Living Stack Community

io.github.equinoxaifinance-rgbPublished by GitHub account equinoxaifinance-rgbv0.4.2Updated Oct 29, 2026

Save your coding agent's exact next step and load it after a restart. Free MCP server, no account.

VerifiedSTDIODesktop onlyDeveloper ToolsSecurity & Monitoring

Overview

AI-generated overview

Free local MCP server that gives an agent a seven-tool proof loop: open a scoped session, authorize a bounded action, record evidence, check a claim, and…

What you can do
  • Check server status and open, inspect or close a scoped session
  • Authorize a bounded action before the host performs it
  • Record typed evidence and outcomes for actions taken
  • Check a claim against the exact subject and outcome
  • Keep a tamper-evident local ledger of the session, with raw values stored only as SHA-256 digests
When to use it
Worth adding when you want an agent to keep a verifiable, local record of what it was authorized to do and what happened, for example to prove the workflow before buying a fuller edition. It is a proof edition, not a durable memory or recovery system.
Requirements
Node.js 22 or newer for the local stdio process, installed via the .mcpb bundle or npx from the GitHub repository. No account, API key, environment variables or headers are declared.
Before you install
The server gates and records host actions but never executes them, so the agent still performs the actions itself. The default policy rejects external and destructive actions. The free edition omits durable memory, recovery checkpoints, signed traces and multi-agent coordination; a paid edition is promoted in the README. Feedback is opt-in via a public GitHub issue, and the runtime itself makes no network calls.

Installation

In SourceWeft

  1. Open Living Stack Community in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.

Other MCP clients

Follow the launch instructions in the repository.

README

Living Stack Community

Buyer guides: run a real proof loop · choose the right edition.

Living Stack Community is the free seven-tool proof loop for MCP-capable agents. It proves the core behavior before a buyer pays: open a scoped session, authorize a bounded action, record typed evidence, check a claim against the exact subject and outcome, and close the session with a tamper-evident local ledger.

It does not include the paid implementation for durable memory, recovery checkpoints, portable signed traces, release-byte verification, adoption receipts, or multi-agent coordination.

Free, 7 tools, no account. Need more? Complete Local has 23 tools for $19 once (compare editions).

Install

Requires Node.js 22 or newer.

Claude Desktop (one click): download the .mcpb bundle and open it. Its SHA-256 is listed in server.json.

Claude Code:

bash
claude mcp add living-stack -- npx -y github:equinoxaifinance-rgb/living-stack-community

Codex:

bash
codex mcp add living-stack -- npx -y github:equinoxaifinance-rgb/living-stack-community

Want the exact commit we tested instead of the newest? The start page has the same commands pinned to one commit.

Cursor, Claude Desktop (manual) and other MCP clients: add this to the client's MCP config (~/.cursor/mcp.json, or claude_desktop_config.json):

json
{  "mcpServers": {    "living-stack": {      "command": "npx",      "args": ["-y", "github:equinoxaifinance-rgb/living-stack-community"]    }  }}

Check it works: npx -y github:equinoxaifinance-rgb/living-stack-community --self-test prints "decision": "PASS".

From a clone, the MCP command is node /path/to/living-stack-community/server.mjs.

Want to help validate the product without buying it? Join the five-user proof pilot. Feedback is opt-in through a public GitHub issue; the Community runtime itself still makes no network calls.

Included tools

  • livingstack.status
  • livingstack.session_start
  • livingstack.session_status
  • livingstack.authorize_action
  • livingstack.record_outcome
  • livingstack.check_claim
  • livingstack.session_close

Upgrade to Complete Local

Complete Local has 23 tools and adds the capabilities that turn the proof loop into an operating runtime: durable context memory, checkpoint recovery, Ed25519-signed portable traces, release verification, opt-in adoption receipts, and all multi-agent Team workflows.

Community is not a time-limited trial. It is the permanent proof edition. Complete Local is the product for ongoing work.

Privacy and boundaries

  • State is local by default.
  • Raw goals, scopes, targets, evidence references, summaries, and close reasons are stored only as SHA-256 digests.
  • The default policy rejects external and destructive actions.
  • The server gates and records host actions; it never executes them.
  • Community has no telemetry or network calls.

See PRODUCT.md and SECURITY.md.

Source: README.md at commit 0ec6832

Tools

0
SourceWeft started this package once in an isolated sandbox, without credentials, to list its tools. That failed: the package type cannot run in the sandbox.

Version history

2
  1. v0.4.2LatestOct 29, 2026
  2. v0.4.1-beta.1Sep 16, 2026

Score

50% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 days25 pts

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • Auth declared10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts