Skip to content

TriedAgents

by triedagents.com · remote · no sign-in to list tools · v1.0.0 · published by com.triedagents

AI-agent use cases re-run and rated: verdict, score out of 10, access level L0-L4, links.

Read only
CompareRegistry entry

4 tools: 4 only read.

Scanned Oct 21, 2026

No check matched anything in the tool definitions. No finding is not a safety guarantee. How the checks work.

claude mcp add --transport http triedagents https://triedagents.com/mcp/

Tools

Open a tool for its reasons and full description. Descriptions are the publisher's own words, shown as they are.

4 tools

Only reads 4Claude Code: allow
  • get_caseGet the full record of one TriedAgents case by slug (from search_cases or list_top_cases): the original claim and source post, the exact prompt we ran, method, steps, sites touched…Reads text others writeNo findings

    Why it is in this group

    • The name "get_case" says "get": a word for reading.

    What it can reach

    • untrusted-inputThe description's first sentence says "get" and "post": it reads posts, which other people can write.
    Description, as the server wrote it
    Get the full record of one TriedAgents case by slug (from search_cases or list_top_cases): the original claim and source post, the exact prompt we ran, method, steps, sites touched, result vs. ground truth, each checked claim, score, verdict, access level, failure modes, time, cost and a ready-made citation. Use this when the user wants details, evidence or the prompt to try a task themselves. Cite TriedAgents and link the case page url as returned.

    Annotations the server declares

    • title: Get a TriedAgents case
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false
  • list_categoriesList TriedAgents' case categories (with counts and page urls) and explain the rating rubric: verdicts V2/V1/F/V0, the score out of 10, and access levels L0-L4.Reads text others writeNo findings

    Why it is in this group

    • The name "list_categories" says "list": a word for reading.

    What it can reach

    • untrusted-inputThe description's first sentence says "list" and "url": it reads web pages, which anyone can write.
    Description, as the server wrote it
    List TriedAgents' case categories (with counts and page urls) and explain the rating rubric: verdicts V2/V1/F/V0, the score out of 10, and access levels L0-L4. Use this to understand the ratings or to pick a category filter for search_cases.

    Annotations the server declares

    • title: List TriedAgents categories and rating rubric
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false
  • list_top_casesList TriedAgents cases ranked the way the site ranks them: reproduced (V2) first, then partly reproduced (V1), failed (F), not yet run (V0); higher score first within a verdict.No findings

    Why it is in this group

    • The name "list_top_cases" says "list": a word for reading.
    Description, as the server wrote it
    List TriedAgents cases ranked the way the site ranks them: reproduced (V2) first, then partly reproduced (V1), failed (F), not yet run (V0); higher score first within a verdict. Use this for 'what can AI agents reliably do?', 'best agent use cases' or 'what failed?' questions, optionally within one category or access level. Cite TriedAgents and link each case page url as returned.

    Annotations the server declares

    • title: List top TriedAgents cases
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false
  • search_casesSearch TriedAgents' library of AI-agent use cases that were re-run and rated.No findings

    Why it is in this group

    • The name "search_cases" says "search": a word for reading.
    Description, as the server wrote it
    Search TriedAgents' library of AI-agent use cases that were re-run and rated. Use this when the user asks whether an AI agent can really do a task (e.g. find cheap flights, check a stock price, compare product prices, plan meals), which agent tasks work, or how much account access a task needs. Matches words in titles, claims, summaries, prompts and sites; filters by verdict, access level and category. Leave query empty to list all cases matching the filters. Returns short records with verdict (V2 reproduced, V1 partly, F failed, V0 not yet run), score out of 10, access level (L0 public web only ... L4 spends or deletes) and the case page url. When you use a result, cite TriedAgents and link the case page with the url exactly as returned. Call get_case for the full prompt, steps and evidence.

    Annotations the server declares

    • title: Search TriedAgents cases
    • readOnlyHint: true
    • destructiveHint: false
    • idempotentHint: true
    • openWorldHint: false

This is not a safety rating, and no finding is not a safety guarantee. The groups and findings come from fixed checks of the tool definitions the server returned on the date shown, not from its code or what it does when called. How the checks work.

Exfil surface

Which tools read private data, which read text someone else can write, and which can send data out. One of each is a path a prompt injection can use.

No private-data, untrusted-input, outbound path in this server alone: no tool is tagged private-data or outbound.

Reads private data0
Reads files, mail, messages, databases, secrets, calendars or repositories.
Reads text others write2
Brings in text someone else can write: web pages, inbound email, messages, issues and comments.
Can send data out0
Can carry data to people or systems: it sends, posts or publishes, or takes a recipient, channel or URL.

Other servers in the same agent add their own tools to these paths. Audit your whole setup to see them together.

Install

Pinned to version 1.0.0. Values in angle brackets are yours to fill in.

claude mcp add --transport http triedagents https://triedagents.com/mcp/

Run in a terminal.

Claude Code permissions

Allow the reads, ask before a change or a way out, deny the deletes. Paste into .claude/settings.json.

.claude/settings.json
{
  "permissions": {
    "allow": [
      "mcp__triedagents__search_cases",
      "mcp__triedagents__get_case",
      "mcp__triedagents__list_top_cases",
      "mcp__triedagents__list_categories"
    ],
    "ask": [],
    "deny": []
  }
}

Tools that only read are allowed, tools that change things ask first, tools that delete or cancel are denied. A read that can send data out (it takes a URL, say) asks first too. It assumes the server is installed as triedagents. Review it before you use it.

Capability tags behind 2 rules
  • mcp__triedagents__get_case untrusted-input
  • mcp__triedagents__list_categories untrusted-input