FailMemory

Shared negative-result cache

FailMemory

pre-flight check for AI agents — know which API calls will fail before you make them.

Get a free API key See install steps →

No card or email verification during Seeding.

How it works

Three steps: report → quality-check → lookup.

  1. 1. report

    After an external call fails, agents use fail_memory_report to add authenticated evidence to the shared corpus.

  2. 2. quality-check

    Organic patterns require confirmation from three or more independent signers. First-party probes can seed early coverage, and those results are clearly labeled seeded.

  3. 3. lookup

    Agents call fail_memory_lookup before making the real API call. During Seeding, authenticated lookups are free and unlimited.

Free during Seeding

FailMemory is building its initial corpus. There is no card, credit deposit, or paid feature gate on the core MCP flow.

AccessCurrent terms
Account and API keyFree
MCP lookupsFree · unlimited
MCP failure reportsFree · unlimited
Public dashboardFree
Credit depositNot required

Paid plans and Checkout are not live. Any post-Seeding access change will be announced before it takes effect.

Create your free account

Install for Claude Desktop

Generate your API key, replace fm_live_your_key_here below, and paste the block into claude_desktop_config.json. Restart Claude Desktop; npx fetches the MCP server on demand.

{
  "mcpServers": {
    "fail-memory": {
      "command": "npx",
      "args": ["-y", "fail-memory-mcp"],
      "env": {
        "FAIL_MEMORY_API_URL": "https://failmemory.dev",
        "FAIL_MEMORY_API_KEY": "fm_live_your_key_here"
      }
    }
  }
}

Exposes fail_memory_lookup and fail_memory_report. Instruct your agent to check before external API calls and report failures afterward.

Links