mcpField
R
rerunplan

rerunplan

Local
github.com/CloudCorpRecordsDeveloper Tools0Updated 1w ago
Repository

About rerunplan

rerunplan is a local tool for deciding whether a failed check should be rerun, fixed, or investigated. It classifies failures using the check name, error signature, and past runs, and returns the evidence behind its recommendation.

Key features

  1. 01Classify infrastructure flakes, flaky checks, and real failures
  2. 02Use matching check history and error signatures
  3. 03Return a recommendation with supporting evidence
  4. 04Normalize common pass and failure statuses
  5. 05Run locally without network or configuration

Use cases

  1. 01Choose whether to rerun or fix a failed check
  2. 02Triage intermittent test failures
  3. 03Identify failures that need further investigation

MCP support

Exposes a `plan_rerun` tool to MCP clients. It accepts the failed check, optional error signature, and past-run history, then returns a classification and recommendation.

Configuration

Local server config is the same for every client. Copy it into your client’s MCP config file.

mcp.json · Universal config
{
  "mcpServers": {
    "cloudcorprecords-rerunplan": {
      "command": "npx",
      "args": [
        "-y",
        "@reneturcios/rerunplan"
      ]
    }
  }
}
Run commandnpx -y @reneturcios/rerunplan

Local servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.

Packages
npm@reneturcios/rerunplan0.1.0
ocighcr.io/cloudcorprecords/rerunplan:0.1.0

Tools

1

Saved tool list · updated 1w ago

ParameterTypeRequiredDescription
checkstringYesThe failed check's name
past_runsarrayYesPast runs; runs for other checks are ignored
error_signaturestringNoThe current failure's error signature (optional)

Details

From the project README

rerunplan

A check failed. Rerun it, or fix it? rerunplan classifies the failure from evidence — the failing check's name, its error signature, and past runs of the same check — and answers with a classification, a recommendation, and the evidence lines that drove the call:

  • infra_flake → rerun — the error signature matches a known infrastructure pattern (timeout, connection reset, DNS, rate limit, runner loss, gateway errors) and the check is not failing that way every time
  • flaky → rerun — the check has passed before and has failed before with this same error signature; the failure comes and goes
  • real_failure → fix — the check fails in every usable past run (with the current signature, when one is given); an infra-looking signature that fails every time is still a real failure
  • insufficient_evidence → investigate — no usable history, or a mixed history no rule claims

Rules fire in a fixed order and each appends its evidence. Statuses are normalised (success/pass/passed, failure/fail/failed/error); runs for other checks are ignored. Pure local computation: no network, no configuration.

CLI

npx @reneturcios/rerunplan plan --check unit-tests --signature "Error: boom" --past past-runs.json
npx @reneturcios/rerunplan --version

past-runs.json is a JSON array of {"check": "unit-tests", "status": "failure", "error_signature": "Error: boom"} objects (check and error_signature optional per run).

Exit codes: 0 on success, 1 on a runtime error (including an unreadable or invalid past-runs file), 2 on a usage error.

MCP server

npx -y @reneturcios/rerunplan mcp

MCP client config:

{
  "mcpServers": {
    "rerunplan": {
      "command": "npx",
      "args": ["-y", "@reneturcios/rerunplan", "mcp"]
    }
  }
}
Tools
  • plan_rerun — Classify a failed check from its past-run history and recommend rerun, fix, or investigate.
    • check (string, required) — the failed check's name
    • error_signature (string, optional) — the current failure's error signature
    • past_runs (array, required) — past runs as {check?, status, error_signature?} objects

Docker

docker run -i ghcr.io/cloudcorprecords/rerunplan
docker run --rm -v "$PWD:/w" ghcr.io/cloudcorprecords/rerunplan plan --check unit-tests --past /w/past-runs.json

Development

npm install
npm run build   # tsc -> dist/
npm test        # tsx --test test/*.test.ts (pure local, no network)
npm run dev     # run the CLI from source

License

MIT — Copyright (c) 2026 Rene Turcios

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.