Drafts CHANGELOG_PENDING.md entries from Conventional Commit subjects on
every push to main -- feat/fix/refactor/perf/security bucketed into
Added/Fixed/Changed, chore/ci/test/docs/style/merge skipped, anything
mentioning internal infra hostnames dropped. No LLM call, no network
access. Never touches CHANGELOG.md -- still needs a human to fold an
entry in. Rolled out fleet-wide from the PeerCortex pilot (2026-07-18).
- TRANSPORT=http starts HTTP server on PORT (default 3100)
- SSE endpoint at /sse, message endpoint at /messages
- Health check at /health
- TRANSPORT=stdio (default) unchanged for local Claude Code usage
- Fix: unused _format variable in matcher.ts
Add a simulated terminal recording as an animated SVG showing the core
PaperCortex workflow: docker startup, status check, semantic search,
receipt extraction, and DATEV export. Embedded in README above the
"What is PaperCortex?" section.
PaperCortex adds semantic search, auto-classification, receipt extraction,
bank statement matching, and DATEV export to Paperless-ngx — powered
entirely by local AI through Ollama. Exposes everything as an MCP Server
for Claude Code and AI agent integration.
- MCP Server with 5 tools (search, classify, receipt, query, export)
- Local Ollama embeddings for semantic document search
- Receipt data extraction (vendor, amount, date, tax, line items)
- DATEV Buchungsstapel CSV export for German accounting
- Bank CSV transaction matching
- Paperless-ngx REST API client
- Docker deployment
- Zero cloud dependencies — 100% self-hosted
