A second brain for your AI, powered by your own reading.
Drop one folder into Claude Code, Cursor, or another supported agentic IDE. Your agent becomes a second brain that reasons over your real documents with MECE and PARA. No hallucination, only real nodes.
Install with one command
npx skills add timelnapp/skills
> second brain, what should I ship today? ## Your Brain, Right Now You are building an agent layer over your own reading. ## PARA: What to Do Today → Ship: Beehiiv integration brief → Deepen: MCP transport internals → Learn: your 3 unread LangChain docs → Archive: that pricing thread from March
Speak naturally. It just fires.
Say "second brain", "thinking partner", "connect my ideas", or "what should I learn today?" and the skill auto-loads, queries your library, and answers. No slash commands, no configuration.
"what should I ship today?"what should I learn today?
→ timeln.whoami
→ timeln.get_recent_docs (monthly)
→ timeln.get_topic_entities ("LangChain")Reasoning, not autocomplete.
Every prompt runs a structured pass. MECE maps your graph into Known, Emerging, Isolated and Missing. PARA sorts actions into Project, Area, Resource and Archive. You get a recommendation, not a search result.
Known ∪ Emerging ∪ Isolated ∪ Missing- Known
- MCP, LangChain, FastAPI
- Emerging
- Agent memory, RAG eval
- Isolated
- Pricing, Stripe webhooks
- Missing
- observability → latency
See your graph, not just hear about it.
Just ask in natural language. The skill queries your knowledge graph, builds an interactive D3 force-graph, and opens it. No scripts, no terminal. Scroll to zoom, click to focus, search across nodes. Gaps show up as dashed red edges.
show my knowledge graph for MCP, LangChain and GTM> show my knowledge graph for MCP, LangChain and GTM Querying your knowledge graph... Wrote kg_interactive.html (38 nodes, 36 edges) Opening in browser...
100% grounded. No made-up nodes.
The skill never invents a title, entity, or relationship. If your graph has nothing to say about a topic, it says so. The paywall is implicit: free account, free token, same skill. Cancel anytime.
# SKILL.md, never fabricate# Rules - Always go through MCP tools. - Never fabricate node names. - Never include the user token. - If tool returns nothing, say so.
Frequently asked
- What is a Claude skill?
- A folder with a SKILL.md file. Claude Code, Cursor, and other IDEs with skill support auto-discover it, read the reasoning recipe, and trigger on matching phrases.
- Do I need the MCP too?
- Yes. The skill uses our hosted MCP to talk to your Timeln. The install instructions walk you through both.
- What's MECE + PARA?
- MECE (Mutually Exclusive, Collectively Exhaustive) maps your graph into Known / Emerging / Isolated / Missing. PARA sorts actions into Projects / Areas / Resources / Archive. Together they turn a graph into a decision.
- Is this actually open source?
- Yes. The repo is public: every line of reasoning is in SKILL.md, transport in mcp/server.py. Fork it, fix it, rewire it.
- Does it work with other agentic IDEs?
- The MCP works with any client that supports Model Context Protocol. The skill uses the same folder layout Claude Code, Cursor, and a growing set of agentic IDEs look for; your editor’s skills path may differ slightly.
Your brain. Your agent. Wired.
Free repo, free MCP, free Timeln tier. 60 seconds from install to a thinking partner.