01Safety
0.0 / 30
Public leaderboard
Public assessment
forgemeshlabs/travel-agent-mcp (@forgemeshlabs/travel-agent-mcp)
What changed in the harness
Selection accuracy 100→98, token cost up 12%, unconfirmed writes 0%→0%.
Category breakdown
Where the score comes from.
Earned points across the four signals Gradable measures. Safety and Legibility are scored out of 30; Economics and Discoverability are scored out of 20.
02Legibility
25.4 / 30
03Economics
18.6 / 20
04Discoverability
11.6 / 20
Highest-impact fix
Estimated gain +30 pointsAdd explicit identity and permission preflight tools
Expose machine-readable principal/tenant confirmation and a non-mutating permission check so agents can verify both before destructive actions.
Description evidence
Defects and rewrites.
10 defects found across the exposed tool descriptions. Suggested rewrites make purpose, inputs, boundaries, and returns easier for an agent to understand.
| Tool | Defect types | Suggested rewrite |
|---|---|---|
list_travel_categories |
no_return_description |
FREE live read from the ForgeMesh travel-agent backend. Returns the available travel flow categories, indicating which are the free airport, timing, or coverage tools and which are the paid x402 planning services, so the caller can pick the next useful call. No input parameters. Read-only, no account setup, no booking side effects. |
get_airport_details |
no_return_description |
FREE lookup that returns the stored details for one airport given its IATA code, so the caller can validate or explain a single airport before route planning; use plan_flight_route for an actual trip route and compare_airport_routes for multiple origin-destination pairs. Read-only, no account setup, no external booking links, and no booking side effects. |
compare_airport_routes |
no_return_description |
Compare two or more airport route pairs side by side. Returns a comparison of the given route pairs so a traveler can pick between alternate origins, destinations, or nearby airports; responses may also include commission-eligible external booking links for each route. Use plan_flight_route for one dated route and get_airport_details for a single airport lookup. Read-only, no account setup, no booking side effects. |
get_travel_timing_advice |
no_return_description |
FREE practical travel timing guidance. Returns timing advice covering booking windows, airport arrival timing, layovers, and seasonal considerations for the traveler's route, using the optional origin and destination codes to determine whether the route is international. Use for questions like when to book or how early to arrive; use plan_flight_route when the traveler wants route planning. Read-only, no account setup, no external booking link, and no booking side effects. |
convert_currency |
no_return_description |
FREE live reference-rate currency conversion from the ForgeMesh travel-agent backend. Returns the converted amount for the given amount and source/target ISO 4217 currency codes, optionally at the requested date, so a traveler or agent gets quick spending context before deciding whether to ask for richer x402 trip guidance. Read-only, no account setup, no payment, and not a transaction quote. |
plan_day_trip |
params_unexplained |
Connect to the live paid x402 day-trip service for a destination, route, or origin-based ask. Use when a traveler wants a relaxed day out with timing, water, food, transit, weather, and backup options. Set the trip with origin and/or destination, or with a coordinate point via lat and lon; radius_miles sets the search radius in miles around the location, max_distance_or_travel_time caps the distance or travel time from the origin, date is the trip day in YYYY-MM-DD, time_window narrows the day (e.g., morning, full day), transportation_modes lists the allowed modes, interests and theme steer what the day should focus on, content_style sets the output style, and budget sets the spending level; include_local_context (boolean) requests local context such as neighborhood tips. The MCP can return live server output or x402 request metadata for clients that can pay with x402. |
plan_weekend_getaway |
params_unexplained |
Connect to the live paid x402 weekend-getaway service for a short overnight escape. Use when the traveler wants a calmer planning brief with lodging posture, water/weather fit, transportation, and easy backup ideas. Provide origin and/or destination, or a coordinate point via lat and lon, to set where the getaway starts and goes; start_date and end_date frame the overnight window in YYYY-MM-DD, transportation_modes lists permitted modes, max_distance_or_travel_time caps the travel scope, budget sets the spending level, lodging_style sets the accommodation preference, trip_style sets the overall style of the getaway, and destination_currency is the currency for cost estimates. The MCP can return live server output or x402 request metadata for clients that can pay with x402. |
plan_weather_aware_trip |
params_unexplained no_return_description |
Connect to the live paid x402 weather-aware planning service for a destination or coordinate pair. Use when the traveler wants rain, heat, wind, and indoor backup awareness before committing to outdoor time. Give destination, or a coordinate point via lat and lon, date in YYYY-MM-DD, time_window (e.g., morning, full day), plan_type to indicate the kind of plan, and optionally transportation_modes and max_distance_or_travel_time to constrain travel. Returns the weather-aware trip plan, or x402 request metadata for clients that can pay with x402. |
find_transit_options |
params_unexplained no_return_description |
Connect to the live paid x402 mobility-options service for a location and transportation type. Use when the traveler wants transit, rail, ferry, trails, bike/scooter options, or a location-filtered short list before planning the rest of the trip. Provide location (the place to search around), or origin and destination to define the connection to check; transportation_type selects the mode (e.g., transit, rail, ferry, trails, bike, scooter) and trip_style tailors the short list. Returns the matching transportation options, or x402 request metadata for clients that can pay with x402. |
request_local_coverage |
params_unexplained no_return_description |
FREE live backend call for requesting coverage for a town, neighborhood, local ritual, free experience, or source set. Use this when you want the system to remember a place worth watching: local vibe, freebie list, sunset walk, opening-soon pattern, or creator angle. Name the place via any of town, location, destination, state, country, county, or neighborhood; signal_types lists the signals to watch for, source_links supplies source material, creator_angle names the content angle, content_style sets the desired style, contact gives a way to reach back, and notes carries extra context; recheck_days (number) sets how often to recheck. Returns confirmation that the coverage request was recorded so the place can be watched. |
Selection evidence
Confusable tool pairs.
6 pairs where similar names or overlapping descriptions may send an agent toward the wrong tool.
| Tool A | Tool B | Confidence | Why they collide |
|---|---|---|---|
plan_day_trip |
plan_weather_aware_trip |
high | Both are paid x402 planning services accepting a destination or coordinates, and both descriptions list weather, timing, and indoor backup options. A task like 'plan a day trip with rain and indoor backup awareness' maps equally well to either because plan_day_trip explicitly includes weather/backups and plan_weather_aware_trip supports day-length plans. |
creator_experiences |
plan_day_trip |
medium | creator_experiences defaults to a day trip and covers timing, food stops, scenic movement, weather-aware backups, and content style, heavily overlapping plan_day_trip's day-out scope. A task like 'plan a shareable day trip with good food stops and photo moments' could trigger either, and both accept destination and content_style. |
plan_weekend_getaway |
plan_weather_aware_trip |
medium | plan_weekend_getaway advertises water/weather fit and easy backup ideas while plan_weather_aware_trip is explicitly about rain/heat/wind and indoor backups with a flexible plan_type. A task like 'plan a weather-aware weekend escape with indoor backups' fits both descriptions without a deciding signal. |
plan_day_trip |
plan_weekend_getaway |
low | Both are paid x402 planning services with nearly identical schemas (origin/destination/transportation/budget) and both describe calm planning with backups; a task like 'plan a short relaxing trip' with no stated duration is ambiguous, though day vs weekend wording usually disambiguates. |
plan_day_trip |
check_trip_disruptions |
low | plan_day_trip includes weather and backup options while check_trip_disruptions covers weather alerts, disruption awareness, and calm backup considerations; a task like 'check weather and backup plans for my day out' could be read as either planning the day or running a disruption pulse check. |
plan_flight_route |
compare_airport_routes |
low | A task like 'show me my route options for flying LAX to JFK' could route to compare_airport_routes because it offers side-by-side options, even though it targets multiple origin-destination pairs rather than flight choices for one pair; the descriptions try hard to disambiguate, so confusion is only occasional. |
Compare the field