evolve

1# /evolve — Improvement Director
2 
3## Orientation
4 
5**Use when:** You want sustained autonomous quality advancement — the director
6forms hypotheses, scouts before attacking, and builds a belief model that
7compounds across cycles. Runs until a natural ceiling, budget exhaustion, or you
8say stop.
9 
10**Don't use when:** You want a single scored loop (`/improve`), a known axis
11attacked directly (`/improve --axis`), or a one-time audit (`/improve --score-only`).
12 
13**Key difference from `/improve`:** `/improve` follows the rubric mechanically.
14`/evolve` asks *why* scores are where they are, validates those theories before
15spending fleet budget, and extracts cross-skill patterns that propagate to skills
16never directly attacked.
17 
18## Invocation
19 
20```
21/evolve {target}                  # run until ceiling, velocity drop, or budget
22/evolve {target} --n={N}          # exactly N director cycles then stop
23/evolve {target} --budget=${X}    # run until cumulative spend reaches $X
24/evolve {target} --continue       # resume from saved director state
25/evolve {target} --status         # show belief model, velocity, spend — no attack
26/evolve {target} --axis={name}    # focus director on one axis (scout + attack only)
27```
28 
29`target` maps to `.planning/rubrics/{target}.md`.
30If no rubric exists, run `/improve {target}` Phase 0 first — `/evolve` requires
31an approved rubric and will not auto-generate one.
32 
33## Campaign Artifacts
34 
35All findings are externalized incrementally — written after every phase, not
36…