Deep Dive
Full-Spectrum Capabilities. Zero Compromise.
Every feature designed for one purpose: make your AI agents impossibly smart about your codebase.
All Core Capabilities
🧠 Memory Core
The foundation of intelligent recall. Every observation is classified, indexed, and instantly retrievable.
- →Semantic memory search with relevance scoring
- →Auto-conflict detection before saving
- →Bulk observation capture (up to 10 at once)
- →Full project context dump — tech stack, hot files, activity
- →Critical warning and gotcha retrieval
- →Convention and pattern surfacing
- →Project brief generation — language, framework, architecture
- →Memory health statistics and session tracking
- →Full-text search across all observations
- →Observation forgetting for outdated knowledge
- →Duplicate detection with similarity scoring
- →Smart save — detect + resolve + store in one action
- →Active file-aware context routing
🕸️ Knowledge Graph
Entity-relationship mapping that understands how your code connects.
- →Graph node creation for code entities
- →Entity relationship mapping (directional links)
- →Observation enrichment on existing entities
- →O(1) instant node retrieval by name
🔍 Code Intelligence
Offline semantic code search, real-time diagnostics, and safe structural edits.
- →Offline ONNX-powered semantic code search
- →Vector codebase indexing (384-dim embeddings)
- →Safe AST-aware block replacement
- →Regex symbol search across project directories
- →Real-time compiler and linter diagnostics
- →Uncommitted diff synchronization
- →Line-range file reading for context preservation
- →Project-wide scan with progress tracking
📊 Analytics & Briefings
Know exactly how your agents are doing. Auto-generated reports, zero effort.
- →Auto-generated daily activity briefings
- →Agent capability utilization metrics
- →Session activity and efficiency tracking
🌍 Global & Cross-Project
Knowledge that transcends a single project. Universal intelligence across your entire codebase.
- →Cross-project knowledge sharing
- →Global semantic recall across all projects
- →Permanent observation pinning (bookmarks)
- →Bookmarked observation retrieval
- →Local git tree awareness (branch, commits, diffs)
🛠️ Utilities & Safety
Backup, rollback, free AI offloading, and external documentation — all built in.
- →Instant memory snapshots (local backups)
- →Checkpoint listing and rollback
- →Free LLM reasoning offload ($0 AI thinking)
- →Free code analysis via cloud models ($0)
- →External documentation resolution
- →Up-to-date library docs fetching
- →Auto-skill compilation from patterns
- →Skill enable/disable management
- →Automated memory compaction (old → summaries)
Observation Engine™
9 Semantic Categories. Zero-Cost Classification.
what-changed
Tracks file modifications and refactors
problem-fix
Error→fix pairs for instant recall
gotcha
Critical warnings — never auto-pruned
decision
Architecture and design choices
trade-off
Documented compromises and rationale
how-it-works
System behavior and internals
convention
Project coding standards
discovery
New patterns and insights
tool-pattern
Recurring tool usage workflows
Intelligence Cortex