khizardevelops

Agent Sesh — Development skill for Claude Code

Development community

A zero-dependency CLI tool that generates a persistent "Project Brain" in your workspace.

How to install Agent Sesh

This entry records only its repository, not the path inside it, so there is no exact command to give. Open khizardevelops/agent-sesh and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Agent Sesh does

A zero-dependency CLI tool that generates a persistent "Project Brain" in your workspace. It solves the context amnesia problem by creating a standardised .agents/ directory and a root AGENTS.md or CLAUDE.md pointer that tells AI agent to read and maintain that context

Alternatives in Development

Storybloq — Cross-session context for Claude Code 696 ★
Alethe Agents — A local-first desktop workspace for running, organizing, and resuming multiple coding agents and shells with r 511 ★
Claude Cognitive — Working memory for Claude Code - persistent context and multi-instance coordination 440 ★

README

🧠 agent-sesh

**Never start an AI coding session from scratch again.** `agent-sesh` generates a persistent "Project Brain" in your workspace. It solves the "Context Amnesia" problem by creating a standardised `.agents/handoff/` directory and root pointer files — `AGENTS.md`, `CLAUDE.md`, `GEMINI.md`… one per agent you use — that tell AIs to read and maintain that context.

[![npm version](https://img.shields.io/npm/v/agent-sesh.svg)](https://www.npmjs.com/package/agent-sesh) [![npm downloads](https://img.shields.io/npm/dm/agent-sesh.svg)](https://www.npmjs.com/package/agent-sesh) [![npm license](https://img.shields.io/npm/l/agent-sesh.svg)](https://www.npmjs.com/package/agent-sesh) [![test](https://github.com/khizardevelops/agent-sesh/actions/workflows/test.yml/badge.svg)](https://github.com/khizardevelops/agent-sesh/actions/workflows/test.yml)

npx agent-sesh@latest

Works with **Claude Code, Cursor, GitHub Copilot, OpenAI Codex, Gemini CLI, Windsurf, Zed, Cline, Qwen Code** and [67 more](#supported-agents) — every agent the [`skills`](https://github.com/vercel-labs/skills) CLI supports, and it never touches the skills that CLI installs. `npx agent-sesh --agents` prints the full list.

What exact problem does this solve?

**Every AI coding session starts at zero. Your project doesn't.**

The knowledge that actually makes an agent useful — why the auth layer is structured that way, which library you already tried and rejected, what's half-finished, what must never change — lives in your head and in a closed chat window. So every session you pay the same tax:

The symptom	What it costs you
Re-explaining architecture at the start of every chat	10–15 minutes and a chunk of your context window, every single time
The agent re-suggests a library you rejected three sessions ago	You re-litigate a settled decision
The agent "fixes" deliberate code because it can't see the reason	Silent regressions in code that was