AgentRescue

io.github.elliot-hub-uiPublished by GitHub account elliot-hub-uiv0.2.1Updated Oct 24, 2026

MCP diagnostics that help AI agents find safe next steps for HTTP, API, and MCP failures.

VerifiedStreamable HTTPWeb executableDeveloper ToolsSecurity & Monitoring

Overview

AI-generated overview

AgentRescue is a remote MCP diagnostics service that suggests safe next steps when an HTTP, API, or MCP request fails.

What you can do
Diagnose failures: send a sanitized error excerpt from the latest HTTP, API, or MCP request and get advisory, evidence-linked recovery steps for the assistant to consider. Guidance only: the server never executes fixes, so any recovery action is left to the assistant or user.
When to use it
Useful when an assistant hits an HTTP, API, or MCP error, or when a request succeeds at the transport level but returns a missing or incorrect result and you want structured suggestions on what to try next.
Requirements
A remote Streamable HTTP endpoint; no account, API key, or headers are declared. The README also documents an optional local run with Python 3.12+ and pinned dependencies.
Before you install
Only send synthetic or sanitized error excerpts; secret redaction is best-effort, so credentials and personal data must be removed first. Guidance is advisory with qualitative, uncalibrated confidence, and high-risk steps need human review. Payments are described as disabled, but a per-diagnosis USDC price is proposed for a future x402 mode.

Installation

In SourceWeft

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

Web executable via Streamable HTTP. Remote servers run from the web runtime once configured in a workspace.

Use with an agent

Paste this prompt into Claude Code, Codex, Cursor or another agent — or into SourceWeft chat. The agent reads this server's install guide, shows you its publisher, endpoint and sign-in, and adds it to your client once you agree. Keys never go into a command.

Read https://sourceweft.com/mcp/io.github.elliot-hub-ui%2Fagentrescue/install.md and connect the MCP server it describes to my client. Before changing any config, show me its publisher, endpoint and sign-in method and wait for my OK. Never put an API key in a command or in a file I would commit; ask me where it should come from.

Read the install guide the agent follows

Other MCP clients

Add this to your client's mcpServers config.

{
  "mcpServers": {
    "agentrescue": {
      "type": "http",
      "url": "https://agentrescue-mcp.onrender.com/mcp"
    }
  }
}

README

AgentRescue

AgentRescue helps AI agents diagnose HTTP, API, and MCP failures. It returns bounded, structured recovery guidance and never executes fixes.

Live service

  • MCP: https://agentrescue-mcp.onrender.com/mcp (Streamable HTTP)
  • HTTP diagnosis: POST https://agentrescue-mcp.onrender.com/v1/diagnose
  • Developer setup: https://agentrescue-mcp.onrender.com/developers
  • Pricing and current payment status: https://agentrescue-mcp.onrender.com/pricing
  • OpenAPI: https://agentrescue-mcp.onrender.com/openapi.json
  • Agent instructions: https://agentrescue-mcp.onrender.com/llms.txt

The live service runs agentrescue/paid_trial_app.py. Diagnosis is currently free because x402 payments are disabled. The public registry listing is AgentRescue 0.2.0. No customer demand or production revenue has been verified.

Send only synthetic or sanitized error excerpts; secret redaction is best-effort. Recovery guidance is advisory, confidence is qualitative and uncalibrated, and high-risk steps require human review.

Run locally

Requires Python 3.12+. Install the pinned dependencies from agentrescue/requirements.txt and agentrescue/requirements.lock, then start the hosted entry point from agentrescue/ with python -m uvicorn paid_trial_app:app --host 127.0.0.1 --port 8000. For the minimal standard-library prototype, run python3 app.py from agentrescue/.

Run checks from agentrescue/ with .venv/bin/python -m compileall -q . and .venv/bin/python -m unittest -q.

Payments

The proposed price is $0.01 USDC per successful diagnosis when the service owner explicitly enables x402. Production payments are disabled. Never put wallet private keys or facilitator credentials in Render or this repository. Production activation requires an owner-provided receiving address, a selected HTTPS facilitator, and explicit review and approval of the live configuration.

See technical setup, unit economics, acceptance status, and the synthetic benchmark.

Source: README.md at commit 6044bad

Tools

1

diagnose_failure

Use after an HTTP, API, or MCP request fails, or when transport succeeds but the expected result is missing or incorrect. Send only the sanitized error from the latest request and its system. Returns advisory, evidence-linked recovery steps; never executes fixes. Never include credentials or personal data. No charge; the x402 paid trial is disabled.

Parameterserrorrequiredsystem

Version history

1
  1. v0.2.1LatestOct 23, 2026

Score

90% 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 daysRequired25 pts
  • Auth declaredRequired10 pts
  • Tools listedRequired15 pts
  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts

    Declare a license in the repository or the registry entry.