A planner you
can look inside.

One week at a time, with a reason.

The engine takes your current starting point, goal priorities, available days, equipment, and commitments. It compares candidate arrangements and explains the placement of sessions. The same inputs produce the same result.

A block gives the broad shape. Sessions are materialized a week at a time, rather than pretending the next three months will happen exactly as planned. The current engine stays within an established baseline; it does not automatically increase your training loads.

Scheduling estimates, not a readiness score.

Running and lifting can create different kinds of training demand. The engine models two estimated scheduling costs: systemic and structural. These help compare arrangements; they do not measure your body, predict performance, or calculate injury risk.

Recovery assumptions and coefficients are inspectable estimates. There is no acute-to-chronic workload ratio, readiness gauge, or learned physiological fatigue model. A logged fatigue skip changes future optional work; it does not diagnose why you feel tired.

Inspect the engine and its constants.

Safety cannot be traded for a better score.

Independent safety checks run after placement scoring. They can restrict or reject work even when the arrangement scores well. Pain reports halt future progression and planning. Fixed commitments cannot be quietly moved or rewritten to improve a score.

These are conservative software policies, not guarantees against injury. They do not replace qualified coaching, technique instruction, or medical advice.

AI can help interpret. It cannot prescribe.

A model you explicitly connect can interpret a free-text goal and suggest compatible exercise cards for your review. The engine, not the model, determines sets, reps, effort, duration, and placement. Invalid or unsupported suggestions are rejected.

There is no bundled model and no hidden AI service. API keys stay in the open tab’s memory, never in training backups. Every core planning and logging function works without a model.

Built for real interruptions.

A time-related skip and a fatigue-related skip mean different things. The former rearranges remaining work without adding catch-up volume. The latter reduces optional work. Deleting a session records a change, not a fatigue observation.

Rest does not reset a streak or erase a badge. There are neither streaks nor badges. Logs describe what happened, not what you owe the plan.

What this release does not do.

Hybrid Coach is an early supporting planner, not a complete race or championship coaching system. Running stays easy; it does not invent threshold paces or intervals from a free-text ambition. A sport example is not a validated sport-specific programme.

Garmin FIT and Apple Health imports are not implemented. There is no account connection, automatic cross-device sync, or technique assessment. Starting weights are not guessed from another exercise.

Evidence and accountability.

Our learning guides link to the research they discuss and distinguish study findings from practical examples. The project does not claim a clinical review, professional coaching certification, or scientific validation of its engine. Editorial responsibility sits with the Hybrid Coach project.

You can inspect the code, discuss a rule, or report a factual error in the public GitHub issues. Please do not post private training records, health details, or API keys. The code is available under the MIT license.