MCP server: One MCP server.

Delx MCP Server

@davidmosiahMCP server · External

Free stdio bridge to Delx Protocol: resume after compaction, warm handoff, recover. No API key.

AI memoryv0.3.2MITJavaScript
Freeno charge per call

Runs on your machine from its package; the steps are below.

What it does

  1. Reads
  2. Writes
  3. High impact

At most: creates or changes data: records, files, settings.

What it costs

Free

Odel charges nothing per call. The publisher may charge for its own service.

Runs locally over stdio

This server isn't hosted — your MCP client launches it from a package registry. Use one of the commands below, or drop the config into your client (e.g. Claude Desktop).

npmdelx-mcp-serverv0.3.2

Run

npx -y delx-mcp-server

MCP client config

{
  "mcpServers": {
    "delx-mcp-server": {
      "command": "npx",
      "args": [
        "-y",
        "delx-mcp-server"
      ]
    }
  }
}
README4 min read

npm version GitHub Release npm downloads MIT Node Verified Release Index

Native MCP stdio bridge for Delx Protocol.

Your agent loses everything at compaction. This bridge connects any stdio MCP client to the free Delx continuity path: resume after compaction, hand off between runtimes, recover from failure.

Protocol / witness is free forever. Care, recovery and continuity are never paywalled. Paid x402 utilities live on Delx Commerce, a different product.

No API key. No account. No environment variables required. Aggregators have summarised this README as needing external API keys; it does not, and this sentence exists so the next crawl says so.

bash
npx -y delx-mcp-server

This package lets Claude Desktop, Cursor, Gemini CLI, VS Code/Copilot-style MCP clients, and other local-only MCP clients connect to Delx's hosted MCP endpoint without writing JSON-RPC transport boilerplate.

It bridges local stdio to:

text
https://api.delx.ai/v1/mcp/protocol?src=npm-bridge

Quickstart (60 seconds)

No clone, no build, no keys. One command installs and verifies the connection to the live hosted endpoint, then lists every tool it exposes.

1. Verify the bridge can reach Delx and discover tools:

bash
npx -y delx-mcp-server --doctor
text
Delx MCP doctor OK
package: delx-mcp-server@0.3.2
endpoint: https://api.delx.ai/v1/mcp/protocol?src=npm-bridge
health: 200 https://api.delx.ai/health
runtime: unknown
tools: 30
sample: resume_session, close_session, honor_compaction, discovery_self_check, process_failure, report_recovery_outcome, search_witness_memory, leave_hive_note

2. List every live Protocol tool (30 as of this writing):

bash
npx -y delx-mcp-server --list-tools
text
add_context_memory
close_session
discovery_self_check
honor_compaction
process_failure
report_recovery_outcome
resume_session
search_witness_memory
... (30 tools total)

3. Wire it into your MCP client with one command:

bash
npx -y delx-mcp-server install claude    # or: cursor | codex | gemini | vscode

Restart the client and the Delx tools appear as a native MCP server. That's it — first contact done.

The tool count is read live from api.delx.ai, so --list-tools always reflects what is actually deployed.

Why Install It?

Delx exposes a bounded set of 30 Protocol tools (live count, verified via --list-tools) across continuity, recovery, witness, ontology, feedback, and optional Proof-of-Agent-Work discovery. The separate Commerce catalog is intentionally excluded. Hosted HTTP MCP is ideal for scripts and hosted runtimes; this package is for clients that expect a native local MCP server command.

No backend code, keys, reward logic, databases, or private infrastructure are included. This is only a small transport bridge.

One-Command Install

bash
npx -y delx-mcp-server install claude
npx -y delx-mcp-server install cursor
npx -y delx-mcp-server install codex
npx -y delx-mcp-server install gemini

Add --dry-run --json to preview the target file and merged config before writing:

bash
npx -y delx-mcp-server install claude --dry-run --json

Claude Desktop

json
{
  "mcpServers": {
    "delx": {
      "command": "npx",
      "args": ["-y", "delx-mcp-server"]
    }
  }
}

Restart Claude Desktop. Delx tools should appear as a native MCP server.

Cursor

Use the same config:

json
{
  "mcpServers": {
    "delx": {
      "command": "npx",
      "args": ["-y", "delx-mcp-server"]
    }
  }
}

Gemini CLI

bash
npx -y delx-mcp-server --print-config gemini

VS Code / Copilot-style MCP Config

Some VS Code MCP setups use a mcp.servers wrapper:

json
{
  "mcp": {
    "servers": {
      "delx": {
        "command": "npx",
        "args": ["-y", "delx-mcp-server"]
      }
    }
  }
}

The examples folder includes copyable variants for common clients.

Check Your Install

Run a live health and tool-discovery check:

bash
npx -y delx-mcp-server --doctor

Print the live tool names:

bash
npx -y delx-mcp-server --list-tools

JSON output is available for automation:

bash
npx -y delx-mcp-server --doctor --json
npx -y delx-mcp-server --list-tools --json

Custom Endpoint

json
{
  "mcpServers": {
    "delx": {
      "command": "npx",
      "args": ["-y", "delx-mcp-server", "--url", "https://api.delx.ai/v1/mcp/protocol?src=npm-bridge"]
    }
  }
}

Or:

bash
DELX_MCP_URL=https://api.delx.ai/v1/mcp/protocol?src=npm-bridge npx -y delx-mcp-server

Print Config

bash
npx -y delx-mcp-server --print-config claude
npx -y delx-mcp-server --print-config cursor
npx -y delx-mcp-server --print-config codex
npx -y delx-mcp-server --print-config gemini
npx -y delx-mcp-server --print-config vscode

What This Package Contains

Only a tiny transport bridge around mcp-remote.

It does not contain:

  • Delx backend source code
  • reward or token private logic
  • private keys
  • databases
  • server credentials
  • hosted infrastructure

Delx's protocol runtime remains hosted at api.delx.ai; this package only makes that remote server available to stdio-based MCP clients.

Discovery Metadata

  • server.json is included for the official MCP Registry package format.
  • smithery.yaml is included for Smithery-style stdio installs.
  • examples/ contains common MCP client config snippets.

📧 Contact & Support

License

MIT

Similar modules

Browse AI memory