Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/bastndev/f1/llms.txt

Use this file to discover all available pages before exploring further.

Every time you start a new agent session, the agent starts cold. It has no idea which folders matter, what conventions the codebase follows, or what rules you want it to obey — so you spend the first few exchanges re-explaining context that hasn’t changed since yesterday. F1 solves this with a two-layer context system that writes your project’s structure and working rules into the workspace before the agent even sees your first message.

​
The problem: cold starts waste time and tokens

AI agents have no persistent memory between sessions. Each launch is a blank slate. Without help, an agent must either scan the entire repository — burning tokens and adding latency — or proceed without context and risk making uninformed edits. For large codebases this compounds quickly: every session, every agent, every time.

​
The two-layer solution

F1 ships two complementary systems that work together to fix cold starts.

My Memory

A lightweight, pure-TypeScript engine that scans your workspace and writes a .f1/ project map — a structural picture of your project that any agent can read instantly. No API keys, no LLM calls, no network traffic. Just a fast filesystem scan that runs before launch.

Smart + Skills

A one-click launch mode that wraps My Memory and adds two more layers: a graphify code graph of your project and a set of built-in working rules written to .f1/smart-rules.md. F1 then types a priming prompt into the agent’s terminal so it actively reads everything before replying.

​
How they relate

Smart + Skills is built on top of My Memory — it does not replace it. Internally, SmartService owns a MemoryService instance and calls setEnabled(true) in its constructor, so the full project-map pipeline always runs as part of a Smart launch. You can also use My Memory on its own (without Smart + Skills) if you only want the structural map and instruction-file sync without the code graph or built-in rules. 
// SmartService always activates MemoryService internally
export class SmartService {
  private readonly memory = new MemoryService();

  constructor() {
    // Smart mode always builds the project map + syncs the AGENTS.md hub
    this.memory.setEnabled(true);
  }
}

​
The result

When Smart + Skills is enabled, the agent’s very first reply — before you’ve typed a single task — is: 
I am ready for work ✅
That line is the agent confirming it has read .f1/smart-rules.md, .f1/project-map.md, and (if graphify ran successfully) graphify-out/GRAPH_REPORT.md. It is oriented, rule-bound, and ready — with no manual context-pasting on your part.
My Memory requires no API keys and makes no LLM calls. It is a pure TypeScript filesystem scan that reads package.json and lists top-level folders. It runs synchronously and never throws, so it cannot block or fail a CLI launch.

Build docs developers (and LLMs) love

Get started for freeTalk to us