KnowFrame — Knowledge Management Framework

The open-source knowledge governance compiler and dynamic graph for VS Code.
Prevents AI agent context window saturation, eliminates markdown drift, and slashes prompt token costs through deterministic quality scoring, BPE token budgeting, and automated structural evolution.
🚀 Why KnowFrame?
When modern developers and autonomous AI coding agents (Claude Code, Antigravity, Codex, Cursor) navigate extensive documentation repositories and markdown knowledge bases, four systemic failures emerge:
- Context Window Saturation: Agents ingest 50k+ token monolithic documents during initial context traces, driving token costs up and causing attention degradation.
- Silent Link Decay & Drift: Renamed headers, broken wikilinks, and orphaned notes silently accumulate, leaving agents with hallucinations and broken code references.
- Knowledge Debt Blindness: Teams lack metrics on documentation cohesion, coupling, and cyclicity.
- Git Regressions: Broken links and unanchored references get committed into
main without automated CI/CD gating.
KnowFrame solves this at compile time. Built on a high-performance Rust core engine (knowframe-core) with a JSON-RPC daemon and an interactive VS Code webview, KnowFrame turns markdown repositories into a living, governed knowledge graph.
🌟 Key Capabilities
1. 🕸️ Interactive Dynamic Knowledge Graph
- Force-Directed Graph Simulation: Real-time D3 physics simulation with smooth layout convergence.
- Radial Hub Pinning: Click any document to make it the central radial hub and explore concentric hop neighborhoods (1-hop direct connections, 2-hop peripheral nodes).
- Cluster Color-Coding: Automatic grouping and coloring by directory hierarchy or tags with customizable opacity falloffs.
- Instant Search & Focus: Fast filter by file path, tag, or connection count with hotkey navigation.
2. 🛡️ Deterministic 5-Pillar Quality Scoring (0–100)
Sub-second, mathematical evaluation of documentation health without non-deterministic LLM hallucinations:
- Pillar 1: Integrity (Weight: 25%): Detects broken wikilinks, circular file paths, unanchored section references, and missing target documents.
- Pillar 2: Structure (Weight: 25%): Identifies God Documents (>3,000 tokens), taxonomy cycles, reference concentration, and AI harness route drifts.
- Pillar 3: Consistency (Weight: 20%): Flags homograph term collisions, synonym ring drift, and metadata schema inconsistencies.
- Pillar 4: Discoverability (Weight: 15%): Discovers isolated silos, unindexed leaves, and high-degree bridge cut vertices.
- Pillar 5: Traceability & Freshness (Weight: 15%): Audits Git commit decay, unreviewed documentation obligations, and stale AST code dependencies.
3. 📊 Active Prompt Load & BPE Token Observatory
- Real-Time BPE Token Counting: Exact byte-pair encoding (cl100k_base) token calculations for every document in the vault.
- Active Prompt Load Gauge: Visualizes total prompt burden with configurable warning thresholds (default: 60k tokens) and critical limits (100k tokens).
- Short Context Tracing: Constrained beam-search route discovery allowing agents to trace only relevant subtrees.
4. 🔒 Knowledge Debt Baselining & Pre-Commit Git Admission Control
- Zero-Tolerance Baseline Ratchet: Record existing legacy debt in
.context-vault/baseline.json. Future commits are guaranteed to never introduce new regressions (knowframe-core validate-candidate).
- Sub-500ms Pre-Commit Admission Hooks: Seamless integration with prek and standard Git hooks to validate file changes before committing.
5. 🧬 18 KOS Structural Evolution Strategies
- Automated detection of structural debt signatures: God Document, Fragmented Concept, Isolated Silo, Implicit Router.
- Evaluates candidate remediation strategies: Topic Decomposition, Section Addressing, Faceted Classification, Thesaurus Relational Reciprocity, Lightweight Routing, and Metadata Materialization.
- Interactive Unified Diff Preview and atomic one-click filesystem transformation with rollback guarantees.
6. ⚡ Dual-Interface Parity (Deterministic Fast-Path + AI Review)
- Deterministic Split-Button
[ Review | ▾ ]: Executes native algorithmic fixes instantly with zero LLM tokens.
- Dropdown Menu
✨ Review with AI: Delegates complex semantic reasoning to configured AI coding agents (Claude Code, Antigravity, Codex, Cursor) with isolated, targeted prompt envelopes.
📦 Installation
From Visual Studio Marketplace
- Open VS Code.
- Press
Ctrl+P (or Cmd+P on macOS) and run:
ext install dscorpsolutions.knowframe
- Alternatively, search for KnowFrame in the Extensions view (
Ctrl+Shift+X).
From VSIX Package
Download the latest knowframe-1.0.3.vsix from GitHub Releases and install via CLI:
code --install-extension knowframe-1.0.3.vsix
🛠️ Quickstart
- Open your Markdown repository or Obsidian vault in VS Code.
- Open the Command Palette (
Ctrl+Shift+P / Cmd+Shift+P) and type:
KnowFrame: Show Graph — Opens the dynamic interactive knowledge graph.
KnowFrame: Open Management & Settings — Opens the Health, Governance, Exclusions, and Evolution drawers.
KnowFrame: Install Git Hooks — Automatically configures the sub-500ms pre-commit admission control hook.
KnowFrame: Fix Broken Links — Automatically scans and fixes broken references across the vault.
KnowFrame ships with standalone Rust CLI binaries for CI/CD pipelines and headless environments:
knowframe-core
# Full repository health audit and defect report
knowframe-core issues
# Validate Git commit candidates against baseline debt
knowframe-core validate-candidate --staged
# Update baseline snapshot
knowframe-core baseline-update
# Run headless background JSON-RPC daemon
knowframe-core daemon --root /path/to/repo --port 8765
knowframe-cli
# Quick quality score summary
knowframe-cli score
# Trace minimal context path between documents
knowframe-cli trace --from docs/index.md --to docs/api.md
⚙️ Configuration Settings
Configure KnowFrame via VS Code Settings (settings.json):
| Setting |
Type |
Default |
Description |
llmWiki.daemonPort |
number |
0 |
Port for the background JSON-RPC daemon (0 = OS-assigned ephemeral). |
llmWiki.gitHookEngine |
string |
"prek" |
Hook orchestrator preference (prek, pre-commit, or script). |
llmWiki.tokenBudgetWarning |
number |
60000 |
Warning threshold for Active Prompt Load tokens. |
llmWiki.tokenBudgetCritical |
number |
100000 |
Critical threshold for Active Prompt Load tokens. |
🏗️ Repository Architecture
KnowFrame is organized as a Cargo workspace with an integrated VS Code extension:
knowframe/
├── Cargo.toml # Cargo Workspace configuration (v1.0.3)
├── crates/
│ ├── core/ # High-performance Rust quality engine & graph compiler
│ ├── cli/ # Headless CLI driver for CI/CD gates and scripting
│ └── engine/ # JSON-RPC 2.0 daemon and WebSocket server
├── extension/ # VS Code Extension Host & Webview UI
│ ├── package.json # Extension manifest (dscorpsolutions.knowframe)
│ ├── src/ # TypeScript extension host (lifecycle, RPC, admission)
│ ├── webview-ui/ # D3 Force graph canvas, management drawer, and components
│ └── bin/ # Platform-specific native binary distribution
└── scripts/
└── distribute.sh # Clean build, package & local installation pipeline
🤝 Contributing
Contributions are warmly welcome! Please feel free to submit issues and pull requests to our GitHub Repository.
- Fork the repository.
- Create your feature branch (
git checkout -b feat/amazing-feature).
- Ensure all tests pass:
cargo test
npm test --prefix extension
- Commit your changes (
git commit -m 'feat: add amazing feature').
- Push to the branch (
git push origin feat/amazing-feature).
- Open a Pull Request.
📄 License
Distributed under the MIT License. See LICENSE for more information.
Authored by DS Corp Solutions — Knowledge Management & Architecture for Autonomous AI Systems.