request_status
LoadingLoading Agent Interface contract
modeled_rows
A planned interface for domain-aware running questions grounded in governed gold models, typed evidence, and explicit analytical boundaries.
interface_status
contract_design: active
production_endpoint: none
MCP tools and resource contracts are in active development. No production MCP server or endpoint is live. This page defines the intended read-only boundary and does not claim a deployed agent runtime, authentication layer, or multi-user serving model.
Each request follows an explicit path from natural-language intent to approved analytical data and application-owned presentation.
step::01
A running question expressed in the consuming application.
step::02
An agent and MCP client identify the bounded analytical task.
step::03
A typed tool validates parameters and selects an approved query path.
step::04
Documented signal, route, session, and calendar models supply governed metrics.
step::05
The interface returns values, provenance, freshness, caveats, and visual intent.
step::06
The mobile or web client renders the chart, map, table, and written analysis.
Bounded tools make common training questions reproducible while keeping metric definitions, time grains, and caveats visible.
query::01
Volume trendsWeekly line chart with rolling context and a concise trend summary.
query::02
Route comparisonsRoute map, comparison table, and pace and heart-rate differences.
query::03
ConsistencyCalendar or weekly bars with active weeks, missed weeks, and streak evidence.
query::04
Fitness efficiencyPace-versus-heart-rate trend with sample context and analytical caveats.
The highlighted tools and resources are representative examples, not an exhaustive final contract. The interface will remain intentionally bounded as additional approved query paths and context are defined.
highlighted_tool_examples
get_training_summaryReturn bounded consistency, volume, and fitness summaries for a requested period.
get_volume_trendReturn daily, weekly, or monthly distance and duration series with defined rollups.
compare_route_runsCompare performance across runs on the same governed route.
get_fitness_trendReturn descriptive pace, heart-rate, efficiency, and recovery context over time.
highlighted_resource_examples
running-signals://definitions/signalsMetric definitions, analytical assumptions, and interpretation boundaries.
running-signals://catalog/gold-modelsThe MCP layer returns typed analytical data. The consuming mobile or web application remains responsible for rendering accessible charts, maps, tables, and written analysis.
summaryevidencevisualization_intentprovenancefreshnesscaveatsGold models define the approved analytical surface. The contract excludes operations and interpretations that are unsafe, ungoverned, or outside this project’s scope.
related_extension
Validated offline findings may improve future gold metrics and the context available to these tools. Experimental outputs are never served directly.
Review ML readinessApproved models, grains, fields, and lineage available to the tool layer.
running-signals://status/freshnessLatest modeled dates and refresh context for evidence returned to a user.
running-signals://status/qualityRelevant test coverage, missing-data flags, and known limitations.