Skip to content

Self-evolving second brain with 35 AI skills, 10 agents, and people CRM. Closed-loop harness: a V-model verification lifecycle where the worker never grades its own homework. Plus paired anti-slop design skills for marketing and product UI. Works with Claude Code, Cursor, Kiro, Gemini CLI, Codex.

Add Gemini CLI + OpenAI Codex support, smarter onboarding, agent team mode

Huy Tieu20578724a4e4
diff --git a/.claude/skills/braindump/SKILL.md b/.claude/skills/braindump/SKILL.md
index 2040602..e002904 100644
--- a/.claude/skills/braindump/SKILL.md
+++ b/.claude/skills/braindump/SKILL.md
@@ -14,6 +14,12 @@ Transform raw thoughts into strategic intelligence through quick capture, system
- User has ideas they want to quickly record
- User mentions wanting to get thoughts out of their head
+## Agent Mode Awareness
+
+**Check `agent_mode` in `00-inbox/MY-PROFILE.md` frontmatter:**
+- If `agent_mode: team` — delegate research, analysis, and writing subtasks to specialist sub-agents (e.g., use Task tool to delegate deep analysis, competitive intel extraction, or pattern recognition to separate agents). Combine results before presenting to user.
+- If `agent_mode: solo` (default) — handle everything directly in the conversation. No delegation.
+
## Pre-Flight Check
**Before executing, check for user profile:**
diff --git a/.claude/skills/daily-brief/SKILL.md b/.claude/skills/daily-brief/SKILL.md
index 20d568c..d4feaac 100644
--- a/.claude/skills/daily-brief/SKILL.md
+++ b/.claude/skills/daily-brief/SKILL.md
@@ -14,6 +14,12 @@ Find verified, relevant news for personalized daily briefings with strict verifi
- User wants to stay updated on their interests
- Morning routine or regular check-in time
+## Agent Mode Awareness
+
+**Check `agent_mode` in `00-inbox/MY-PROFILE.md` frontmatter:**
+- If `agent_mode: team` — delegate news research across different interest areas to parallel sub-agents (e.g., one agent per topic cluster). Each agent searches, verifies sources, and returns findings. Combine and synthesize results into the final brief.
+- If `agent_mode: solo` (default) — handle all research and synthesis directly in the conversation. No delegation.
+
## Pre-Flight Check
**Before executing, check for user profile:**
diff --git a/.claude/skills/knowledge-consolidation/SKILL.md b/.claude/skills/knowledge-consolidation/SKILL.md
index d53032e..5954705 100644
--- a/.claude/skills/knowledge-consolidation/SKILL.md
+++ b/.claude/skills/knowledge-consolidation/SKILL.md
@@ -15,6 +15,12 @@ Transform scattered insights from braindumps, daily briefs, and check-ins into c
- User wants to extract patterns from accumulated braindumps
- Before major decisions that could benefit from framework consultation
+## Agent Mode Awareness
+
+**Check `agent_mode` in `00-inbox/MY-PROFILE.md` frontmatter:**
+- If `agent_mode: team` — delegate scanning and pattern extraction to parallel sub-agents (e.g., one per domain: personal braindumps, professional braindumps, project-specific content, daily briefs). Each agent identifies themes and patterns, then a synthesis agent combines findings into frameworks.
+- If `agent_mode: solo` (default) — handle all scanning, pattern recognition, and framework building directly. No delegation.
+
## Process Flow
### 1. Data Gathering
diff --git a/.claude/skills/onboarding/SKILL.md b/.claude/skills/onboarding/SKILL.md
index cee437e..f305050 100644
--- a/.claude/skills/onboarding/SKILL.md
+++ b/.claude/skills/onboarding/SKILL.md
@@ -14,6 +14,17 @@ Welcome new users and collect essential information to personalize their COG exp
- User wants to update their profile or add new projects
- Any time profile customization is needed
+## Core Design Principle: Smart, Low-Friction Onboarding
+
+**The onboarding MUST feel like a natural conversation, NOT a form to fill out.**
+
+Key rules:
+- **Ask open-ended questions, not option-pickers.** Never present numbered lists of choices for the user to pick from.
+- **Ask as few questions as possible.** Infer what you can from context and the user's natural responses.
+- **Never ask redundant questions.** If you can extract the answer from something the user already said, don't ask again.
+- **Parse intelligently.** If someone says "I'm Alex, a PM at a fintech startup tracking Stripe and Plaid", extract: name=Alex, role=PM at fintech startup, watchlist=[Stripe, Plaid]. Don't ask follow-up questions for info already given.
+- **Confirm, don't re-ask.** If you're unsure about something the user said, confirm your interpretation rather than asking the question fresh.
+
## Process Flow
### 1. Welcome Message
@@ -21,68 +32,71 @@ Greet the user warmly and explain what COG is:
```
Welcome to COG - your self-evolving second brain powered by Claude + Obsidian + Git!
-COG helps you:
-- Capture thoughts and insights through brain dumps
-- Get daily intelligence briefings tailored to your interests
-- Build and consolidate knowledge over time
-- Track patterns in your thinking and development
-
-Before we begin, I'll ask you a few questions to personalize your experience. This will take about 3-5 minutes.
+COG helps you capture thoughts, get daily intelligence briefings, and build knowledge over time - all stored as simple markdown files you own.
-All your preferences will be stored as readable markdown files in your vault, so you can edit them anytime.
+Let's get you set up. Tell me a bit about yourself - your name, what you do, and what topics or areas you're most interested in staying sharp on. Feel free to share as much or as little as you'd like.
```
+**This single open-ended prompt replaces the old sequential questions.** The user can naturally mention their name, role, interests, sources, projects, and competitors all at once - or just share a few things.
+
### 2. Check for Existing Profile
Look for `00-inbox/MY-PROFILE.md`. If it exists:
```
-I found an existing profile! Would you like to:
-1. Update your profile
-2. Add new projects
-3. Update interest areas
-4. View current profile
-5. Start fresh (archive old profile)
-
-What would you like to do? (1-5)
+I found an existing COG profile! What would you like to update? Just tell me what you'd like to change - your interests, projects, profile info, or anything else.
```
-### 3. Information Collection (Keep it Simple!)
+**Don't present a numbered menu.** Let them describe what they want in natural language.
-Ask only essential questions in a conversational way:
+### 3. Intelligent Information Extraction
-**Question 1: What's your name?**
-- Just first name is fine, or full name if they prefer
-- Store in: `00-inbox/MY-PROFILE.md`
+After the user responds, extract as much as possible from their natural language:
-**Question 2: What do you do? (Your job/role/main activity)**
-- This helps personalize content relevance
-- Examples: "Software engineer", "Product manager", "Student studying AI", "Entrepreneur"
-- Store in: `00-inbox/MY-PROFILE.md`
+| Field | How to Extract |
+|-------|---------------|
+| **Name** | Look for self-introduction patterns ("I'm Alex", "My name is...", "Call me..."). Use first name by default. |
+| **Role** | Look for job/activity mentions ("I'm a PM", "I work in...", "software engineer at..."). |
+| **Interests** | Look for topic mentions ("interested in AI", "following crypto", "love design"). Also infer from role context. |
+| **News Sources** | Look for source mentions ("I read HN", "follow on Twitter"). If not mentioned, skip - it's optional. |
+| **Projects** | Look for project mentions ("working on a SaaS app", "building..."). If not mentioned, skip. |
+| **Competitive Watch** | Look for company/person mentions ("tracking Stripe", "watching what OpenAI does"). If not mentioned, skip. |
-**Question 3: What topics are you interested in?**
-- Ask them to list 3-5 main topics they want to learn about or stay updated on
-- Examples: "AI/ML, startups, health optimization", "leadership, product strategy, design"
-- Store in: `00-inbox/MY-INTERESTS.md`
-- Keep it natural - don't make them choose from categories
+### 4. Smart Follow-Up (Only If Needed)
-**Question 4: Where do you like to get your news and information?**
-- Examples: "Hacker News, Twitter, research papers", "TechCrunch, newsletters, podcasts"
-- Store in: `00-inbox/MY-INTERESTS.md` under "Preferred Sources"
-- This helps COG understand what sources to prioritize
+After extracting what you can, check what's missing from the **required** fields only:
+- **Name** (required)
+- **Role** (required)
+- **Interests** (required - need at least 2-3 topics)
-**Question 5: Do you have any active projects you're working on?**
-- Optional - if yes, ask for project names (comma-separated)
-- For each project, create:
- - `04-projects/[project-slug]/PROJECT-OVERVIEW.md` with basic structure
- - Full directory structure
-- If no projects, skip this entirely
+If any required field is missing, ask ONE follow-up that covers all gaps. For example:
+```
+Thanks! I got your name and role. What topics are you most interested in staying updated on? (e.g., AI, startups, design, health - whatever matters to you)
+```
-**Question 6: Any companies, competitors, or people you want to keep an eye on?** (Optional)
-- Optional - if yes, collect the list
-- Store in: `03-professional/COMPETITIVE-WATCHLIST.md`
-- Used for automatic extraction in braindumps
+**Optional fields** (news sources, projects, competitive watch) should NEVER generate follow-up questions. If the user didn't mention them, skip them. They can always add them later by editing the files or running onboarding again.
-### 4. Generate Profile Documents
+### 5. Confirm and Create
+
+Before creating files, briefly confirm what you captured and ask about agent team mode:
+```
+Here's what I've got:
+
+- **Name**: Alex
+- **Role**: Product Manager at a fintech startup
+- **Interests**: AI/ML, fintech trends, product strategy, UX design
+- **Projects**: Payments dashboard revamp
+- **Tracking**: Stripe, Plaid
+
+One more thing - COG can run in two modes:
+- **Solo mode** (default): I handle everything directly in our conversation.
+- **Agent team mode**: I delegate research, analysis, and writing to specialist sub-agents for deeper, more thorough results. Works best with Claude Code.
+
+Which do you prefer? (Solo is great for most people - team mode is for power users who want maximum depth.)
+```
+
+**Wait for confirmation**, then generate everything. If they say "looks good" or similar, proceed. If they correct something, update and proceed without re-confirming. Default to `solo` if they don't express a preference.
+
+### 6. Generate Profile Documents
Create the following markdown files:
@@ -92,6 +106,7 @@ Create the following markdown files:
type: profile
created: YYYY-MM-DD
onboarding_completed: true
+agent_mode: solo
tags: ["#profile", "#config", "#cog"]
---
@@ -102,13 +117,16 @@ tags: ["#profile", "#config", "#cog"]
- **Role**: [Job/role/main activity]
- **Profile Created**: [Date]
+## Settings
+- **Agent Mode**: solo *(solo = handle everything directly; team = delegate to specialist sub-agents for deeper results)*
+
## Active Projects
-[If they have projects:]
+[If they mentioned projects:]
- [[04-projects/[slug]/PROJECT-OVERVIEW|Project Name 1]]
- [[04-projects/[slug]/PROJECT-OVERVIEW|Project Name 2]]
[If no projects:]
-*No active projects yet. Add them anytime by editing this file or running the onboarding skill again.*
+*No active projects yet. Add them anytime by editing this file or running onboarding again.*
## Related
- [[MY-INTERESTS|My Interests & News Sources]]
@@ -142,11 +160,15 @@ tags: ["#interests", "#daily-brief", "#config"]
- [Topic 5]
## Preferred News Sources
+[If sources were mentioned:]
*Where I like to get information:*
- [Source 1]
- [Source 2]
- [Source 3]
+[If no sources mentioned:]
+*No specific sources set. COG will search broadly for your topics. Add preferred sources here anytime.*
+
## Notes
*Add any additional context about your interests here.*
@@ -155,7 +177,7 @@ tags: ["#interests", "#daily-brief", "#config"]
*Update this file anytime as your interests evolve. Just edit and save—COG will pick up the changes.*
```
-#### `03-professional/COMPETITIVE-WATCHLIST.md` (if applicable)
+#### `03-professional/COMPETITIVE-WATCHLIST.md` (only if they mentioned companies/people to track)
```markdown
---
type: competitive-intelligence
@@ -214,7 +236,7 @@ tags: ["#project", "#overview"]
*This overview helps COG organize your project-related thoughts and updates.*
```
-### 5. Create Directory Structure
+### 7. Create Directory Structure
Based on configuration, create personalized structure:
**Base Structure (Always):**
@@ -237,6 +259,7 @@ Based on configuration, create personalized structure:
consolidated/
patterns/
timeline/
+ booklets/
06-templates/
```
@@ -251,7 +274,7 @@ Based on configuration, create personalized structure:
resources/
```
-### 6. Create Welcome Guide
+### 8. Create Welcome Guide
Generate: `00-inbox/WELCOME-TO-COG.md`
@@ -334,53 +357,36 @@ When you use the braindump skill, select the project to automatically file your
*You can archive or delete this welcome guide once you're comfortable with COG.*
```
-### 7. First Action Prompts
-After setup, guide the user to their first action:
+### 9. Wrap-Up (No Menu!)
+After setup, summarize what was created and suggest a natural next action:
```
-Great! Your COG is now configured.
-
-I've created these profile documents for you:
-- MY-PROFILE.md (your basic preferences)
-- MY-INTERESTS.md (topics for daily briefs)
-[If applicable:] - COMPETITIVE-WATCHLIST.md (companies to track)
-[If applicable:] - PROJECT-OVERVIEW.md files for each project
-
-All files are in your vault and can be edited anytime.
+You're all set! I've created your profile, interests, and project files. Everything is in your vault and editable anytime.
-Would you like to:
-
-1. **Try your first braindump** - Capture what's on your mind right now
-2. **Get your daily brief** - See today's intelligence report
-3. **Review your profile** - Open MY-PROFILE.md to see/edit settings
-4. **Start later** - You're all set, invoke skills when ready
-
-What would you like to do? (1-4)
+If you want to jump right in, try a braindump - just tell me what's on your mind and I'll capture it. Or ask for your daily brief to see what's happening in your interest areas today.
```
+**Don't present a numbered menu of next actions.** Just suggest one or two natural things and let them decide.
+
## Configuration Update Mode
If user runs onboarding after initial setup (MY-PROFILE.md exists):
+Don't show a menu. Just ask:
```
-You've already completed onboarding! Would you like to:
-
-1. **Update your profile** - Edit MY-PROFILE.md with new preferences
-2. **Add new interests** - Update MY-INTERESTS.md with new topics
-3. **Add new projects** - Create new project structures
-4. **View current profile** - See your current MY-PROFILE.md
-
-What would you like to do? (1-4)
+You've already completed onboarding! What would you like to update? Just tell me what needs changing.
```
+Then intelligently handle whatever they say - whether it's adding projects, changing interests, updating their role, etc.
+
## Success Criteria
Onboarding is successful when:
-1. ✅ `MY-PROFILE.md` created in `00-inbox/`
-2. ✅ `MY-INTERESTS.md` created in `00-inbox/`
-3. ✅ Project directories and overviews created (if applicable)
-4. ✅ `WELCOME-TO-COG.md` guide created
-5. ✅ User understands next steps and where their profile is stored
+1. `MY-PROFILE.md` created in `00-inbox/`
+2. `MY-INTERESTS.md` created in `00-inbox/`
+3. Project directories and overviews created (if applicable)
+4. `WELCOME-TO-COG.md` guide created
+5. User understands next steps and where their profile is stored
## Error Handling
@@ -408,12 +414,12 @@ All configuration data is stored as markdown files in:
- `04-projects/[project]/PROJECT-OVERVIEW.md` - Project details
Benefits of markdown storage:
-- ✅ Human-readable and editable
-- ✅ Version controlled with Git
-- ✅ Searchable in Obsidian
-- ✅ Linkable from other notes
-- ✅ No parsing required, just read as text
-- ✅ Can be archived, moved, organized like any other note
+- Human-readable and editable
+- Version controlled with Git
+- Searchable in Obsidian
+- Linkable from other notes
+- No parsing required, just read as text
+- Can be archived, moved, organized like any other note
## Philosophy
diff --git a/.claude/skills/url-dump/SKILL.md b/.claude/skills/url-dump/SKILL.md
index 8dce5c6..bf30997 100644
--- a/.claude/skills/url-dump/SKILL.md
+++ b/.claude/skills/url-dump/SKILL.md
@@ -14,6 +14,12 @@ Transform raw URLs into structured, insightful knowledge entries through intelli
- User pastes a URL and wants to capture it
- User wants to organize web resources into their knowledge base
+## Agent Mode Awareness
+
+**Check `agent_mode` in `00-inbox/MY-PROFILE.md` frontmatter:**
+- If `agent_mode: team` — delegate content extraction, analysis, and categorization to a sub-agent while handling user interaction directly. The sub-agent fetches URL content, generates insights, and returns structured results for filing.
+- If `agent_mode: solo` (default) — handle everything directly in the conversation. No delegation.
+
## Pre-Flight Check
**Before executing, check for user profile:**
diff --git a/.claude/skills/weekly-checkin/SKILL.md b/.claude/skills/weekly-checkin/SKILL.md
index 3a8c974..b8f3d6c 100644
--- a/.claude/skills/weekly-checkin/SKILL.md
+++ b/.claude/skills/weekly-checkin/SKILL.md
@@ -14,6 +14,12 @@ Comprehensive weekly review and analysis integrating insights across all domains
- End of week reflection time
- User wants to analyze patterns across the week
+## Agent Mode Awareness
+
+**Check `agent_mode` in `00-inbox/MY-PROFILE.md` frontmatter:**
+- If `agent_mode: team` — delegate domain-specific analysis to sub-agents (e.g., one for personal patterns, one for professional review, one per active project). Each agent scans relevant braindumps/briefs and returns insights. Combine into the weekly synthesis.
+- If `agent_mode: solo` (default) — handle all scanning and analysis directly in the conversation. No delegation.
+
## Pre-Flight Check
**Check for user profile (optional but enhances experience):**
diff --git a/.gemini/commands/braindump.toml b/.gemini/commands/braindump.toml
new file mode 100644
index 0000000..efe7584
--- /dev/null
+++ b/.gemini/commands/braindump.toml
@@ -0,0 +1,6 @@
+description = "Quick capture raw thoughts with intelligent domain classification and competitive intelligence extraction"
+prompt = """
+You are running the COG braindump workflow. Follow this playbook exactly:
+
+@{.gemini/skills/braindump.md}
+"""
diff --git a/.gemini/commands/daily-brief.toml b/.gemini/commands/daily-brief.toml
new file mode 100644
index 0000000..7b13d48
--- /dev/null
+++ b/.gemini/commands/daily-brief.toml
@@ -0,0 +1,6 @@
+description = "Generate personalized news intelligence briefing with verified sources (7-day freshness)"
+prompt = """
+You are running the COG daily brief workflow. Follow this playbook exactly:
+
+@{.gemini/skills/daily-brief.md}
+"""
diff --git a/.gemini/commands/knowledge-consolidation.toml b/.gemini/commands/knowledge-consolidation.toml
new file mode 100644
index 0000000..c0b1967
--- /dev/null
+++ b/.gemini/commands/knowledge-consolidation.toml
@@ -0,0 +1,6 @@
+description = "Build frameworks from scattered insights across all braindumps and notes"
+prompt = """
+You are running the COG knowledge consolidation workflow. Follow this playbook exactly:
+
+@{.gemini/skills/knowledge-consolidation.md}
+"""
diff --git a/.gemini/commands/onboarding.toml b/.gemini/commands/onboarding.toml
new file mode 100644
index 0000000..3ca27dd
--- /dev/null
+++ b/.gemini/commands/onboarding.toml
@@ -0,0 +1,6 @@
+description = "Set up your COG profile - name, role, interests, projects, and preferences"
+prompt = """
+You are running the COG onboarding workflow. Follow this playbook exactly:
+
+@{.gemini/skills/onboarding.md}
+"""
diff --git a/.gemini/commands/url-dump.toml b/.gemini/commands/url-dump.toml
new file mode 100644
index 0000000..26d57a1
--- /dev/null
+++ b/.gemini/commands/url-dump.toml
@@ -0,0 +1,6 @@
+description = "Quick capture URLs with automatic content extraction, insights, and categorization"
+prompt = """
+You are running the COG URL dump workflow. Follow this playbook exactly:
+
+@{.gemini/skills/url-dump.md}
+"""
diff --git a/.gemini/commands/weekly-checkin.toml b/.gemini/commands/weekly-checkin.toml
new file mode 100644
index 0000000..6e2e478
--- /dev/null
+++ b/.gemini/commands/weekly-checkin.toml
@@ -0,0 +1,6 @@
+description = "Cross-domain pattern analysis and strategic reflection for weekly review"
+prompt = """
+You are running the COG weekly check-in workflow. Follow this playbook exactly:
+
+@{.gemini/skills/weekly-checkin.md}
+"""
diff --git a/.gemini/skills/braindump.md b/.gemini/skills/braindump.md
new file mode 100644
index 0000000..4f77d9d
--- /dev/null
+++ b/.gemini/skills/braindump.md
@@ -0,0 +1,47 @@
+# COG Braindump Playbook
+
+## Goal
+Capture raw thoughts and transform them into structured, classified intelligence.
+
+## Pre-Flight
+1. Read `00-inbox/MY-PROFILE.md` for name, role, projects, agent_mode
+2. Read `03-professional/COMPETITIVE-WATCHLIST.md` for tracked entities
+3. If no profile exists, suggest running `/onboarding` first
+
+## Steps
+
+### 1. Prompt for Input
+Ask the user to share their thoughts. Accept any format — stream of consciousness is fine.
+
+### 2. Classify Domain
+Analyze content and classify into:
+- **Personal** → `02-personal/braindumps/`
+- **Professional** → `03-professional/braindumps/`
+- **Project-specific** → `04-projects/[project-slug]/braindumps/`
+- **Mixed/unclear** → `00-inbox/`
+
+### 3. Extract & Analyze
+From the raw input, extract:
+- **Key themes** (3-5 main topics)
+- **Questions raised** (explicit or implicit)
+- **Decisions needed** (with options if apparent)
+- **Action items** with Obsidian Tasks dates: `- [ ] Task 📅 YYYY-MM-DD`
+- **Competitive intelligence** (if entities from watchlist are mentioned)
+- **Emotional context** (energy level, stress indicators)
+- **Connections** to previous braindumps or knowledge
+
+### 4. Generate Output File
+Save to appropriate domain folder as `braindump-YYYY-MM-DD-[slug].md` with:
+```yaml
+type: braindump
+created: YYYY-MM-DD
+domain: personal|professional|project
+project: [if applicable]
+themes: [list]
+tags: [auto-generated]
+```
+
+Include sections: Raw Capture, Key Themes, Strategic Insights, Questions & Decisions, Action Items, Connections
+
+### 5. Extract Competitive Intelligence
+If watchlist entities are mentioned, save intel to `04-projects/[project]/competitive/[entity-slug].md`
diff --git a/.gemini/skills/daily-brief.md b/.gemini/skills/daily-brief.md
new file mode 100644
index 0000000..48be695
--- /dev/null
+++ b/.gemini/skills/daily-brief.md
@@ -0,0 +1,50 @@
+# COG Daily Brief Playbook
+
+## Goal
+Generate a personalized news intelligence briefing with verified sources.
+
+## Pre-Flight
+1. Read `00-inbox/MY-INTERESTS.md` for topics and preferred sources
+2. Read `00-inbox/MY-PROFILE.md` for role context and agent_mode
+3. Read `03-professional/COMPETITIVE-WATCHLIST.md` for entities to track
+4. If no interests file, suggest running `/onboarding` first
+
+## Verification Standards (MANDATORY)
+- All news MUST be from within the last 7 days
+- Minimum 2 credible sources per claim
+- Confidence levels explicitly stated (High/Medium/Low)
+- Source credibility assessed: Tier 1 (major outlets), Tier 2 (trade press), Tier 3 (blogs/social)
+
+## Steps
+
+### 1. Research
+Search for news across user's interest areas using web search tools. Prioritize:
+- User's preferred sources (from MY-INTERESTS.md)
+- Tier 1 and Tier 2 sources
+- Topics directly relevant to user's role and projects
+
+### 2. Verify & Filter
+- Confirm dates are within 7 days
+- Cross-reference claims across multiple sources
+- Discard unverifiable or outdated information
+- Assess strategic relevance to user's role
+
+### 3. Analyze
+For each story:
+- Strategic relevance to user's interests/projects
+- Opportunities and threats
+- Competitive implications (if watchlist entities involved)
+- Recommended actions
+
+### 4. Generate Brief
+Save to `01-daily/briefs/daily-brief-YYYY-MM-DD.md` with:
+```yaml
+type: daily-brief
+created: YYYY-MM-DD
+topics_covered: [list]
+sources_count: N
+confidence: high|medium|mixed
+tags: ["#daily-brief", "#intelligence"]
+```
+
+Include sections: Top Stories, Industry Updates, Competitive Intelligence, Opportunities & Threats, Action Items
diff --git a/.gemini/skills/knowledge-consolidation.md b/.gemini/skills/knowledge-consolidation.md
new file mode 100644
index 0000000..6bf1c5d
--- /dev/null
+++ b/.gemini/skills/knowledge-consolidation.md
@@ -0,0 +1,51 @@
+# COG Knowledge Consolidation Playbook
+
+## Goal
+Transform scattered insights from braindumps, briefs, and check-ins into coherent frameworks and knowledge documents.
+
+## Pre-Flight
+1. Read `00-inbox/MY-PROFILE.md` for context and agent_mode
+2. Identify last consolidation date (check `05-knowledge/consolidated/` for most recent report)
+
+## Steps
+
+### 1. Scan Vault
+Gather all unprocessed content since last consolidation:
+- `02-personal/braindumps/`
+- `03-professional/braindumps/`
+- `04-projects/*/braindumps/`
+- `01-daily/briefs/`
+- `01-daily/checkins/`
+
+### 2. Pattern Recognition
+Apply multi-layer analysis:
+- **Frequency patterns**: Topics that appear 3+ times
+- **Temporal patterns**: How thinking evolves over time
+- **Domain correlations**: Insights that span personal + professional
+- **Contradictions**: Where thinking conflicts across notes
+
+### 3. Framework Building
+For significant patterns, create or update frameworks:
+- Name the pattern/framework
+- Define its components
+- Provide actionable application guidelines
+- Link to source braindumps
+
+### 4. Generate Outputs
+
+**Frameworks** → `05-knowledge/consolidated/[framework-name]-framework.md`
+```yaml
+type: framework
+created: YYYY-MM-DD
+sources: [list of source braindumps]
+tags: ["#framework", "#knowledge"]
+```
+
+**Patterns** → `05-knowledge/patterns/pattern-[name].md`
+
+**Timeline** → `05-knowledge/timeline/[topic]-evolution-YYYY-MM.md`
+
+**Report** → `05-knowledge/consolidated/consolidation-YYYY-MM-DD.md`
+
+### 5. Mark Processed
+Add `consolidated: YYYY-MM-DD` to frontmatter of processed braindumps.
diff --git a/.gemini/skills/onboarding.md b/.gemini/skills/onboarding.md
new file mode 100644
index 0000000..acf504a
--- /dev/null
+++ b/.gemini/skills/onboarding.md
@@ -0,0 +1,65 @@
+# COG Onboarding Playbook
+
+## Goal
+Welcome the user and create their COG profile through natural conversation.
+
+## Design Principle
+Ask open-ended questions, not option menus. Infer what you can. Never ask redundant questions.
+
+## Steps
+
+### 1. Check for Existing Profile
+- Look for `00-inbox/MY-PROFILE.md`
+- If exists: Ask "What would you like to update? Just tell me what needs changing."
+- If not: Proceed with setup below
+
+### 2. Single Open-Ended Prompt
+```
+Welcome to COG - your self-evolving second brain!
+
+Tell me a bit about yourself - your name, what you do, and what topics you're most interested in staying sharp on. Feel free to share as much or as little as you'd like.
+```
+
+### 3. Extract from Response
+Required: **Name**, **Role**, **Interests** (2-3 minimum)
+Optional (only if mentioned): News Sources, Projects, Competitive Watchlist
+
+### 4. Follow-Up (Only If Needed)
+Only ask ONE follow-up if required fields are missing. Never ask about optional fields.
+
+### 5. Ask About Agent Mode
+During confirmation, ask:
+- **Solo** (default): Handle everything directly in the conversation
+- **Team**: Delegate to specialist sub-agents for deeper results
+
+### 6. Confirm and Create
+Show summary, wait for confirmation, then create:
+
+**`00-inbox/MY-PROFILE.md`** with frontmatter:
+```yaml
+type: profile
+created: YYYY-MM-DD
+onboarding_completed: true
+agent_mode: solo
+tags: ["#profile", "#config", "#cog"]
+```
+Include: Name, Role, Profile Created date, Agent Mode setting, Active Projects (if any)
+
+**`00-inbox/MY-INTERESTS.md`** with topics and preferred sources
+
+**`03-professional/COMPETITIVE-WATCHLIST.md`** (only if they mentioned competitors)
+
+**`04-projects/[project-slug]/PROJECT-OVERVIEW.md`** for each mentioned project
+
+### 7. Create Directory Structure
+```
+00-inbox/, 01-daily/briefs/ + checkins/, 02-personal/braindumps/ + development/ + wellness/
+03-professional/braindumps/ + leadership/ + strategy/ + skills/
+04-projects/, 05-knowledge/consolidated/ + patterns/ + timeline/ + booklets/, 06-templates/
+```
+
+### 8. Create Welcome Guide
+Generate `00-inbox/WELCOME-TO-COG.md` with quick start instructions.
+
+### 9. Wrap Up
+Suggest a natural next action (braindump or daily brief). No numbered menus.
diff --git a/.gemini/skills/url-dump.md b/.gemini/skills/url-dump.md
new file mode 100644
index 0000000..34f7f1c
--- /dev/null
+++ b/.gemini/skills/url-dump.md
@@ -0,0 +1,55 @@
+# COG URL Dump Playbook
+
+## Goal
+Transform URLs into structured knowledge entries with automatic content extraction, insights, and categorization.
+
+## Pre-Flight
+1. Read `00-inbox/MY-PROFILE.md` for interests, projects, agent_mode
+2. Read `00-inbox/MY-INTERESTS.md` for relevance assessment
+
+## Steps
+
+### 1. Accept URL(s)
+User provides one or more URLs. Validate they are accessible.
+
+### 2. Fetch & Extract
+For each URL:
+- Fetch page content
+- Extract: title, author, publication date, main content
+- Identify content type (article, tool, reference, research, video, etc.)
+
+### 3. Categorize
+Auto-assign to one of:
+- Articles & Blogs
+- Tools & Resources
+- Reference & Documentation
+- Research & Papers
+- Inspiration & Design
+- Videos & Media
+- News & Updates
+- Project-Specific (if relevant to an active project)
+
+### 4. Analyze
+Generate:
+- **Summary** (2-3 sentences)
+- **Key Insights** (3-5 bullet points)
+- **Relevance** to user's interests and projects
+- **Action items** if applicable: `- [ ] Task 📅 YYYY-MM-DD`
+- **Related** links to existing vault content
+
+### 5. Save
+**Standard** → `05-knowledge/booklets/[category]/[title-slug]-YYYY-MM-DD.md`
+**Project-specific** → `04-projects/[project-slug]/resources/[title-slug].md`
+**Unclear** → `00-inbox/[title-slug].md`
+
+File format:
+```yaml
+type: booklet
+created: YYYY-MM-DD
+source_url: [URL]
+category: [category]
+author: [if known]
+tags: [auto-generated]
+```
+
+Include sections: Summary, Key Insights, Relevance, Full Notes, Source Info
diff --git a/.gemini/skills/weekly-checkin.md b/.gemini/skills/weekly-checkin.md
new file mode 100644
index 0000000..b8a68ec
--- /dev/null
+++ b/.gemini/skills/weekly-checkin.md
@@ -0,0 +1,52 @@
+# COG Weekly Check-in Playbook
+
+## Goal
+Comprehensive weekly review integrating insights across all domains with pattern recognition.
+
+## Pre-Flight
+1. Read `00-inbox/MY-PROFILE.md` for name, role, projects, agent_mode
+2. Scan recent content from the past week:
+ - `02-personal/braindumps/` (personal domain)
+ - `03-professional/braindumps/` (professional domain)
+ - `04-projects/*/braindumps/` (each active project)
+ - `01-daily/briefs/` (daily briefs from this week)
+
+## Steps
+
+### 1. Gather Context
+Read all braindumps, briefs, and notes from the past 7 days.
+
+### 2. Guide Reflection
+Ask the user conversational reflection questions:
+- How was your week overall? (energy, mood, productivity)
+- What went well? What was challenging?
+- Any breakthroughs or insights?
+
+### 3. Domain Review
+For each active domain, summarize:
+- **Personal**: Wellness, growth, personal goals
+- **Professional**: Accomplishments, challenges, learnings
+- **Projects** (each): Progress, blockers, next steps
+
+### 4. Pattern Recognition
+Identify patterns across the week:
+- Recurring themes across braindumps
+- Cross-domain connections
+- Energy and productivity patterns
+- Evolving priorities
+
+### 5. Forward Planning
+Help set priorities for next week:
+- Top 3 priorities
+- Action items with dates: `- [ ] Task 📅 YYYY-MM-DD`
+- Carry-over items from this week
+
+### 6. Generate Check-in
+Save to `01-daily/checkins/weekly-checkin-YYYY-MM-DD.md` with:
+```yaml
+type: weekly-checkin
+created: YYYY-MM-DD
+week_rating: N/10
+domains_reviewed: [list]
+tags: ["#weekly-checkin", "#reflection"]
+```
diff --git a/.kiro/powers/cog-onboarding/POWER.md b/.kiro/powers/cog-onboarding/POWER.md
index 5751e08..f464f86 100644
--- a/.kiro/powers/cog-onboarding/POWER.md
+++ b/.kiro/powers/cog-onboarding/POWER.md
@@ -7,7 +7,7 @@ keywords: ["onboarding", "setup COG", "setup profile", "get started", "configure
# COG Onboarding Power
-Welcome new users and collect essential information to personalize their COG (Claude + Obsidian + Git) second brain experience.
+Welcome new users and collect essential information to personalize their COG (Cognition + Obsidian + Git) second brain experience.
## When This Power Activates
@@ -15,35 +15,66 @@ Welcome new users and collect essential information to personalize their COG (Cl
- User is new and hasn't completed onboarding yet
- User wants to update their profile or add new projects
+## Core Principle: Smart, Low-Friction Onboarding
+
+**The onboarding MUST feel like a natural conversation, NOT a form.**
+
+- Ask open-ended questions, never numbered option lists
+- Ask as few questions as possible - infer from natural responses
+- Never ask redundant questions - if info was already given, don't re-ask
+- Parse intelligently - extract name, role, interests, projects, watchlist from a single paragraph if the user provides them
+- Confirm your interpretation rather than asking the question fresh
+
## Onboarding Steps
### 1. Check for Existing Profile
Look for `00-inbox/MY-PROFILE.md` in the vault:
-- If exists: Offer update mode (update profile, add interests, add projects, view current, start fresh)
+- If exists: Ask "What would you like to update? Just tell me what needs changing." (no numbered menus)
- If not found: Proceed with full onboarding
-### 2. Collect Information (Keep it Simple!)
+### 2. Single Open-Ended Prompt
+
+Instead of asking 6 sequential questions, start with ONE open-ended message:
+
+```
+Welcome to COG - your self-evolving second brain!
+
+Tell me a bit about yourself - your name, what you do, and what topics you're most interested in staying sharp on. Feel free to share as much or as little as you'd like.
+```
+
+### 3. Intelligent Extraction
+
+From the user's response, extract:
+- **Name** (required) - from self-introduction patterns
+- **Role** (required) - from job/activity mentions
+- **Interests** (required, 2-3 minimum) - from topic mentions, also infer from role
+- **News Sources** (optional) - from source mentions, skip if not mentioned
+- **Projects** (optional) - from project mentions, skip if not mentioned
+- **Competitive Watch** (optional) - from company/person mentions, skip if not mentioned
+
+### 4. Smart Follow-Up (Only If Needed)
+
+Only ask a follow-up if required fields (name, role, interests) are missing. Ask ONE question covering all gaps. Never ask about optional fields the user didn't mention.
+
+### 5. Confirm and Create
-Ask only essential questions conversationally:
+Show a brief summary of what was captured. Also ask about agent mode:
+- **Solo mode** (default): Handle everything in a single conversation
+- **Agent team mode**: Delegate to specialist sub-agents for deeper results (works best with Claude Code)
-1. **Name** - First name or full name
-2. **Role** - Job/role/main activity (e.g., "Software engineer", "Product manager")
-3. **Interests** - 3-5 topics to learn about or stay updated on
-4. **News Sources** - Preferred information sources (e.g., "Hacker News, Twitter")
-5. **Active Projects** - Optional, comma-separated project names
-6. **Competitive Watch** - Optional, companies/people to track
+Default to `solo` if user doesn't express a preference. Store as `agent_mode: solo|team` in MY-PROFILE.md frontmatter.
-### 3. Generate Profile Documents
+### 6. Generate Profile Documents
Create these markdown files:
**`00-inbox/MY-PROFILE.md`** - Basic profile with name, role, active projects
**`00-inbox/MY-INTERESTS.md`** - Topics and preferred news sources
-**`03-professional/COMPETITIVE-WATCHLIST.md`** - If tracking competitors
-**`04-projects/[project-slug]/PROJECT-OVERVIEW.md`** - For each project
+**`03-professional/COMPETITIVE-WATCHLIST.md`** - Only if they mentioned competitors
+**`04-projects/[project-slug]/PROJECT-OVERVIEW.md`** - Only if they mentioned projects
-### 4. Create Directory Structure
+### 7. Create Directory Structure
```
00-inbox/
@@ -55,9 +86,9 @@ Create these markdown files:
06-templates/
```
-### 5. Generate Welcome Guide
+### 8. Generate Welcome Guide and Wrap Up
-Create `00-inbox/WELCOME-TO-COG.md` with quick start instructions.
+Create `00-inbox/WELCOME-TO-COG.md` with quick start instructions. Suggest a natural next action (braindump or daily brief) without presenting a numbered menu.
## Success Criteria
diff --git a/agents.md b/AGENTS.md
similarity index 87%
rename from agents.md
rename to AGENTS.md
index 1298504..74b9728 100644
--- a/agents.md
+++ b/AGENTS.md
@@ -10,7 +10,7 @@ This document defines the available commands/skills for AI agents interacting wi
### /onboarding
-**Description:** Personalize COG for your workflow - creates profile, interests, and watchlist files with guided setup.
+**Description:** Personalize COG for your workflow - creates profile, interests, and watchlist files with a smart, conversational setup.
**Triggers:**
- `/onboarding`
@@ -19,15 +19,24 @@ This document defines the available commands/skills for AI agents interacting wi
- "setup my profile"
- "get started"
-**Purpose:** Welcome new users and collect essential information to personalize their COG experience. Creates profile documents stored as markdown files within the vault.
+**Purpose:** Welcome new users and collect essential information to personalize their COG experience through natural conversation - not sequential form-filling. Creates profile documents stored as markdown files within the vault.
-**What it does:**
-1. Collects user name, role, interests, and preferred news sources
-2. Creates `00-inbox/MY-PROFILE.md` with basic preferences
-3. Creates `00-inbox/MY-INTERESTS.md` with topics for daily briefs
-4. Optionally creates project structures in `04-projects/`
-5. Optionally creates `03-professional/COMPETITIVE-WATCHLIST.md`
-6. Generates a welcome guide
+**How it works:**
+1. Asks ONE open-ended question: "Tell me about yourself - name, role, and what you're interested in"
+2. Intelligently parses the response to extract name, role, interests, projects, news sources, and competitive watchlist
+3. Only asks a follow-up if required info (name, role, interests) is still missing
+4. Asks about agent mode preference (solo vs team) during confirmation
+5. Confirms extracted info before creating files
+6. Creates `00-inbox/MY-PROFILE.md` with basic preferences and agent_mode setting
+7. Creates `00-inbox/MY-INTERESTS.md` with topics for daily briefs
+8. Optionally creates project structures in `04-projects/` and `03-professional/COMPETITIVE-WATCHLIST.md` (only if mentioned)
+9. Generates a welcome guide
+
+**Agent modes:**
+- **Solo** (default): All skills handle everything directly in one conversation
+- **Team**: Skills delegate research, analysis, and writing to specialist sub-agents for deeper results (works best with Claude Code)
+
+**Design principle:** Never ask redundant questions. Never show numbered option menus. Infer what you can from context.
**Run this first** if you're new to COG.
diff --git a/GEMINI.md b/GEMINI.md
new file mode 100644
index 0000000..5c78657
--- /dev/null
+++ b/GEMINI.md
@@ -0,0 +1,38 @@
+# COG: Agentic Second Brain
+
+You are operating inside a **COG second brain** — a self-evolving knowledge management system built on Obsidian markdown files and Git.
+
+## Your Role
+
+You are the user's personal knowledge agent. Help them capture thoughts, stay informed, reflect, and build knowledge — all stored as plain markdown files they own.
+
+## Vault Structure
+
+```
+00-inbox/ → Landing zone, profile files (MY-PROFILE.md, MY-INTERESTS.md)
+01-daily/ → briefs/, checkins/
+02-personal/ → braindumps/
+03-professional/ → braindumps/, COMPETITIVE-WATCHLIST.md
+04-projects/ → [project-slug]/ with braindumps/, competitive/, resources/
+05-knowledge/ → consolidated/, patterns/, timeline/, booklets/
+06-templates/ → Document templates
+```
+
+## User Profile
+
+Read `00-inbox/MY-PROFILE.md` for user name, role, and active projects.
+Read `00-inbox/MY-INTERESTS.md` for topics and preferred news sources.
+Read `03-professional/COMPETITIVE-WATCHLIST.md` for companies to track.
+
+## Available Skills
+
+Use `/onboarding`, `/braindump`, `/daily-brief`, `/weekly-checkin`, `/knowledge-consolidation`, or `/url-dump` to invoke COG skills. Each has a detailed playbook in `.gemini/skills/`.
+
+## Rules
+
+- All output files use Obsidian-compatible markdown with YAML frontmatter
+- Tasks use [Obsidian Tasks emoji format](https://publish.obsidian.md/tasks/Reference/Task+Formats/Tasks+Emoji+Format): `- [ ] Task 📅 YYYY-MM-DD`
+- News must be verified with sources and within last 7 days
+- Respect domain separation: personal, professional, project-specific
+- Never fabricate sources or dates
+- All files are editable by the user — treat configuration as knowledge
diff --git a/README.md b/README.md
index 3fa514a..3ce9dad 100644
--- a/README.md
+++ b/README.md
@@ -2,7 +2,7 @@
**COG = Cognition + Obsidian + Git**
-A self-evolving second brain powered by AI agents, Obsidian markdown files, and Git version control. Works with **Claude Code**, **Kiro**, **OpenAI agents**, and any AI that reads markdown. No database, no vendor lock-in—just `.md` files that think.
+A self-evolving second brain powered by AI agents, Obsidian markdown files, and Git version control. Works with **Claude Code**, **Kiro**, **Gemini CLI**, **OpenAI Codex**, and any AI that reads markdown. No database, no vendor lock-in—just `.md` files that think.
## What is COG?
@@ -13,7 +13,7 @@ COG is a personal productivity and knowledge management system that:
- 📝 **Plain text**: Just markdown files (no database, no vendor lock-in)
- 📱 **Multi-device**: Syncs via iCloud to iPhone, iPad, Mac
- 🔒 **Privacy-first**: Your data, your files, your control
-- 🤖 **Agent-agnostic**: Works with Claude Code, Kiro, OpenAI, and more
+- 🤖 **Agent-agnostic**: Works with Claude Code, Kiro, Gemini CLI, OpenAI Codex, and more
- 🎯 **Specialized skills**: Custom AI capabilities for specific workflows
- ✅ **Verification-first**: No AI hallucinations, sources required
- 📋 **Obsidian Tasks compatible**: Tasks include `📅 YYYY-MM-DD` emoji format for dashboard queries
@@ -25,7 +25,9 @@ COG is a personal productivity and knowledge management system that:
- **AI Agent** (choose one or more):
- [Claude Code](https://claude.ai/download) - Uses `.claude/skills/`
- [Kiro](https://kiro.dev/) - Uses `.kiro/powers/`
- - Any OpenAI-compatible agent - Uses `agents.md`
+ - [Gemini CLI](https://github.com/google-gemini/gemini-cli) - Uses `GEMINI.md` + `.gemini/commands/`
+ - [OpenAI Codex](https://github.com/openai/codex) - Uses `AGENTS.md`
+ - Any other AI agent - Uses `AGENTS.md`
- [Obsidian](https://obsidian.md/) - Recommended for viewing/editing your notes
- Git - Already on your system if you can run the clone command below
@@ -45,8 +47,14 @@ COG is a personal productivity and knowledge management system that:
# Kiro:
# Open folder, mention "onboarding" or "setup COG"
+ # Gemini CLI:
+ gemini # Then type: /onboarding
+
+ # OpenAI Codex:
+ codex # Then ask: "Run onboarding"
+
# Other agents:
- # Reference agents.md and ask to run onboarding
+ # Reference AGENTS.md and ask to run onboarding
```
That's it! The onboarding will personalize COG for your needs in 2-3 minutes.
@@ -56,7 +64,8 @@ That's it! The onboarding will personalize COG for your needs in 2-3 minutes.
- Skills are defined in multiple formats for agent compatibility:
- `.claude/skills/` - 6 Claude Code skills
- `.kiro/powers/` - 6 Kiro powers
- - `agents.md` - Universal agent documentation
+ - `GEMINI.md` + `.gemini/commands/` - 6 Gemini CLI commands
+ - `AGENTS.md` - Universal agent documentation (OpenAI Codex and others)
- Onboarding will create your personalized directory structure
**Want Obsidian integration?**
@@ -87,18 +96,24 @@ COG provides specialized AI capabilities that work across different agents. The
**How Skills Work:**
-Skills are self-contained AI capabilities that combine orchestration and specialized processing. Unlike the old command/subagent split, each skill now:
-- Automatically activates when Claude detects relevant user intent
+Skills are self-contained AI capabilities that combine orchestration and specialized processing. Each skill:
+- Automatically activates when the AI agent detects relevant user intent
- Handles the complete workflow from user interaction to file creation
- Contains all necessary instructions, analysis frameworks, and verification protocols
-- Uses progressive disclosure to load information only as needed
+- Works across all supported agents (Claude Code, Kiro, Gemini CLI, OpenAI Codex)
For example:
-- Ask "What's on my mind?" → Claude invokes the **braindump** skill
-- Say "Give me my daily brief" → Claude invokes the **daily-brief** skill
-- Request "Weekly review" → Claude invokes the **weekly-checkin** skill
-
-This architecture leverages Claude Code's native skill system for better performance and simpler maintenance.
+- Ask "What's on my mind?" → Agent invokes the **braindump** skill
+- Say "Give me my daily brief" → Agent invokes the **daily-brief** skill
+- Request "Weekly review" → Agent invokes the **weekly-checkin** skill
+
+**Agent-Specific Formats:**
+| Agent | Skill Format | Invocation |
+|-------|-------------|------------|
+| Claude Code | `.claude/skills/*/SKILL.md` | Auto-detected from intent |
+| Kiro | `.kiro/powers/*/POWER.md` | Auto-detected from intent |
+| Gemini CLI | `.gemini/commands/*.toml` → `.gemini/skills/*.md` | `/command` slash commands |
+| OpenAI Codex | `AGENTS.md` | Natural language request |
## Key Features
@@ -165,7 +180,8 @@ Tasks automatically appear in:
```
your-obsidian-vault/
-├── agents.md # Universal agent documentation (OpenAI, etc.)
+├── AGENTS.md # Universal agent documentation (OpenAI Codex, etc.)
+├── GEMINI.md # Gemini CLI context file
├── .claude/
│ └── skills/ # 6 Claude Code skills
│ ├── onboarding/
@@ -174,6 +190,15 @@ your-obsidian-vault/
│ ├── weekly-checkin/
│ ├── knowledge-consolidation/
│ └── url-dump/
+├── .gemini/
+│ ├── commands/ # 6 Gemini CLI slash commands
+│ │ ├── onboarding.toml
+│ │ ├── braindump.toml
+│ │ ├── daily-brief.toml
+│ │ ├── weekly-checkin.toml
+│ │ ├── knowledge-consolidation.toml
+│ │ └── url-dump.toml
+│ └── skills/ # 6 Gemini skill playbooks
├── .kiro/
│ └── powers/ # 6 Kiro powers
│ ├── cog-onboarding/
@@ -217,22 +242,20 @@ your-obsidian-vault/
In Claude Code, ask: "Run onboarding" or "Set up my COG profile"
-The onboarding skill will ask:
-- Your name
-- What you do (your job/role)
-- Topics you're interested in (3-5 main areas)
-- Where you get your news (preferred sources)
-- Active projects you're working on (optional)
-- Companies/people you want to track (optional)
+The onboarding skill asks ONE open-ended question: "Tell me about yourself." It intelligently extracts your name, role, interests, projects, and competitive watchlist from your natural response — no sequential form-filling. It also asks about **agent mode** (solo vs team) for power users.
**Takes 2 minutes. Everything is stored as readable markdown files.**
Onboarding creates:
-- `00-inbox/MY-PROFILE.md` - Your basic info and projects
+- `00-inbox/MY-PROFILE.md` - Your basic info, projects, and agent mode setting
- `00-inbox/MY-INTERESTS.md` - Your topics and news sources
- `03-professional/COMPETITIVE-WATCHLIST.md` - Companies/people you're tracking (if any)
- `04-projects/[project-name]/` folders for each active project
+**Agent modes** (set during onboarding, changeable anytime in MY-PROFILE.md):
+- **Solo** (default): Skills handle everything directly in one conversation
+- **Team**: Skills delegate research, analysis, and writing to specialist sub-agents for deeper results (works best with Claude Code)
+
After onboarding, COG will:
- Personalize daily briefs to your topics
- Offer your projects when you braindump
@@ -395,11 +418,12 @@ After 3 months of use:
A: COG adds self-evolving intelligence. It doesn't just store—it learns, analyzes, and synthesizes insights automatically.
**Q: Do I need Claude Code specifically?**
-A: No! COG now supports multiple AI agents:
+A: No! COG supports multiple AI agents with native integrations:
- **Claude Code**: Uses `.claude/skills/` (native support)
- **Kiro**: Uses `.kiro/powers/` (native support)
-- **OpenAI/Other agents**: Use `agents.md` for skill documentation
-- Any AI that reads markdown can use COG's skills
+- **Gemini CLI**: Uses `GEMINI.md` + `.gemini/commands/` (native slash commands)
+- **OpenAI Codex**: Uses `AGENTS.md` (auto-loaded at startup)
+- Any AI that reads markdown can use COG's skills via `AGENTS.md`
**Q: How much does it cost?**
A: COG is free (MIT license). You only pay for Claude Code usage (Anthropic's pricing).
@@ -437,6 +461,8 @@ COG is free and open-source (MIT license). Your support helps keep it that way w
Built with:
- [Claude Code](https://claude.ai/code) by Anthropic
- [Kiro](https://kiro.dev/) by AWS
+- [Gemini CLI](https://github.com/google-gemini/gemini-cli) by Google
+- [OpenAI Codex](https://github.com/openai/codex) by OpenAI
- [Obsidian](https://obsidian.md/) for markdown management
- Inspiration from Zettelkasten, Building a Second Brain, and GTD methodologies
@@ -446,6 +472,6 @@ Built with:
---
-**TL;DR:** COG is a self-evolving agentic second brain using AI agents + Obsidian + Git. Works with Claude Code, Kiro, OpenAI, and more. Just `.md` files that sync via iCloud and learn your patterns. Setup in 2 steps, zero maintenance required.
+**TL;DR:** COG is a self-evolving agentic second brain using AI agents + Obsidian + Git. Works with Claude Code, Kiro, Gemini CLI, OpenAI Codex, and more. Just `.md` files that sync via iCloud and learn your patterns. Setup in 2 steps, zero maintenance required.
⭐ Star this repo if you find it useful!