session-protocol

Save and load JSON session state with /save-session and /load-session commands.

Updated Oct 25, 2025
One-click install
npx skills add https://github.com/bartoszwarzocha/kalahari --skill session-protocol-bartoszwarzocha
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: session-protocol
Source: https://github.com/bartoszwarzocha/kalahari/tree/main/.claude/skills/session-protocol
Command: npx skills add https://github.com/bartoszwarzocha/kalahari --skill session-protocol-bartoszwarzocha

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Defines a structured session state format and lifecycle to save, load, and resume work without losing context.

Core Features & Use Cases

  • Session State Location: .claude/session-state.json
  • Session State Format (FULL): JSON schema describing mode, timestamps, git info, and task progress
  • Save/Load Commands: /save-session and /load-session guidance
  • Usage Guidance: how to handle blockers and progress tracking in sessions

Quick Start

Create and modify .claude/session-state.json to save your current session, then use /load-session to restore.

Frequently Asked Questions about session-protocol

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I save and restore session state in interactive workflows?

Session state management saves your workflow progress to .claude/session-state.json, capturing mode, timestamps, git info, and task progress. Use /save-session to persist state and /load-session to resume work without losing context across local and remote environments.

What format should I use to structure persistent session data?

Session state uses a JSON schema stored at .claude/session-state.json that records status fields, blockers, task progress, and workflow metadata. This structured format ensures reliable state serialization and desynchronization prevention across long-running tasks and multi-step sessions.

Can I use session state to track progress in multi-step AI workflows?

Yes. Session state management is designed for interactive AI workflows and multi-step sessions where persistent state must be preserved between runs. It tracks task progress, blockers, and context to enable reliable resumption without manual reconstruction.

How do I prevent data loss when resuming interrupted work?

Explicitly save session state with /save-session commands before stopping work, then load it with /load-session to restore full context. The JSON schema captures all necessary state—timestamps, progress markers, and blockers—to resume exactly where you left off.

What happens if my session state file becomes out of sync?

Session state management prevents desynchronization by using explicit save and load commands with a single source of truth at .claude/session-state.json. Structured JSON formatting and status tracking ensure consistency across local and remote execution environments.