[binlecode]

Summary

At _GRID_MIN_WIDTH (96 cols) each dashboard box gets ~44 content columns, and the longest rows are wider than that, so Textual clipped their tails. Observed on the published hero GIF (dashboard split against the process table, ~47 cols/column):

• P-CPU 36% @2121MHz (117°C) avg 35% · max 40 — missing %
• GPU 100% @1578MHz (103°C) avg 100% · max — max value gone entirely
• P-CPU [bar] idle64 low1 mid5 — high16 gone
• ↑ Out 33.8 KB/s avg 33.9 KB/s · max 66.0 — missing unit

Carried open since 1.5.0.

Behavior change

Metric rows are width-adaptive instead of hard-clipped. The rolling context degrades through three tiers, each row picking the widest that fits its own measured width:

avg N% · max N%   ->   ⌀N ▲N%   ->   ▲N%   ->   (nothing)

Applies to every labelled row (P-CPU/E-CPU, GPU, ANE, RAM, Mem BW, the four I/O rates, CPU/GPU/Package power). The reading itself is never shortened to make room for its annotation. Residency rows size their density bar to what a worst-case breakdown leaves (16 chars down to 8, dropped below that), so the four percentages always survive.

The fit is computed from each row's high-water widths, not the current frame's, so a row sitting on a tier boundary holds its form instead of flipping between shapes as digits come and go (avg 8% -> avg 10%, 987MHz -> 1987MHz). Verified live: 12s at 96 cols on an M4 Max, form stable throughout.

? help and docs/SPEC-system.md §6.3 document the ⌀/▲ compact form.

Also in this PR

• Hero GIF re-recorded off the fixed build (v1.9.8 banner, 4.4 MB / 176 frames); README alt text corrected — the tape has recorded grid + per-core + theme cycling since 1.9.x, but the alt text still described the old stack take.
• Release-noise rule (CLAUDE.md): normal PRs no longer bump the version, only release PRs do; a new "When to cut a release" section gates releases on user-visible change. 1.9.2–1.9.7 — six releases in forty minutes, each fixing the release pipeline that shipped the one before it — are recorded there as the failure mode. The 1.9.8 notes tell users to skip 1.9.1–1.9.7.

Validation

• .venv/bin/pytest -q — 150 passed
• .venv/bin/ruff check . / ruff format --check . — clean
• .venv/bin/python -m actop.actop --help
• Ran on macOS 15 / Apple M4 Max at 60, 80, 96, 110, 120, 160 cols (grid and stack, with and without swap in use): no row overflows its column, residency breakdowns intact.
• New test test_narrow_grid_columns_degrade_context_instead_of_clipping_rows mounts the real dashboard at 96 cols via App.run_test() and drives real SystemSnapshots through update_metrics; test_residency_bar_has_no_gaps_or_overflow_at_full_width renamed and given a settled layout, since a fixed 16-char bar is no longer the contract.

🤖 Generated with Claude Code