mindforge-research

MindForge

An agentic intelligence framework for Claude Code — orchestrates multi-agent workflows with governance, memory, and autonomous execution. Production-hardened with true parallelism, streaming SDK, and zero-trust security. Install once, get structured AI-driven development with built-in quality gates.

Latest: v11.3.1

• v11.3.1 — Packaging hotfix. Restores the full published payload: every npx mindforge-cc install now delivers all 174 slash commands, 73 skills, 154 subagents, and the complete .mindforge/ framework. (v11.3.0 shipped a too-narrow npm allowlist that silently dropped commands and skills — fixed here, with a tarball regression test so it cannot recur.)
• v11.3.0 — "Legion". Imports 154 specialized Claude-Code-native subagents across 10 categories into .claude/agents/, fully rebranded and collision-safe. Additive and backward-compatible.

See CHANGELOG.md for full release history.

v11.0.0 — Sovereign Stability

MindForge v11.0.0 "Sovereign Stability" is a production-hardening release focused on reliability, performance, and real-world deployment readiness. Key highlights:

• Memory-safe operations — LRU-bounded caches, atomic writes, log rotation, and snapshot garbage collection eliminate resource leaks in long-running sessions.
• True wave parallelism — Semaphore-based concurrent execution with configurable max concurrency replaces sequential task dispatch.
• Streaming SDK — WebSocket event streaming, streamExecution() with AsyncIterable, and batchExecute() for high-throughput integrations.
• Hardened security — Ephemeral enclave keys, session-scoped agent isolation, time-limited RBAC elevation, dashboard rate limiting, and structured ZK proof returns.
• Production observability — /api/v1/system health endpoint, P95 latency tracking, heap health monitoring, and real EIS client with retry logic.
• Graduated intelligence — Adaptive tier escalation (+1/+2/MAX) with cost-awareness, 3-tier stuck detection, and adaptive context windows.

This release ships 211 personas, 73 skills, 154 specialized subagents, 174 commands, 18 pillars, and 49 swarm templates across 12 engineering domains.

Installation & Setup

🚀 Quick Start (No Install)

Run MindForge immediately for a specific runtime without a permanent installation:

# For Claude Code
npx mindforge-cc@latest --claude --global

# For Antigravity
npx mindforge-cc@latest --antigravity --global

🌍 Global Installation

Enable system-wide /mindforge commands for your primary AI coding runtime:

npm install -g mindforge-cc@latest

📂 Local Project Setup

Initialize MindForge in an existing repository with specialized agent identities:

# For Claude Code
npx mindforge-cc@latest --claude --local

# For Antigravity
npx mindforge-cc@latest --antigravity --local

• Production Hardening (v11.0.0) — LRU caches, atomic JSON writes, log rotation, HANDOFF validation, and temporal snapshot GC for crash-safe long-running sessions.
• True Wave Parallelism (v11.0.0) — Semaphore-based concurrent wave execution with configurable max concurrency replaces sequential dispatch.
• Streaming SDK (v11.0.0) — WebSocket event streaming, streamExecution() AsyncIterable, batchExecute(), model streaming across Anthropic/OpenAI/Gemini providers.
• Graduated Intelligence (v11.0.0) — Adaptive tier escalation (+1/+2/MAX) with cost-awareness, 3-tier stuck detection, and adaptive context windows (10/20/30).
• Security Hardening (v11.0.0) — Ephemeral enclave keys, session-scoped ZTAI, time-limited RBAC elevation, dashboard rate limiting (100 req/min/IP), token expiration.
• Observability (v11.0.0) — /api/v1/system health endpoint, P95 latency ring buffer, heap health monitoring, real EIS client with exponential backoff.
• Grounded Wave Execution (v9.0.0) — AutoRunner reads HANDOFF.json wave groups, dispatches tasks with audit tracing, persists progress, and resumes on restart (Pillar XXIV).
• Model Topology Modernization (v9.0.0) — All model references updated to the Claude 4.x family: claude-opus-4-7, claude-sonnet-4-6, claude-haiku-4-5 (Pillar XXV).
• Unified Memory Architecture (v9.0.0) — Knowledge and graph edges consolidated into SQLite (celestial.db) with FTS5 search. Four JSONL stores replaced by one queryable store (Pillar XXVI).
• Schema Migration Engine (v9.0.0) — _migrations table with versioned migration tracking and transaction-wrapped bulk imports (Pillar XXVII).
• Integration Test Chain (v9.0.0) — 27-assertion end-to-end test suite validating the full execution pipeline (Pillar XXVIII).
• Sovereign Identity Synthesis (v8.1.1) — Autonomous SOUL.md generation derived from reasoning traces and interaction sentiment (Pillar XIX).