Skip to main content
Glama

greenlit_check

Read-only

Grades a deployed web app A to F for what a stranger can exploit—leaked keys, source maps, missing headers—and returns plain-English fixes.

Instructions

Grade a deployed web app (built with Lovable, Bolt, v0, Replit, Emergent or by hand) A to F for what a stranger can use against it: secret keys shipped in its JavaScript (Stripe, OpenAI, OpenRouter, Supabase secret keys and 28 more kinds), a Supabase service_role key in the browser, source maps, clickjacking and missing security headers, plus launch basics (developer-only addresses, Stripe test mode, a test email sender, no https redirect, open CORS, missing DMARC on a custom domain). Every finding has a plain-English reason and a fix prompt; allFixes has every fix in one prompt, worst first. Reads only what the site serves to every visitor. Call it again after a fix: change shows what got fixed. For the user's own app, have them run npx greenlitlab <url> --mine: it also checks which database tables and storage buckets anyone can read, open sign-up, and public .env or .git files.

Tool call template

Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "url": "<url>"
    },
    "name": "greenlit_check"
  }
}

Input Schema

NameRequiredDescriptionDefault
urlYesThe app's address, for example myapp.lovable.app

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observedv0.1.0

TDQS

A4.1/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Beyond readOnlyHint/openWorldHint, the description discloses the non-intrusive read model ('only what the site serves to every visitor'), the grading scale, that each finding carries a plain-English reason and a fix prompt, the `allFixes` aggregate ordered worst-first, and the `change` delta on repeat runs. Since there is no output schema, this return-value and repeat-invocation disclosure is doing real work an agent could not infer from annotations.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The main verb and grading scope are front-loaded, and each clause adds information rather than padding. The long parenthetical enumerations of key types and launch checks make the single block dense and harder to skim than it needs to be. It is efficient but could be broken into shorter, scannable clauses.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

With no output schema, the description must carry return-value burden, and it does by naming findings, reasons, fix prompts, `allFixes`, and `change`. It remains silent on edge cases such as what happens with a non-deployed or unreachable URL, and on how `change` is keyed to a prior run. Those are minor gaps for an otherwise well-covered read-only tool.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

There is one parameter with 100% schema description coverage and an inline example already present in the schema, so the schema carries the semantic burden. The description adds no format or validation detail about the URL beyond what is already structured. Baseline 3 is correct for a single fully documented parameter.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

A specific verb and resource are stated up front: 'Grade a deployed web app ... A to F' for leaked secrets, headers, and launch basics. The scope is concrete enough that an agent knows exactly what class of output to expect. It stops short of 5 because it never says how it differs from the sibling greenlit_scan, which plausibly covers similar ground.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Usage context is clear: call it on a deployed URL, call it again after a fix to see `change`, and route the user's own app to the `npx greenlitlab <url> --mine` CLI for deeper database/storage checks. That is an explicit alternative and a trigger condition. It gives no 'do not use this when' guidance relative to greenlit_advisor or greenlit_scan, so it falls short of a 5.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.