Capture & Organize Personal Knowledge19 skills
Skills for daily thought capture, news intelligence, and URL saving with domain-based organization.
Knowledge Management Blueprint
Build a Self-Evolving Second Brain With AI
huytieuMay 2026
A bundle of 17 skills and 6 worker agents that turn markdown notes into an agentic second brain with Git and multi-agent support.
Personal Knowledge CaptureTeam Intelligence & SyncProduct Management WorkflowStrategic ResearchSpecialist WorkersPeople & Team Profiles
Install Command
Run this command to deploy the blueprint to your environment.
What problem does this solve?
Turns your scattered notes and daily thoughts into an evolving knowledge system that learns your patterns and synthesizes insights automatically. Most knowledge systems require constant manual organization, external databases, and lock-in to a single platform. COG keeps everything local as markdown files while AI agents handle the learning and connection-building.
How does it work?
1. Clone the repo and run onboarding in Claude Code, Kiro, Gemini CLI, or any markdown-reading agent. 2. Start capturing thoughts using braindump, daily briefs, and URL saves. All are saved as markdown in organized domains. 3. Specialist worker agents (6 parallel Sonnet instances) handle data collection from GitHub, Linear, Slack, and PostHog while your main agent does reasoning. 4. Weekly consolidation and monthly synthesis build frameworks from scattered notes. Output: a self-organizing vault that connects your raw thinking to strategic insights and team intelligence.
Included Skill Groups
Generate Cross-Platform Team Intelligence6 skills
Skills for engineering and product leaders to synthesize GitHub, Linear, Slack, PostHog, and meeting data into unified intelligence briefs.
Manage Product Lifecycle End to End7 skills
Skills for product managers to generate PRDs, create user stories, release notes, and manage knowledge bases with integrations to Linear, Jira, and Confluence.
Conduct Deep Multi-Agent Research3 skills
A research skill that decomposes complex questions into parallel research threads with web search, source citation, and multi-agent investigation.
Run Parallel Data-Heavy Operations5 skills
Six worker agents that handle structured data extraction, file operations, and publishing while main agent handles reasoning.
Track Team Profiles With Evidence-Based Tiers1 skill
A people relationship system that builds profiles from meeting notes, braindumps, and team interactions with progressive tiered enrichment (Tier 3, 2, or 1).
Personalize With Role Packs & Integration Discovery5 skills
Onboarding system that matches your role to curated skill recommendations and integrations, with guided discovery for external tools.
What's the biggest win?
Every insight you capture feeds into larger patterns automatically, with no manual cross-referencing or external tool switching required.
What should I know technically?
Built for Claude 3 Opus (main reasoning) and Sonnet (worker agents) to optimize cost-quality tradeoff. Requires Node 18+ for some CLI operations. Works on macOS, Linux, and Windows via WSL. Obsidian Tasks plugin (optional) enables emoji-format task queries with due dates. iCloud sync requires macOS/iOS but Git sync works anywhere.
What should I watch out for?
Team intelligence skills require CLI tools (gh, linear CLI) and MCP integrations (GitHub, Linear, Slack, PostHog). They degrade gracefully but work best when integrations are set up. Full setup takes approximately 15 minutes if starting fresh with integrations. The system is designed for individual or small team use, not large-scale organizational knowledge bases.
Key Features
Multi-Agent Support
Works with Claude Code (17 native skills), Kiro (7 powers), Gemini CLI (7 commands), Cursor, or any markdown-reading AI agent. No vendor lock-in.
Worker Agent Architecture
Parallel specialist agents (data collector, researcher, executor, publisher) run in background while main agent synthesizes, reducing costs and latency.
Local & Version-Controlled
Everything stored as markdown files with Git version history and optional iCloud sync to iPhone, iPad, and Mac. Privacy-first, no external servers.
Team Intelligence
Cross-references GitHub PRs, Linear issues, Slack discussions, PostHog metrics, and meetings into unified team briefs with two-way Linear sync-back.
Role-Based Personalization
Onboarding matches your role (Product Manager, Engineering Lead, Founder, Marketer, etc.) to pre-configured role packs that prioritize relevant skills and integrations.
Tools in this Blueprint
About This Blueprint
- License
- MIT
- Industry
- Technology
- Skills
- 6 workflows, 0 sub-skills, 11 standalone