agents/momentum: auto-infer per-user momentum baseline #114

New Issue

alvis · 2026-05-05T09:26:02Z

alvis commented

2026-05-05 09:26:02 +00:00

Goal

Auto-infer a per-user momentum baseline (rolling completions/day) so "momentum" snippets reflect deviation from each user'''s normal — not an absolute count.

Why

A 3-task day is a slump for a power user and a peak for a casual one. The current MomentumAgent has no way to know which.

Inferred parameters

Key	Shape	Default
`baselineCompletionsPerDay`	float	1.0
`stdev`	float	1.0
`momentumWindow`	days int	7

Method (sketch)

Rolling 28-day window of tip_feedback.action='''done''' count by day.
Mean + stdev → baseline.
Snippet text keys off (today_count − baseline) / stdev.

Done when

Inference function plugged into the framework with TTL = 7d (cheap, daily recompute).
Cold-start uses default until ≥ 14 days of history.
MomentumAgent rewords snippet using z-score language ("above your usual", "slowing down").
Tests cover: power user, casual user, returning-from-break user.

Depends on

meta: shared context-inference framework for agents (#111)

## Goal Auto-infer a per-user momentum baseline (rolling completions/day) so "momentum" snippets reflect deviation from each user'''s normal — not an absolute count. ## Why A 3-task day is a slump for a power user and a peak for a casual one. The current `MomentumAgent` has no way to know which. ## Inferred parameters | Key | Shape | Default | |---|---|---| | `baselineCompletionsPerDay` | float | 1.0 | | `stdev` | float | 1.0 | | `momentumWindow` | days int | 7 | ## Method (sketch) - Rolling 28-day window of `tip_feedback.action='''done'''` count by day. - Mean + stdev → baseline. - Snippet text keys off (today_count − baseline) / stdev. ## Done when - Inference function plugged into the framework with TTL = 7d (cheap, daily recompute). - Cold-start uses default until ≥ 14 days of history. - `MomentumAgent` rewords snippet using z-score language ("above your usual", "slowing down"). - Tests cover: power user, casual user, returning-from-break user. ## Depends on - #111

alvis added this to the M2 — AI tips + multi-source signals milestone 2026-05-05 09:26:02 +00:00

alvis referenced this issue from a commit

2026-05-05 11:21:14 +00:00

feat(agents): per-agent inference — momentum, overdue-task, recent-patterns, focus-area (ADR-0014 step 7)

alvis closed this issue

2026-05-06 05:27:14 +00:00

alvis referenced this issue from a commit

2026-05-06 06:55:01 +00:00

feat(agents): per-user baseline + stdev inference for momentum agent (#114)

Sign in to join this conversation.