Forecall icon
Forecall icon

Forecall

Forecall is for people who build MCP servers and APIs that AI agents call. It checks the text an agent reads before it picks a tool, and measures what models actually do with it.

The landing page: score your MCP tool descriptions for free, or get a tools/list with the CLI.

Cost / License

  • Freemium (Subscription)
  • Proprietary

Platforms

  • Online
  • Software as a Service (SaaS)
A sample result: Notion's official MCP server scored, with its average, tools and near-identical descriptions.
Scores for 55 public MCP servers, listed by name.
+1
The MCP Census: what the tools of public MCP servers are for, what calling them does, and how they authenticate.
Forecall alternatives  
0likes
3alternatives
0articles
Save

Features

Features

 Tags

Highlights All activities

Forecall News & Activities

Recent activities

Forecall information

  • Developed by

    JP flagMirai Studio, Inc.

  • Licensing

    Proprietary and Freemium product.

  • Pricing

    Subscription ranging between $29 and $199 per month + free version with limited functionality.

  • Alternatives

    3 alternatives listed

  • Badge

    Get an embeddable badge for Forecall

  • Supported Languages

    • English
    • Japanese

AlternativeTo Categories

Development, AI Tools & Services

Popular alternatives

View all
Forecall was added to AlternativeTo by Hiroki Nomura on and this page was last updated .

No comments or reviews, maybe you want to be first?

Official Links

What is Forecall?

Forecall is for people who build MCP servers and APIs that AI agents call. It checks the text an agent reads before it picks a tool, and measures what models actually do with it.

  • Free linter: paste a server's tools/list JSON, or the URL of a public server, and get a 0–100 score per tool with what to fix: missing usage context, undocumented arguments and return values, near-identical descriptions. No sign-up. The linter and its CLI are open source (Apache-2.0).
  • Evaluations (Pro and Team plans): Claude and GPT models answer utterances written for each tool, and you get tool-choice accuracy, argument validity and false calls, plus rewrite proposals for the tools they missed, measured again before they are proposed.
  • Failure KB: an MCP server that agents use to look up known tool-call failures and their workarounds, and to report new ones. A record is published once another user reproduces it, and server owners can prove their npm package and answer reports officially.

The score says what a description covers; it does not predict which tool a model will pick. Forecall also publishes scores for 55 public MCP servers and an MCP Census of their 1,274 tools.