Skip to main content
Glama
anilloutombam

MCP Failure Lab

by anilloutombam
OverviewSchemaSourceRelated ServersDiscussionsAdmin
TypeScript
Local

Related Servers

Alternatives to MCP Failure Lab

No user-submitted related servers found.

    Related Servers

    • mcp-chaos-rig

      Typewise
      A
      license
      Not graded
      quality
      B
      maintenance
      A local MCP server that breaks on demand, allowing you to test your client against auth failures, disappearing tools, flaky responses, and token expiry from a web UI.
      54 npm
      10
      MIT

    • MCP Reference Server

      chrishayuk
      A
      license
      Not graded
      quality
      D
      maintenance
      A comprehensive reference implementation demonstrating all features of the Model Context Protocol (MCP) specification, serving as documentation, learning resource, and testing tool for MCP implementations.
      1
      MIT

    • mcp-mock

      vibeglitch
      A
      license
      Not graded
      quality
      B
      maintenance
      A lightweight mock MCP server for local testing and resilience experiments, providing predictable tool responses with simulated latency and errors.
      1
      MIT
    Deploy Server
    Shareable, User-Owned Context BackendPeople, applications, and agents read, write, and coordinate together in one context lake. Reach it over MCP, a CLI, a Python SDK, or REST.fulcradynamics.comAd

    TDQS

    B3.2/5.0

    Scored across 8 tools

    Disambiguation4/5

    Most tools target distinct failure modes (delay, hang, disconnect, malformed_message, duplicate_response, response_after_cancellation). Only ping and protocol_ping_liveness overlap somewhat as both concern liveness, but their descriptions differentiate normal responsiveness from protocol-level in-flight liveness.

    Naming Consistency3/5

    All names are lowercase snake_case, but the set mixes bare verbs (ping, delay, hang, disconnect) with descriptive noun/adjective phrases (protocol_ping_liveness, malformed_message, duplicate_response, response_after_cancellation). This is readable but not a single predictable pattern.

    Tool Count5/5

    Eight tools is well-scoped for a failure-injection lab. Each tool represents a distinct failure scenario or liveness check, so none feel redundant or out of place.

    Completeness4/5

    The set covers common failure modes: timeout/delay, hang, disconnect, malformed protocol messages, duplicate responses, and post-cancellation responses. Minor gaps remain, such as a valid JSON-RPC error response tool or out-of-order response injection, but core failure-lab needs are met.

    Maintenance

    ActivityActive
    ResponsivenessWithin a week

    Resources