Intelligence Engine — current public evidence record

Verification pin

Field	Value
Verified	2026-08-02
Private source repository	fbratten/intelligence-engine
Source branch	main
Source commit	3cdf7099c32c614a376a03324f545cfe8bc3a8b4
Public showcase repository	fbratten/intelligence-engine-showcase
Pre-refresh showcase head	6d577816b14517e7dd61c23c6ca6877da271414d
Product version stated by source	0.21.0

The source repository is private. This showcase publishes a bounded description, screenshots and demonstrations, not the private source tree or private indexed data.

Current source-supported capability summary

At the source pin above, the repository documents and implements a schema-driven intelligence engine with:

• three product domains: code, archaeology and written_content;
• one synthetic conformance_probe fixture that is deliberately not user-indexable;
• Python/FastAPI backend and React/TypeScript/Sigma.js frontend;
• Tree-sitter parsing for Python, JavaScript, TypeScript/TSX, Java, Go, HTML and CSS;
• KuzuDB graph storage with NetworkX as an alternate backend and parity oracle;
• LanceDB semantic storage and BM25 lexical retrieval;
• hybrid BM25, semantic and graph-context retrieval;
• project-scoped and shared multi-project storage modes;
• 14 MCP tools plus self-describing MCP resources;
• 46 REST endpoints across the current source surface;
• LLM-backed summaries and Q&A with provider and endpoint-class controls;
• incremental indexing, index collision prevention, preservation of AI overlays and cross-project analysis;
• a written_content extraction and originality-audit path for self-vs-self analysis.

The private source README reports 1261+ tests. That figure is recorded here as a source claim at the pinned commit. This showcase refresh did not re-run the private test suite.

Evidence classes

Claim	Evidence class	Public evidence
Schema-driven domains	Implemented and source-documented	Architecture and domain sections in this showcase
Code knowledge graph	Runtime-demonstrated in earlier showcase captures	Graph, Cypher and dashboard screenshots
Hybrid retrieval	Implemented and test-supported in private source	Architecture and search-selector demo
Web application	Runtime-demonstrated	Public screenshots and UI documentation
MCP integration	Implemented and test-supported in private source	Current tool inventory in this showcase
Multi-project/shared storage	Implemented and repeatedly repaired/verified in private source	Bounded description; no private estate data published
Written-content domain	Implemented and test-supported in private source	Public architecture description only in this refresh
Archaeology domain	Implemented MVP	Public schema and entity examples

Recruiter-readable product flow

structured source
-> domain extractor
-> domain-scoped entities and relationships
-> KuzuDB + LanceDB + BM25
-> hybrid retrieval and graph expansion
-> FastAPI / MCP
-> React graph explorer, dashboards and AI-assisted inspection

Important boundaries and non-claims

• The public showcase is not the private source repository.
• Screenshots may represent earlier runtime snapshots and are not claimed to show the exact current source pin pixel-for-pixel.
• written_content currently supports self-vs-self structural and originality auditing. External-source plagiarism checking is planned, not claimed as complete.
• The conformance_probe domain is a synthetic architecture fixture and is not a user-indexable product domain.
• The source test count is source-reported. This public refresh did not independently execute the private suite.
• The engine has been used on a private project estate, but private project names, source content, paths, graph data and credentials are not published here.
• Product counts are date- and commit-bound. They must not be treated as timeless marketing figures.

Freshness rule

Any future public update that changes domains, endpoint counts, MCP tool counts, product version or test claims must update this file with:

1. a new source commit pin;
2. a new verification date;
3. the exact source of each changed count;
4. explicit distinction between re-run evidence and source-reported evidence;
5. preservation of the prior evidence record in Git history.