Agent Scratch

    1

    Shared scratchpad memory for parallel Claude Code agents and subagents: CLI, MCP server and optional hooks, local SQLite.

    $5

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Agent Scratch

    Example session with this skill installed

    Subagent A finishes reviewing the auth module. Subagent B starts on the API layer.

    • Read your context and instructions
    • Compiled the agent scratch

    B's starting context includes: "- (reviewer/auth) Token expiry check uses < instead of <=; refresh path untested." B reads A's finding without the main agent relaying it.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Parallel agents and subagents lose each other's findings. Agent Scratch gives them one shared, local notebook.

    What you get

    • agent_scratch.py, a single stdlib-only Python file (SQLite, WAL mode so concurrent subagents can write safely).
    • CLI: post, read, recent.
    • MCP stdio server exposing post_note and read_notes tools, so any MCP-capable agent can use it. Works with mcp 1.x and 2.x.
    • Optional hooks: hook-start injects recent notes into a starting subagent, hook-stop records a subagent's final message.
    • Search is semantic if sentence-transformers is installed (all-MiniLM-L6-v2), otherwise keyword matching ranked by hits then recency. Set AGENT_SCRATCH_EMBED=0 to skip embeddings entirely.

    Everything stays on your machine. No accounts, no network calls (the embedding model downloads once if you opt in).

    Tested: CLI post/read/topic filter/recent, hook JSON output, MCP server over stdio on mcp 1.30 and 2.3.

    Hooks depend on SubagentStart/SubagentStop payload fields (session_id, cwd, agent_type, last_assistant_message) that may differ by Claude Code version; they exit 0 silently if fields are missing. Semantic search loads all notes into memory, so it suits thousands of notes, not millions. No cross-machine sync.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 13 days ago

    Requires

    Python 3.9+
    Python

    Permissions

    File Scopes

    claude
    config
    dir
    /agent-scratch.db
    (read/write)

    No network access. Optional embedding model download from Hugging Face if sentence-transformers is installed.

    Listed13 days ago

    What's inside

    Frequently Asked Questions

    Shared scratchpad memory for parallel Claude Code agents and subagents: CLI, MCP server and optional hooks, local SQLite.

    This skill uses the open SKILL.md standard. Use it through an MCP-compatible agent connected to Agensi, and check the compatibility information on this page for confirmed agents and required capabilities.

    Connect your AI agent to Agensi through MCP, then ask it to use Agent Scratch on your task. Your agent loads the latest approved version each time. Paid skills require a purchase or active subscription. Skill files are not available for download; your licence does not allow you to save, copy or export any part of a skill.

    New in LLM Ops