mindforge-cc 1.0.5 → 2.0.0-alpha.4

package/docs/reference/sdk-api.md

@@ -1,18 +1,23 @@
-# MindForge SDK API — Reference (v1.0.0)
+# MindForge SDK API — Reference (v2.0.0-alpha.4)
 
 ## Package
+
 `@mindforge/sdk`
 
 ## Exports
+
 From `sdk/src/index.ts`:
+
 - `MindForgeClient`
 - `MindForgeEventStream`
 - `commands`
+- `MindForgeMemory` (Day 11)
 - Types: `MindForgeConfig`, `PhaseResult`, `TaskResult`, `SecurityFinding`,
   `GateResult`, `HealthReport`, `HealthIssue`, `MindForgeEvent`, `CommandOptions`
 - `VERSION`
 
 ## MindForgeClient
+
 High-level API for reading local project state.
 
 Methods:

package/docs/user-guide.md

@@ -1,4 +1,4 @@
-# MindForge User Guide (v1.0.0)
+# MindForge User Guide (v2.0.0-alpha.4)
 
 This guide gets you from install to productive, with the minimum needed to run
 MindForge in a real project. It assumes Node.js 18+.
@@ -68,7 +68,91 @@ This generates `.planning/ARCHITECTURE.md` and an inferred conventions file.
 /mindforge:ship 1
 ```
 
-## 6. Update and migration
+## 6. Autonomous Execution (v2)
+For complex phases that require no human intervention, use the autonomous engine.
+
+### Start autonomous phase
+```bash
+/mindforge:auto --phase 1
+```
+
+### Steering mid-execution
+If you need to guide the agent while it is running in another terminal:
+```bash
+/mindforge:steer "Focus on bug fixes in the auth module"
+```
+
+---
+
+## 7. Browser Runtime & Visual QA (v2)
+MindForge can now control a persistent browser for visual verification and QA.
+
+### Navigate and interact
+```bash
+/mindforge:browse --navigate https://example.com
+/mindforge:browse --click "#login-btn"
+```
+
+### Run visual QA
+To scan for UI bugs and generate regression tests:
+```bash
+/mindforge:qa --diff
+```
+
+---
+
+## 8. Multi-Model Intelligence (v2)
+MindForge v2 can route tasks between Claude, GPT, and Gemini for the best balance of context, cost, and quality.
+
+### Cross-Model Review
+To get an adversarial review from multiple high-tier models:
+```bash
+/mindforge:cross-review
+```
+This will aggregate findings, detect consensus, and generate a `CROSS-REVIEW-[N].md` report.
+
+### Deep Research
+To leverage Gemini 1.5 Pro's huge context window for codebase-wide research:
+```bash
+/mindforge:research "How does the governance layer handle emergency overrides?"
+```
+
+### Cost Tracking
+Monitor your daily spend and model-specific usage:
+```bash
+/mindforge:costs --summary
+```
+MindForge will block further calls if `MODEL_COST_HARD_LIMIT_USD` is reached.
+
+---
+
+## 9. Persistent Knowledge Graph (v2)
+MindForge now has long-term memory. It captures architectural decisions, bug patterns, and team preferences, making them available across project sessions.
+
+### Manual Memory entry
+To explicitly record a project-wide preference:
+```bash
+/mindforge:remember --add "preference: Use direct imports for internal modules" --tags "#nodejs"
+```
+
+### Search and retrieval
+To find related knowledge manually:
+```bash
+/mindforge:remember --search "Tailwind best practices"
+```
+
+### Global Promotion
+To make a specific project insight available across ALL your MindForge projects on this machine:
+```bash
+/mindforge:remember --promote [ENTRY_ID]
+```
+
+### Automatic Loading
+Relevant memories are automatically loaded at the start of every session, providing context about past decisions and patterns.
+
+---
+
+## 10. Update and migration
 ### Check for updates
 ```
 /mindforge:update
@@ -84,7 +168,7 @@ This generates `.planning/ARCHITECTURE.md` and an inferred conventions file.
 /mindforge:migrate --from v0.6.0 --to v1.0.0
 ```
 
-## 7. Plugins
+## 11. Plugins
 ### List / validate
 ```
 /mindforge:plugins list
@@ -96,7 +180,7 @@ This generates `.planning/ARCHITECTURE.md` and an inferred conventions file.
 /mindforge:plugins install mindforge-plugin-<name>
 ```
 
-## 8. Skills
+## 12. Skills
 ```
 /mindforge:skills list
 /mindforge:skills validate
@@ -104,7 +188,7 @@ This generates `.planning/ARCHITECTURE.md` and an inferred conventions file.
 
 To publish or install a skill, see `docs/skills-publishing-guide.md`.
 
-## 9. Token usage profiling
+## 13. Token usage profiling
 ```
 /mindforge:tokens --profile
 /mindforge:tokens --summary
@@ -113,7 +197,7 @@ To publish or install a skill, see `docs/skills-publishing-guide.md`.
 Token optimization policies are defined in:
 - `.mindforge/production/token-optimiser.md`
 
-## 10. Configuration (MINDFORGE.md)
+## 14. Configuration (MINDFORGE.md)
 Key settings live in `MINDFORGE.md`. See:
 - `docs/reference/config-reference.md`
 
@@ -123,18 +207,18 @@ Common settings:
 - `TOKEN_WARN_THRESHOLD`, `TOKEN_LEAN_MODE`
 - `MINDFORGE_AUTO_CHECK_UPDATES`
 
-## 11. Troubleshooting
+## 15. Troubleshooting
 - Health issues: run `/mindforge:health --repair`
 - Schema drift: run `/mindforge:migrate --dry-run` then apply
 - Installer issues: re-run with `--force`
 - CI mode: set `CI=true` and check `.mindforge/ci/` docs
 
-## 12. Security
+## 16. Security
 MindForge never stores credentials in files. See:
 - `docs/security/SECURITY.md`
 - `docs/security/threat-model.md`
 
-## 13. Reference docs
+## 17. Reference docs
 - Commands: `docs/reference/commands.md`
 - SDK: `docs/reference/sdk-api.md`
 - Skills: `docs/reference/skills-api.md`

package/docs/usp-features.md

@@ -1,7 +1,7 @@
-# MindForge v1.0.0 — Unique Selling Points, Features, and Best Practices
+# MindForge v2.0.0 — Unique Selling Points, Features, and Best Practices (v2.0.0-alpha.4)
 
-This document summarizes what makes MindForge v1.0.0 distinct, what features
-are included in the first stable release, and how to use them effectively.
+This document summarizes what makes MindForge v2.0.0 distinct, what features
+are included in the latest alpha release, and how to use them effectively.
 
 ---
 
@@ -31,6 +31,18 @@ are included in the first stable release, and how to use them effectively.
 7. **Extensible with plugins and skills**
    - Plugins add commands and skills without changing core files, keeping upgrades safe.
 
+8. **Autonomous “Walk-Away” Execution (v2)**
+   - `/mindforge:auto` allows for full phase/milestone completion with intelligent stuck detection and node repair (RETRY → DECOMPOSE → ESCALATE).
+
+9. **Persistent Visual QA (v2)**
+   - Headful/headless browser runtime with named session persistence and systematic visual diff verification.
+
+10. **Multi-Model Intelligence Layer (v2)**
+    - Dynamic routing between Anthropic, OpenAI, and Gemini based on task persona and security tier. Adversarial code reviews ensure maximum coverage.
+
+11. **Persistent Knowledge Graph (v2)**
+    - Captures and ranks engineering context (decisions, bug patterns, preferences) across project sessions using TF-IDF and confidence reinforcement.
+
 ---
 
 ## Feature Set (v1.0.0)
@@ -141,18 +153,54 @@ preserving scope (local vs global).
 
 ---
 
-### 10. Token Usage Optimiser
-**What it does:** Profiles and reduces token usage across sessions.
+### 11. Autonomous Execution (v2)
+**What it does:** Enables handoff-free execution of complex phases and milestones.
 
 **How to use:**
+```bash
+/mindforge:auto --phase 1
+/mindforge:steer "Focus on security hardening next"
 ```
-/mindforge:tokens --profile
-/mindforge:tokens --summary
+
+---
+
+### 12. Browser Runtime & Visual QA (v2)
+**What it does:** Playwright-powered browser control with session persistence and automated UI bug detection.
+
+**How to use:**
+```bash
+/mindforge:browse --navigate https://example.com
+/mindforge:qa --diff
+```
+
+---
+
+### 13. Multi-Model Intelligence (v2)
+**What it does:** Unified API client for Anthropic, OpenAI, and Gemini with persona-based routing and automated cost tracking.
+
+**How to use:**
+```bash
+/mindforge:cross-review # Adversarial multi-model review
+/mindforge:research "Deep search query" # 1M context research
+/mindforge:costs # Usage and budget summary
+```
+
+---
+
+### 14. Persistent Knowledge Graph (v2)
+**What it does:** Long-term memory system that ensures architectural decisions and bug patterns are never forgotten. Supports TF-IDF search, confidence scoring, and global promotion.
+
+**How to use:**
+```bash
+/mindforge:remember --add "insight" # Manual capture
+/mindforge:remember --search "query" # Manual retrieval
+/mindforge:remember --promote [ID] # Promote to machine-wide global store
+/mindforge:remember --stats # View memory health
 ```
 
 ---
 
-### 11. SDK (TypeScript)
+### 15. SDK (TypeScript)
 **What it does:** Programmatic access to health, audit log, event stream, and commands.
 
 **How to use:**

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "mindforge-cc",
-  "version": "1.0.5",
+  "version": "2.0.0-alpha.4",
   "description": "MindForge - Enterprise Agentic Framework for Claude Code and Antigravity",
   "bin": {
     "mindforge-cc": "bin/install.js"
@@ -39,7 +39,8 @@
     "@eslint/js": "^9.0.0",
     "@types/node": "^20.0.0",
     "eslint": "^9.0.0",
-    "globals": "^15.0.0"
+    "globals": "^15.0.0",
+    "playwright": "^1.40.0"
   },
   "engines": {
     "node": ">=18.0.0"