Skip to Content
Docs

Overview

StruggleTime tracks productive struggle: the loops where something breaks, you dig in, and it works again. Steady back-and-forth coding and fixing errors get more credit than large instant code drops you never ran, or long stretches away from the keyboard.

What gets credit

(edit | prompt) -> (test | build | run) -> (follow-up edit | commit)
PatternWeight (v0.4.0-alpha)
Struggle / fail→fix→pass1.00x
AI write + dwell + follow-up0.85x
Human edits, never run0.70x
Unverified multi-file dump0.20x
Quiet follow-up after real work0.70 → 0 over ~5–12 minutes
Idle after the active spanGradual decay (no hard 240s cliff)

How trust works

  • Passive file & git tracking: the desktop tracker watches saved file changes and git activity in your project folder. It does not screenshot your screen or read file contents.
  • Scrambled file IDs: paths are mixed with a secret salt (sha256(salt:path)) before anything is sent, so raw paths and source never leave your machine.
  • Tamper-evident history: each sealed time slice stores prev_hash of the previous entry. Quiet edits to past history are easy to spot in review.
  • Shared scoring rules: daemon and server both use st-core with the same FORMULA_VERSION.

Privacy

struggletime daemon is a native binary. Wire payloads are allowlisted metadata (~25–50 KB) with scrambled file IDs. See Privacy.

Pages

On this page