MCP endpoint · you are looking at a server
This host does not serve a website. It answers one question, for coding agents.
Point your agent here and ask whether the app it built is fit to carry real users. It runs 118 read-only detectors against your own repository and gets back a graded readiness report — every gap named, every consequence spelled out. The scan is free.
Install it, in one line
claude mcp add --transport http prodready https://mcp.prodready.si/mcp --scope userThen ask, in plain language: use prodready to check whether this app is production ready. Lines for Codex, Cursor, Windsurf and Cline are on the documentation page below.
Two tools for the free scan
prodready_probe(stack?)
Returns the read-only commands and file globs your host should run locally — greps, file counts, git ls-files. Nothing that writes or installs, and nothing a command does calls out.
prodready_assess(evidence)
Takes what your agent collected and returns findings mapped to numbered rules, a 0–100 score across five dimensions with a grade, and the next step.
sdlc_* — the SDLC, with the licence
Our delivery process on your own project: refined backlog, sprints, three independent reviewers on every change, audits and release gates, with a ledger of every gate. Listed for every account; they unlock with the SDLC licence. What the SDLC does.
The rule this server is built on
The report names the gap and its consequence. It never names the remedy. “Three tables carrying an owner column have row security switched off” is a finding; what you do about it is the work we sell. You will not find a patch, a migration, a config snippet or a step list anywhere in the output.
What never happens here
- This server never reads your disk. It is remote and has no way to.
- Your source code is never sent. What crosses the wire is paths, line numbers, counts and matched identifiers — and you can read it before it goes.
- Secret values are never collected. Credential detectors list files and count matches; they never print what matched.
- Evidence is never stored and never logged: the report is written and the input is gone. What is kept is your sign-in, the report page for seven days and, if you use the SDLC tools, your project documents until you delete them.
The full documentation is on the site.
What each tool returns, what a report looks like line by line, the five scoring dimensions and the grade bands, what the scan cannot see, how the scope band is worked out, and install lines for Claude Code, OpenAI, Cursor, Windsurf and Cline.