Changelog

0.7.0 — 2026-03-23

Knowledge MCP tools — get_rationale, find_owners, search_knowledge
Incremental ingestion — content-hash-based change detection, --incremental flag, --watch mode
Schema versioning and migrations — :Meta node versioning, navegador migrate CLI
Enhanced init — config.toml with storage, LLM, and cluster settings
Text-based graph export — deterministic JSONL format for git-friendly diffs
Editor integrations — MCP config generation for Claude Code, Cursor, Codex, Windsurf
CI/CD mode — navegador ci ingest/stats/check with JSON output, exit codes, GitHub Actions annotations
Python SDK — Navegador class wrapping all internal modules
Sensitive content detection — API key, password, token redaction before graph storage
VCS abstraction — GitAdapter and FossilAdapter with auto-detection
MCP security hardening — query validation, complexity limits, --read-only mode
Shell completions — bash, zsh, fish tab completion
LLM backend abstraction — unified provider interface for Anthropic, OpenAI, Ollama
AST optimizations — LRU tree cache, incremental re-parsing, graph node diffing, parallel ingestion

Language expansion — Kotlin, C#, PHP, Ruby, Swift, C, C++
FrameworkEnricher base class — auto-discovery, node promotion, semantic edges
Framework enrichers — Django, FastAPI, React/Next.js, Express.js, React Native, Rails, Spring Boot, Laravel
Monorepo support — Turborepo, Nx, Yarn, pnpm, Cargo, Go workspace detection
Git diff integration — map uncommitted changes to affected symbols and knowledge
Code churn correlation — git history analysis for behavioural coupling

Impact analysis — blast-radius traversal with MCP tool and CLI
Execution flow tracing — call chain precomputation from entry points
Dead code detection — unreachable functions, classes, and files
Test coverage mapping — link test functions to production code via TESTS edges
Circular dependency detection — DFS-based cycle detection in import and call graphs
Multi-repo support — register, ingest, and search across repositories
Coordinated rename — graph-assisted multi-file symbol refactoring with preview
CODEOWNERS integration — parse ownership files to Person and Domain nodes
ADR ingestion — MADR-format Architecture Decision Records
OpenAPI / GraphQL ingestion — API contract schemas as graph nodes
PlanOpticon pipeline — end-to-end meeting-to-knowledge with auto-linking
PM tool integration — GitHub issues ingestion (Linear/Jira stubs)
External dependency nodes — npm/pip/cargo package tracking
Fossil SCM support — full VCS implementation
Submodule traversal — parent + submodule linked ingestion
Multi-repo workspace — unified and federated knowledge graph modes

Semantic search — embedding-based similarity search with LLM providers
Community detection — label propagation over heterogeneous graph
LLM integration — natural language queries, community naming, documentation generation
Documentation generation — template and LLM-powered docs from graph context

Graph explorer — HTTP server with browser-based force-directed visualization
Test coverage — 96% coverage across 1902 tests

First public release.

7-language AST ingestion — Python, TypeScript, JavaScript, Go, Rust, Java via tree-sitter
Property graph storage — FalkorDB-lite (SQLite, zero-infra) or Redis-backed FalkorDB
Context bundles — file, function, class, concept, and explain context loading
MCP server — 7 tools for AI agent integration (ingest_repo, load_file_context, load_function_context, load_class_context, search_symbols, query_graph, graph_stats)
CLI — ingest, context, function, class, explain, search, decorated, query, stats, add, annotate, domain, concept, wiki ingest, planopticon ingest, mcp
Knowledge ingestion — concepts, rules, decisions, persons, domains, wiki pages, PlanOpticon video analysis outputs
Wiki ingestion — local Markdown directories, GitHub repo docs via API or git clone