUT
Masked text
chunks · token counts
LEGEND
INPUT
STAGE
HIDDEN TEXT · INJECTION
CONFIG AND OUTPUT
NO OUTBOUND