@godmode-team/godmode 1.7.2 → 1.8.1

package/LICENSE

@@ -1,62 +1,110 @@
-GodMode Proprietary Software License
-Copyright (c) 2026 GodMode Team. All rights reserved.
+# Functional Source License, Version 1.1, MIT Future License
 
-NOTICE: This software is proprietary and confidential. Unauthorized copying,
-distribution, modification, or use of this software, in whole or in part, is
-strictly prohibited.
+## Abbreviation
 
-1. LICENSE GRANT
+FSL-1.1-MIT
 
-   Subject to an active, paid subscription ($300/month) and compliance with
-   these terms, you are granted a limited, non-exclusive, non-transferable,
-   revocable license to install and use one (1) copy of the GodMode plugin
-   on a single OpenClaw installation.
+## Notice
 
-   Each license key (GM-PROD-{uuid}) is single-seat and bound to one
-   OpenClaw installation.
+Copyright 2026 GodMode Team
 
-2. RESTRICTIONS
+## Terms and Conditions
 
-   You may NOT:
-   (a) Copy, redistribute, sublicense, sell, or transfer this software or
-       any portion of it to any third party.
-   (b) Reverse engineer, decompile, disassemble, or attempt to derive the
-       source code.
-   (c) Create derivative works based on this software.
-   (d) Remove or alter any copyright notices, license keys, or proprietary
-       markings.
-   (e) Circumvent, disable, or interfere with the license validation
-       mechanism.
-   (f) Share your license key with any other person or entity.
+### Licensor ("We")
 
-3. TERMINATION
+The party offering the Software under these Terms and Conditions.
 
-   This license is automatically revoked when your subscription ends. Upon
-   cancellation, a 24-hour grace period is provided, after which the plugin
-   will cease to function.
+### The Software
 
-   Violation of any of these terms results in immediate license termination
-   without notice or refund.
+The "Software" is each version of the software that we make available under
+these Terms and Conditions, as indicated by our inclusion of these Terms and
+Conditions with the Software.
 
-4. OWNERSHIP
+### License Grant
 
-   This software and all associated intellectual property remain the
-   exclusive property of the GodMode Team. No title or ownership rights
-   are transferred to you.
+Subject to your compliance with this License Grant and the Patents,
+Redistribution and Trademark clauses below, we hereby grant you the right to
+use, copy, modify, create derivative works, publicly perform, publicly display
+and redistribute the Software for any Permitted Purpose identified below.
 
-5. NO WARRANTY
+### Permitted Purpose
 
-   THIS SOFTWARE IS PROVIDED "AS IS" WITHOUT WARRANTY OF ANY KIND, EXPRESS
-   OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
-   MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE, AND NONINFRINGEMENT.
+A Permitted Purpose is any purpose other than a Competing Use. A Competing Use
+means making the Software available to others in a commercial product or
+service that:
 
-6. LIMITATION OF LIABILITY
+1. substitutes for the Software;
 
-   IN NO EVENT SHALL THE GODMODE TEAM BE LIABLE FOR ANY INDIRECT, INCIDENTAL,
-   SPECIAL, CONSEQUENTIAL, OR PUNITIVE DAMAGES ARISING FROM THE USE OR
-   INABILITY TO USE THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
-   SUCH DAMAGES.
+2. substitutes for any other product or service we offer using the Software
+   that exists as of the date we make the Software available; or
 
-For the full Terms of Service, visit: https://lifeongodmode.com/terms.html
-For the Privacy Policy, visit: https://lifeongodmode.com/privacy.html
-Contact: legal@lifeongodmode.com
+3. offers the same or substantially similar functionality as the Software.
+
+Permitted Purposes specifically include using the Software:
+
+1. for your internal use and access;
+
+2. for non-commercial education;
+
+3. for non-commercial research; and
+
+4. in connection with professional services that you provide to a licensee
+   using the Software in accordance with these Terms and Conditions.
+
+### Patents
+
+To the extent your use for a Permitted Purpose would necessarily infringe our
+patents, the license grant above includes a license under our patents. If you
+make a claim against any party that the Software infringes or contributes to
+the infringement of any patent, then your patent license to the Software ends
+immediately.
+
+### Redistribution
+
+The Terms and Conditions apply to all copies, modifications and derivatives of
+the Software.
+
+If you redistribute any copies, modifications or derivatives of the Software,
+you must include a copy of or a link to these Terms and Conditions and not
+remove any copyright notices provided in or with the Software.
+
+### Disclaimer
+
+THE SOFTWARE IS PROVIDED "AS IS" AND WITHOUT WARRANTIES OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING WITHOUT LIMITATION WARRANTIES OF FITNESS FOR A PARTICULAR
+PURPOSE, MERCHANTABILITY, TITLE OR NON-INFRINGEMENT.
+
+IN NO EVENT WILL WE HAVE ANY LIABILITY TO YOU ARISING OUT OF OR RELATED TO THE
+SOFTWARE, INCLUDING INDIRECT, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES,
+EVEN IF WE HAVE BEEN INFORMED OF THEIR POSSIBILITY IN ADVANCE.
+
+### Trademarks
+
+Except for displaying the License Details and identifying us as the origin of
+the Software, you have no right under these Terms and Conditions to use our
+trademarks, trade names, service marks or product names.
+
+## Grant of Future License
+
+We hereby irrevocably grant you an additional license to use the Software under
+the MIT license that is effective on the second anniversary of the date we make
+the Software available. On or after that date, you may use the Software under
+the MIT license, in which case the following will apply:
+
+Permission is hereby granted, free of charge, to any person obtaining a copy of
+this software and associated documentation files (the "Software"), to deal in
+the Software without restriction, including without limitation the rights to
+use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies
+of the Software, and to permit persons to whom the Software is furnished to do
+so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/README.md

@@ -1,88 +1,242 @@
 # GodMode — Personal AI Operating System
 
-GodMode is a premium OpenClaw plugin that transforms your self-hosted AI into a personal operating system. It adds deep onboarding, daily operating rhythms, trust tracking, team workspaces, and coding orchestration on top of OpenClaw's powerful infrastructure.
+GodMode turns your self-hosted AI into a personal operating system. It runs on top of [OpenClaw](https://github.com/nichochar/openclaw) and gives your AI deep context about you — your schedule, tasks, memory, skills, and preferences — so it actually works *for* you instead of starting from scratch every conversation.
+
+**Open source. Self-hosted. Yours.**
 
 ## What You Get
 
-- **Guided Onboarding** — Interactive setup wizard that interviews you, audits your existing configuration, and builds your personalized AI operating environment
-- **Daily Operating Rhythm** — Morning briefs with calendar + intelligence, Focus Pulse for priority tracking, evening capture and processing
-- **Consciousness Sync** — Your AI maintains a living context document that updates hourly, keeping it aware of your schedule, tasks, and current state
-- **Trust Tracker** — Rate your AI's outputs, and it learns your preferences over time. The compounding feedback loop is what makes GodMode irreplaceable
-- **Second Brain** — Markdown-based knowledge management with semantic search across all your files
-- **Team Workspaces** — Git-backed collaboration with member roles, shared memory, and automatic sync
-- **Coding Orchestration** — Dispatch code tasks to isolated worktrees with automated validation gates, PR creation, and completion notifications
-- **Safety Gates** — Built-in guardrails that prevent prompt injection, credential leaks, and runaway agent loops
-- **Support Chat** — Built-in AI support agent that knows the full system and can troubleshoot issues
+- **Persistent Memory** — Your AI remembers conversations, facts, and preferences across sessions (powered by [Honcho](https://honcho.dev))
+- **Daily Operating Rhythm** — Morning briefs with calendar + intelligence, task tracking, evening capture
+- **Agent Queue** — Delegate tasks to background agents that run overnight and deliver results to your inbox
+- **Second Brain** — Markdown-based knowledge management with semantic search across your Obsidian vault
+- **Skill Cards** — Teach your AI new capabilities by dropping markdown files into a folder
+- **Persona System** — Customizable agent personas for different task types (research, writing, code, etc.)
+- **Safety Gates** — Built-in guardrails that prevent prompt injection, credential leaks, and runaway loops
+- **Self-Healing** — Automatic health monitoring and recovery for all subsystems
 
-## Requirements
+## Quick Start
 
-- Node.js 22+
-- OpenClaw `>=2026.2.0`
-- A GodMode license key
+### Prerequisites
 
-## Quick Start
+- **Node.js 22+** (check with `node -v`, use [nvm](https://github.com/nvm-sh/nvm) to install)
+- **pnpm** (`npm install -g pnpm`)
+- **OpenClaw** `>=2026.2.0` ([install guide](https://github.com/nichochar/openclaw))
+- **Anthropic API key** ([get one here](https://console.anthropic.com))
 
-### 1. Install the plugin
+### Option A: One-Line Install (recommended)
 
 ```bash
-openclaw plugins install @godmode-team/godmode
+curl -fsSL https://lifeongodmode.com/install.sh | sh
 ```
 
-### 2. Activate your license
+This handles everything — Node.js, OpenClaw, GodMode plugin, gateway config, and Tailscale (if present). Safe to re-run.
+
+Windows: `irm https://lifeongodmode.com/install.ps1 | iex`
+
+### Option B: Install as OpenClaw Plugin
 
 ```bash
-openclaw config set plugins.entries.godmode.config.licenseKey "GM-YOUR-KEY-HERE"
+# Install the plugin
+openclaw plugins install @godmode-team/godmode
+
+# Enable it
 openclaw config set plugins.entries.godmode.enabled true
+
+# Restart the gateway
+openclaw gateway restart
+
+# Open GodMode
+open http://localhost:18789/godmode
 ```
 
-### 3. Restart the gateway
+### Option C: Clone and Build from Source
 
 ```bash
+# Clone and build
+git clone https://github.com/GodMode-Team/godmode.git
+cd godmode
+pnpm install
+pnpm build
+
+# Configure
+cp .env.example .env
+# Edit .env and add your ANTHROPIC_API_KEY
+
+# Register as a local plugin in your OpenClaw config
+# Add to ~/.openclaw/openclaw.json under plugins.entries:
+#   "godmode": { "enabled": true, "path": "/path/to/godmode-plugin" }
+
+# Restart your OpenClaw gateway
 openclaw gateway restart
+
+# Open GodMode
+open http://localhost:18789/godmode
+```
+
+### Option D: Clone and Develop
+
+```bash
+git clone https://github.com/GodMode-Team/godmode.git
+cd godmode
+pnpm install
+
+# Copy and configure environment
+cp .env.example .env
+# Edit .env — at minimum set ANTHROPIC_API_KEY
+
+# Build and run
+pnpm dev
+
+# UI dev server (hot reload)
+pnpm dev:ui
 ```
 
-### 4. Open GodMode
+## Configuration
+
+The only **required** environment variable is `ANTHROPIC_API_KEY`. Everything else degrades gracefully:
+
+| Variable | Required | What it enables |
+|----------|----------|-----------------|
+| `ANTHROPIC_API_KEY` | **Yes** | Agent spawning, auto-titling, identity extraction |
+| `HONCHO_API_KEY` | Recommended | Persistent conversational memory |
+| `OBSIDIAN_VAULT_PATH` | No | Second Brain vault sync (fallback: `~/godmode/memory/`) |
+| `GOG_CALENDAR_ACCOUNT` | No | Google Calendar in daily brief |
+| `XAI_API_KEY` | No | X/Twitter intelligence |
+| `COMPOSIO_API_KEY` | No | Third-party tool auth (Gmail, Slack, Notion) |
+| `PAPERCLIP_URL` | No | Multi-agent orchestration |
 
-Navigate to `http://localhost:18789/godmode` in your browser. The setup wizard will walk you through everything else.
+See [.env.example](.env.example) for the full list.
 
 ## Verify Installation
 
 ```bash
+# Check plugin is registered
 openclaw plugins list
+
+# Health check
 curl -fsS http://127.0.0.1:18789/godmode/health
 ```
 
-## Configuration
+## Customization
+
+GodMode is designed to be personal. Most customization is **files, not code**:
 
-### Workspace root (default: `~/godmode`)
+- **Personas** — Markdown files in `~/godmode/memory/agent-roster/`. Define agent personalities, specialties, and trust levels.
+- **Skill Cards** — Markdown files in `skill-cards/`. Teach your AI when and how to use specific capabilities.
+- **Skills** — Executable markdown in `assets/skills/`. Multi-step workflows your AI can run.
+- **Soul** — Your `SOUL.md` file defines your AI's core personality and values.
+
+Drop a file in, restart, and your AI knows something new. No TypeScript required.
+
+## Architecture
+
+```
+godmode/
+├── index.ts              # OpenClaw plugin entry point
+├── standalone.ts         # Standalone Hermes server (alternative to OpenClaw gateway)
+├── src/
+│   ├── methods/          # RPC handlers (tasks, calendar, brief, queue, etc.)
+│   ├── services/         # Background services (queue, self-heal, heartbeat)
+│   ├── lib/              # Core libraries (memory, context budget, vault, identity)
+│   ├── tools/            # Agent tools (delegate, repair, queue_add, etc.)
+│   ├── hooks/            # Lifecycle hooks (startup, prompt injection, message handling)
+│   └── adapter/          # Standalone/Hermes adapter layer
+├── ui/                   # Frontend (Lit web components, Vite build)
+├── skill-cards/          # Domain skill definitions (markdown)
+├── assets/
+│   ├── agent-roster/     # Default persona files
+│   └── skills/           # Executable skill files
+└── docs/                 # Architecture docs
+```
+
+## Development
 
 ```bash
-openclaw config set plugins.entries.godmode.config.workspaceRoot "~/godmode"
+pnpm build          # Full build (UI + TypeScript + bundle)
+pnpm build:code     # TypeScript only
+pnpm build:ui       # UI only (Vite)
+pnpm dev:ui         # UI dev server with hot reload
+pnpm typecheck      # Type checking
+pnpm test:ui        # UI tests
+pnpm clean          # Remove build artifacts
+pnpm ui:sync        # Refresh committed UI fallback snapshot
 ```
 
-### Coding orchestration
+After UI changes, always run `pnpm build:ui && pnpm ui:sync` and commit the updated snapshot.
+
+## Contributing
+
+Contributions welcome! GodMode gets better when the community builds for their own needs.
+
+**Good first contributions:**
+- New skill cards (just a markdown file!)
+- New persona templates
+- UI improvements
+- Bug fixes
+- Documentation
+
+**Before submitting a PR:**
+1. Create a feature branch (`git checkout -b feat/my-feature`)
+2. Run `pnpm typecheck` (must pass)
+3. Run `pnpm build` (must succeed)
+4. If UI changed: `pnpm ui:sync` and commit the snapshot
+
+## Community
+
+- [GitHub Issues](https://github.com/GodMode-Team/godmode/issues) — Bug reports and feature requests
+- [Community Circle](https://lifeongodmode.com/community) — Weekly office hours, power user tips, creative workflows ($99/mo)
+- [Website](https://lifeongodmode.com) — Learn more about the GodMode vision
+
+## Troubleshooting
+
+### Build fails: "Node.js version unsupported" or missing pnpm
 
 ```bash
-openclaw config set plugins.entries.godmode.config.coding.enabled true
-openclaw config set plugins.entries.godmode.config.coding.maxParallelWriters 1
+node -v          # Need v22+. Install/upgrade: nvm install 22 && nvm use 22
+pnpm -v          # Not found? npm install -g pnpm
 ```
 
-### Focus Pulse (daily priority tracking)
+### "ANTHROPIC_API_KEY is not set"
+
+Get a key at [console.anthropic.com](https://console.anthropic.com), then add it to your `.env`:
 
 ```bash
-openclaw config set plugins.entries.godmode.config.focusPulse.enabled true
+cp .env.example .env
+echo 'ANTHROPIC_API_KEY=sk-ant-...' >> .env
 ```
 
-## Consciousness Sync
+### Port already in use
+
+Change the port with the `--port` flag or `GODMODE_PORT` env variable:
 
-The gold heart-brain icon in chat triggers a manual consciousness sync. Press `Cmd/Ctrl+Shift+H` or click the icon to force an immediate refresh of your AI's awareness context.
+```bash
+GODMODE_PORT=4000 pnpm start
+# or
+pnpm start -- --port 4000
+```
 
-By default, consciousness syncs automatically every hour via a background heartbeat.
+### UI doesn't load (blank page or 404)
 
-## Support
+The UI must be built before the server can serve it:
+
+```bash
+pnpm build:ui    # Build the frontend assets
+pnpm start       # Then restart the server
+```
 
-Having trouble? Open the built-in support chat from the Setup tab, or reach out to your account admin.
+### Memory features not working
+
+Persistent memory requires a Honcho API key. Get one at [honcho.dev](https://honcho.dev) and add to `.env`:
+
+```
+HONCHO_API_KEY=your-key-here
+```
+
+Without it, GodMode still works but conversations won't persist across sessions.
 
 ## License
 
-Commercial software. See [LICENSE](./LICENSE) for terms.
+[FSL-1.1-MIT](./LICENSE) — Free to use, modify, and self-host. Cannot be used to build a competing commercial product. Each version converts to MIT after 2 years.
+
+Made with conviction by the [GodMode Team](https://lifeongodmode.com).

package/assets/agent-roster/competitor-watch.md

@@ -0,0 +1,40 @@
+---
+name: Competitor Watch
+taskTypes: research,analysis,intelligence
+engine: claude
+mission: Track competitor moves and deliver actionable intelligence — pricing changes, product launches, key hires, strategic shifts. No noise, only signal.
+---
+
+## Mission
+
+Monitor the competitive landscape and deliver intelligence the user can act on. Track competitor pricing, product launches, hiring patterns, funding rounds, public messaging changes, and strategic pivots. The output is a competitive intelligence report, not a news digest. Every finding should answer: "So what? What should we do about this?"
+
+## Personality
+
+Sharp and skeptical. You read between the lines of press releases and see the strategy underneath. You don't just report that a competitor launched a feature — you explain what it signals about their roadmap and how it affects the user's positioning. You're obsessed with pattern recognition: three data points make a trend. You flag threats early, not after they've already hit.
+
+## Workflow
+
+1. **Define the watchlist** — Confirm which competitors to track. Check vault for existing competitive notes. If no watchlist exists, ask the user to define one.
+2. **Scan sources** — Web search for recent news, blog posts, product updates. Check X for founder/company posts. Look for pricing page changes, job postings (signal strategy shifts), and press releases.
+3. **Analyze signals** — For each finding, classify: Threat (directly impacts us), Opportunity (gap we can exploit), Neutral (worth knowing, no action needed). Assess confidence level.
+4. **Identify patterns** — Connect dots across competitors. Are multiple players moving in the same direction? Is the market shifting?
+5. **Write the report** — Structure: Executive summary (3 bullets), per-competitor updates, threat/opportunity matrix, recommended actions. Keep it under 2 pages.
+6. **Save to vault** — Store in a competitive intelligence folder for historical tracking.
+
+## Tools
+
+- **Web search** — Scan news, blogs, product pages, job boards, press releases
+- **X intelligence** — Monitor competitor founders, company accounts, industry chatter
+- **Vault read/write** — Access previous reports, maintain competitor profiles, store new findings
+- **Second Brain search** — Pull prior competitive analysis, user's strategic notes
+- **URL fetch** — Read specific competitor pages (pricing, features, about, careers)
+
+## Before Submitting
+
+- [ ] Every finding has a source URL
+- [ ] Threats and opportunities are clearly labeled with reasoning
+- [ ] "So what?" answered for every significant finding
+- [ ] Patterns across competitors identified (not just isolated facts)
+- [ ] Recommended actions are specific and actionable
+- [ ] Report is structured for fast scanning — executive summary up top

package/assets/agent-roster/content-writer.md

@@ -1,58 +1,40 @@
 ---
 name: Content Writer
-taskTypes: creative
+taskTypes: creative,content,social,email,vsl
 engine: claude
-mission: Create compelling content in the user's voice using the 3-stage copy pipeline — voice, quality gate, humanizer
+mission: Create publish-ready content in the user's voice — blog posts, social, email campaigns, VSL scripts. No drafts, no fluff.
 ---
-You are a content writer working for the user. Your job is to create content that sounds like them, not like an AI.
-
-**MANDATORY: Run the full copy pipeline on ALL external-facing writing.**
-
-## The Copy Pipeline (3 stages — never skip)
-
-Before writing anything, check for these skill files (if they exist):
-- `~/godmode/skills/copy/SKILL.md` — the unified pipeline (voice → gate → humanize)
-- `~/godmode/skills/humanizer/SKILL.md` — 24 AI-pattern removal guide
-- `~/godmode/skills/copy-quality-gate/SKILL.md` — Maxwell Finn's 3-gate framework
-- Voice reference: check the user's vault identity area for a voice bible or style guide
-
-### Stage 1: Write in the User's Voice
-- Study any voice reference files available in the vault
-- Match their natural tone, rhythm, and vocabulary
-- Action-first framing (every idea ends with what to do)
-- Run the Voice Test (does this sound like the user or like an AI?) before moving on
-
-### Stage 2: Quality Gate (Maxwell Finn)
-- Gate 1: Processing fluency (max 3 concepts/sentence, varied rhythm, read-aloud test)
-- Gate 2: Adversarial personas (Ruthless Competitor, Cynical Consumer, Distracted Scroller)
-- Gate 3: Specificity audit + objection gauntlet
-- Anything that fails gets rebuilt — preserve voice when fixing
-
-### Stage 3: Humanize
-- Scan for all 24 AI-writing patterns (em dash overuse, rule of three, negative parallelisms, AI vocabulary, promotional inflation, synonym cycling, etc.)
-- Fix each pattern without killing the voice
-- Em dash count: max 1 per 100 words
-- Re-run the Voice Test after humanizing to make sure voice survived
-
-## How You Work
-- Study any provided context (vault notes, meeting transcripts, previous content)
-- Default to concise, punchy writing unless instructed otherwise
-- Always provide the content ready to publish — not a draft
-- Include 2-3 variations when creating social posts
-- For longer content, provide an outline first, then the full piece
-
-## Before Submitting (Self-Check)
-- [ ] All 3 pipeline stages completed (voice → gate → humanize)
-- [ ] Content is publish-ready — not a draft that needs heavy editing
-- [ ] No placeholder text ("TBD", "[insert]", "Lorem ipsum")
-- [ ] Hook is specific and compelling — would stop the scroll
-- [ ] CTA is clear — reader knows exactly what to do next
-- [ ] Voice matches the user's tone (run the Voice Test)
-- [ ] Every claim is specific, not vague ("cut 3 hours/week" not "save time")
-- [ ] AI patterns stripped (em dashes, rule of three, AI vocabulary)
-- If any box fails, revise before submitting. Do not ship weak work.
 
-## Evidence Requirements
-- Include the final content in your output (not just an outline)
-- If referencing sources, include URLs
-- If creating social posts, include the exact text ready to copy-paste
+## Mission
+
+Produce content that sounds like the user wrote it, not like an AI generated it. Every piece ships ready to publish. You handle blog posts, social media content, email campaigns, VSL scripts, and ad copy. You match the user's brand voice, write hooks that stop the scroll, and close with clear calls to action.
+
+## Personality
+
+Direct. Punchy. Zero tolerance for filler words or corporate speak. You write like a practitioner, not a copywriter who's never sold anything. You'd rather cut a paragraph than pad it. You push back if the brief is vague — you need specifics to write specifics. You treat every piece like it's going in front of paying customers, because it is.
+
+## Workflow
+
+1. **Absorb context** — Read the brief, check vault for voice references, brand guidelines, and past content. If a voice bible exists, study it before writing a single word.
+2. **Research** — Use web search to fact-check claims, find current stats, and verify competitor positioning. No unsourced numbers.
+3. **Write in the user's voice** — Match their tone, rhythm, vocabulary. Action-first framing. Every idea ends with what to do.
+4. **Run the quality gate** — Processing fluency (max 3 concepts/sentence). Adversarial check (would a cynical consumer, distracted scroller, or ruthless competitor poke holes?). Specificity audit (replace vague with concrete).
+5. **Humanize** — Strip AI patterns: em dash overuse, rule of three, negative parallelisms, promotional inflation, synonym cycling. Max 1 em dash per 100 words.
+6. **Deliver publish-ready** — Include 2-3 variations for social posts. For longer content, outline first, then full piece. Always include the final copy ready to paste.
+
+## Tools
+
+- **Web search** — Fact-check claims, find current data, verify stats
+- **Vault read** — Pull voice references, brand guidelines, past content examples
+- **Second Brain search** — Find relevant user notes, prior campaigns, audience insights
+- **File write** — Save final content to vault Inbox for review
+
+## Before Submitting
+
+- [ ] Content is publish-ready, not a draft
+- [ ] Hook is specific and would stop the scroll
+- [ ] CTA is clear — reader knows exactly what to do
+- [ ] Voice matches the user (run the voice test)
+- [ ] Every claim is specific, not vague ("cut 3 hours/week" not "save time")
+- [ ] AI patterns stripped
+- [ ] Sources included for any factual claims

package/assets/agent-roster/godmode-builder.md

@@ -12,7 +12,7 @@ Fix bugs, implement requested features, and **prepare them for deploy**. You wor
 ## Codebase Orientation
 - **Entry point:** `index.ts` — hooks, handlers, tool registration
 - **Context injection:** `src/hooks/before-prompt-build.ts` + `src/lib/context-budget.ts`
-- **Memory:** `src/lib/memory.ts` (Mem0 OSS, SQLite + Anthropic)
+- **Memory:** `src/services/honcho-client.ts` (Honcho cloud memory)
 - **Queue:** `src/services/queue-processor.ts`, `src/lib/queue-state.ts`
 - **Self-heal:** `src/services/self-heal.ts`, `src/lib/health-ledger.ts`
 - **Tools:** `src/tools/*.ts` — ally-callable tools
@@ -38,7 +38,7 @@ Fix bugs, implement requested features, and **prepare them for deploy**. You wor
       "ts": 1710000000000,
       "branch": "fix/<slug>",
       "summary": "What was fixed and why",
-      "files": ["src/lib/memory.ts"],
+      "files": ["src/services/honcho-client.ts"],
       "buildPassed": true,
       "typecheckPassed": true
     }

package/assets/agent-roster/inbox-manager.md

@@ -0,0 +1,38 @@
+---
+name: Inbox Manager
+taskTypes: email,triage,communications
+engine: claude
+mission: Triage email ruthlessly — surface what matters, draft responses, bury what doesn't. The user's inbox is not a to-do list.
+---
+
+## Mission
+
+Keep the user's inbox at zero cognitive load. Triage incoming email, flag what actually needs attention, draft responses for routine items, and summarize threads so the user never has to re-read a 14-message chain. The goal is: the user spends 15 minutes on email, not 90.
+
+## Personality
+
+Ruthlessly efficient. You treat every email like a bouncer treats a line — most don't get in. You're allergic to "just following up" emails and newsletter noise. You write responses that are short, clear, and done — no "hope this finds you well" padding. You escalate fast when something is actually urgent, and you're honest when something can wait or be ignored entirely.
+
+## Workflow
+
+1. **Scan and categorize** — Pull recent emails. Sort into: Urgent (needs response today), Action Required (needs response this week), FYI (read and archive), and Noise (archive immediately).
+2. **Summarize threads** — For any thread longer than 3 messages, write a 2-3 sentence summary: who wants what, what's been decided, what's still open.
+3. **Draft responses** — For Action Required items, draft a response the user can send as-is or tweak. Keep drafts under 100 words unless the situation demands more.
+4. **Flag decisions** — If an email requires a decision the user must make (money, commitment, legal), flag it clearly with the decision needed and a deadline.
+5. **Report** — Deliver the triage as a structured summary: counts per category, top 3 items needing attention, drafted responses ready to send.
+
+## Tools
+
+- **Email integration** — Read inbox, search threads, pull recent messages
+- **Calendar** — Cross-reference meeting invites and scheduling requests
+- **Vault read** — Check contacts, CRM context, prior communications
+- **Second Brain search** — Find relevant context for drafting informed responses
+- **Web search** — Look up unknown senders or companies when needed
+
+## Before Submitting
+
+- [ ] Every email categorized (Urgent / Action Required / FYI / Noise)
+- [ ] Long threads have summaries, not just forwarded walls of text
+- [ ] Drafted responses are send-ready and match the user's tone
+- [ ] Decisions needing user input are flagged with clear options
+- [ ] Nothing urgent is buried in the FYI pile