Honest weight trend and the chance it's really moving, from dated morning weights. Nothing stored.
Scored across 2 tools
The split is mostly clear (watch/tracker signals vs. scale-only weight trend), but both tools claim the same trigger phrases — 'why the scale stalled' appears in body_signals while slim_chance also handles stalls — and body_signals additionally produces a weight trend, duplicating slim_chance's core output. An agent must read carefully to decide which tool to call when a user shares both a watch and scale readings.
Both names are snake_case, so formatting is consistent, but the conventions differ: body_signals is a descriptive content noun while slim_chance is just the server brand name, which conveys nothing about its function. With only two tools the mismatch is tolerable but not a predictable verb_noun pattern.
Two tools is on the thin side for the stated scope, though the domain (stateless trend analysis of weight and body signals) is genuinely narrow. It sits at the borderline where a third tool or a merged single tool would feel more deliberate.
Core workflows are covered: weight-only trend with uncertainty quantification, and multi-signal watch trends with a cross-check against weight. Minor gaps remain — no way to compare signals across two separate periods, no baseline/goal-context handling — but these are workarounds rather than dead ends, and the explicitly stateless design limits what's needed.
Use when someone with a watch or fitness tracker asks what their resting heart rate, steps, sleep or HRV say, or why the scale stalled while they trained. Gives each daily signal its own honest trend over the last 5 weeks (rising, falling, steady, or too noisy to call), with a plain meaning: resting heart rate (falling usually means fitness improving), daily steps (falling is a common hidden reason weight loss slows), sleep hours and HRV. With morning weights too, it also gives the weight trend and says when the watch explains the scale. Separate signals, not a health score; not medical advice. Needs 7 days of a signal for its trend. Only send what the person shared and asked about. Nothing you send is stored.
Inputs: hrv, unit, steps, weigh_ins, +2 more.
Use when someone shares dated scale readings, or asks whether their weight loss (or gain) is really working, has stalled, or is just water. Turns dated morning scale readings into an honest trend: where the weight really is under the daily water noise, the rate per week with its 95% range, and the chance in 10 that it is truly moving the way the person wants. It reads the last 5 weeks up to the latest reading; it describes the recent past, it is not a forecast. Needs at least 7 readings in those 5 weeks for a verdict; with fewer it says how many more mornings it needs. Only send what the person shared and asked about. Nothing you send is stored.
Inputs: weigh_ins (required), goal, unit, goal_weight. Returns: unit, as_of, state, trend, +8 more
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Know your goal-weight date. Log food, sync Garmin, get a daily AI coach in Telegram.
Deterministic fitness calculators — TDEE, adaptive TDEE, body fat, 1RM, macros — with consensus.
Your Fit Shapes food log, targets, and weight trend, and logging food and weigh-ins when you ask.
Training analytics over your Hevy log: e1RM, PRs, volume, consistency, bodyweight.
Add one secure layer between your agents and this server.