[gogoout]

Stacked on #176, which is stacked on #175. Retarget this PR to main before merging it, after #176 is merged. A squash merge into the Monokai branch would strand it there.

What changes

• Two new themes: flexoki and flexoki-light, Steph Ango's Flexoki (MIT). There are now thirteen themes.
• Bundled syntax themes. two-face does not ship Flexoki, so both syntax themes are bundled .tmTheme files, the mechanism from ADR 0040. They are written here from Flexoki's published palette and generated from one table, so both use the same token colours. Both credit Flexoki in their headers.
• Seeds: Flexoki's 400 accents on the dark ground (#100F0F), and its 600 accents on paper (#FFFCF0).
• Five inks moved to pass the legibility tests. Each seed file records the number that forced it:
  • Dark: the red #D14D41 (4.42:1) and the blue #4385BE are lifted, with hue kept. The blue fell to 2.81:1 when muted for a foreign hunk.
  • Light: the green #66800B (4.39:1) and the yellow #AD8301 (3.39:1) are darkened.
  • The light green also turns 4° toward green. The reviewed mark is this green made 0.10 darker, and Flexoki's own hue cannot reach the 0.10 chroma floor in sRGB at that lightness.
• Search fill: Flexoki's orange on dark and its 400 yellow on light, so the fill differs from the skim tier's colour.

Docs

• Every theme list is updated: both READMEs, docs/cli.md, spec/tui.md and spec/consumers.md.
• Gallery entries are added, and ADR 0040 notes the two Flexoki themes.
• assets/themes.sh records them. assets/themes/flexoki.png and flexoki-light.png are new.

Tests

• cargo test, clippy -D warnings and fmt are clean.
• The legibility, chroma, distinctness and own-colour tests run over both new themes.
• The config tests and the theme dropdown test cover the two new names; the G key now lands on flexoki-light.
• Privacy sweep of the diff and commit message: clean.

[gogoout]

@claude review

[claude]

Claude finished @gogoout's task in 2m 11s —— View job

---

Review of flexoki / flexoki-light themes

• Gather context (diff vs origin/feat/monokai-pro, read CLAUDE.md rules)
• Read changed files (theme seeds, config, docs, ADR, tests)
• Check legibility/contrast claims and test coverage
• Check docs/spec/ADR consistency
• Post review feedback

Verification limitation: this sandbox required approval for every non-trivial shell command (cargo test, cargo clippy, even python3), and none was granted, so I could not execute the test suite, clippy/fmt, or independently recompute the contrast numbers. The review below is from reading the diff and the existing derive/test machinery in crates/tui/src/theme/mod.rs. Please confirm CI is green before merging.

Overall

Clean, well-scoped addition that follows the established seed/derive pattern exactly (compare flexoki.rs/flexoki_light.rs against the existing monokai.rs). Nothing here reaches for a new abstraction, and the bundled-.tmTheme mechanism is exactly ADR 0040's escape hatch, used the way monokai already uses it.

• crates/tui/src/theme/flexoki.rs / flexoki_light.rs: seeds are commented with the same discipline as the rest of the palette set — each ink shift states the measured contrast that forced it and that hue/chroma were kept, which is exactly what derive's test suite (every_theme_is_legible_on_its_own_ground, a_semantic_ink_is_tellable_from_ordinary_text, a_search_hit_is_readable_and_unmistakable) checks for every theme in ALL. Since both new variants are folded into that array via ThemeName::Flexoki/FlexokiLight, they get the full legibility/chroma/distinctness battery for free — no bespoke tests needed.
• crates/tui/src/theme/mod.rs: Flexoki/FlexokiLight are appended after Monokai (not alphabetically inserted), matching the "adding a theme is adding a variant" convention elsewhere — and crates/tui/tests/tui.rs was correctly updated so the G-key-cycles-to-last-theme test now expects FlexokiLight instead of Monokai.
• crates/engine/src/config.rs: ThemeName enum and its round-trip/name tests were extended consistently; the "list every valid name in the error message" test (user_config_parses_the_theme_and_names_the_valid_ones) already includes flexoki/flexoki-light.
• Docs (README.md, crates/tui/README.md, docs/cli.md, spec/tui.md, spec/consumers.md) and adr/0040...md are all bumped from "eleven" to "thirteen" and list the two new names — I didn't find a stale "eleven" or missing mention anywhere in the diff.
• assets/themes.sh / assets/theme.vhs: theme list and comments updated in step; new flexoki.png/flexoki-light.png screenshots added.

Minor observations (non-blocking)

• Both bundled .tmTheme files set findHighlight/findHighlightForeground (e.g. flexoki.tmTheme:32-35), but nothing in derive() reads those keys — the reviewer's own highlight_bg/highlight_fg come from the seed's highlight field, not the syntect theme's find-highlight. This matches monokai.tmTheme's existing shape, so it's not a regression, just metadata carried for .tmTheme-format completeness rather than something consumed. Worth double-checking it wasn't meant to line up with seed.highlight (dark seed uses Flexoki's orange #DA702C, while the bundled file's findHighlight is the yellow #D0A215) — but since it's inert, it's cosmetic at most.
• The PR body's own note stands out and matters: this branch targets feat/monokai-pro, not main — it must be retargeted after [feat] tui: monokai wears Monokai Pro's palette #176 merges, or a squash merge would strand it on that intermediate branch per the stacking note.

Nothing else jumped out as incorrect. The contrast/chroma deltas described in the seed comments (e.g. red 4.42:1 → lifted, green 4.39:1 → darkened, yellow 3.39:1 → darkened) are internally consistent with the direction deepen/mix push things in derive, and the pattern of "measure, then justify the literal in a comment" matches how every other theme in this file already documents its exceptions.

[gogoout]

The findHighlight note is fixed: the two Flexoki files (and monokai.tmTheme, on #176) now name the seed's own search fill. The key stays inert, since derive() reads the fill from the seed. Rebased onto #176.

[gogoout]

Superseded: GitHub cannot change the base of a stacked PR, and this one targeted the branch #176 was merged into. The same commits are rebased onto #179 in the replacement PR.

[gogoout]

Replacement: #180.