Milestones

List view

• D. Portability and packaging

  Answers 'can someone else run RAD?' rather than 'does RAD's loop work?'. Deliberately deferred behind milestone 2 so it does not compete for attention.

  No due date

  •0/5 issues closed
  0% complete5 open0 closed

• P2. Friction and housekeeping

  Small, unblocked, touches nothing on the spine. Good work for short sessions.

  No due date

  •0/3 issues closed
  0% complete3 open0 closed

• P1. Planning quality

  Independent of the signal spine. Improves what enters the pipeline and how plans are shaped. Parallelizable at any time.

  No due date

  •1/9 issues closed
  11% complete8 open1 closed

• 4. Actuation and auto-clear

  Convert suggestions into work items, then let calibrated signal drive approval. Last by necessity: auto-clear requires calibration data from milestone 3.

  No due date

  •0/3 issues closed
  0% complete3 open0 closed

• 3. Folds and readouts

  Pure read-side folds over the enriched event log, plus the surfaces that render them. All cheap once milestone 2 lands, all misleading before it.

  No due date

  •0/8 issues closed
  0% complete8 open0 closed

• 2. Trace enrichment

  Record wave outcomes richly enough to learn from (per-task statuses, failure snapshots, contract-level usage). Single keystone issue; hard dependency for every fold in milestone 3.

  No due date

  •0/1 issues closed
  0% complete1 open0 closed

• 1. Trustworthy signal

  Make RAD's recorded outcomes and documented claims mean what they say. Everything downstream mines this data — until it is honest, the insights and green-lane arcs are built on noise. Blocks milestone 2.

  No due date

  •6/11 issues closed
  54% complete5 open6 closed