widgets: thirteen tools that run in the reader's browser, each from its own .t27 spec

[gHashTag]

Closes #1378 (epic #1377)

Thirteen widget tools for the WIDGETS tab, each a static page at t27.ai/widgets/<id>/ generated from specs/widgets/<id>.t27. The reader's files never leave the browser; no cookies; no network beyond same-origin files.

widget	tests	hook (from its spec)
bit-diff	5	Check the byte-identical claim yourself: drop t27's .bit and openXC7's and see every frame compared, or the first one that is not.
build-receipt	4	From a ternary spec to an Artix-7 200T bitstream with no Vivado: every stage timed, versioned and hashed.
cost-curves	7	A 64x64 multiplier with -nodsp costs 9,772 LUTs, 7.26% of an XC7A200T. Let it use DSPs and it is 16 DSP48E1 and 243 LUTs.
gf16-calc	6	Where does phi land in 16 bits? Type any number and watch it rounded to GF16, field by field, with the error printed exactly.
idcode	7	Paste your board's IDCODE, get its bit fields and its device as a link: #id=0x03636093 is the Artix-7 200T on our bench.
lut-explain	5	One routed design, 4738 LUTs: 1305 are XOR or XNOR, 2251 have no short name. Guess the gate from its 64 INIT bits.
pin-map	6	Our own Wukong XDC puts 13 of its 49 ports on balls with no user IO. Paste yours and see where every pin lands on the XC7A200T.
placement-map	4	A 373-cell design on a 33,650-slice die: see exactly which sites nextpnr picked, then drop your own routed JSON.
race-the-bee	4	Three times this week a person and a bee went for the same issue. The person started first every time; the bee merged first every time.
resource-bars	4	How much of a 200T does a UART-command blinker take? 0.12% of the LUTs. Drop your routed JSON and see your own bars.
slack-waterfall	5	Where do 4.8 ns go in a 20 ns clock? Every path of a real Artix-7 run, segment by segment, worst first; drop your report for yours.
test-touch	5	Bees landed 367 PRs on t27 in 14 days. 11 deleted tests or asserts; one titled 'Implement the 1 empty function body' deleted all 26 tests in its file.
test-waves	5	A spec's test block, run as gates and drawn as waves: every edge is from a real simulation, and your own .vcd opens in the same view.

The gallery spec lists them as tool in category runs; seven new ideas from the competitor sweep replace the seven now built.

Found while landing

• test-waves/demo.vcd was kept out of git by the root *.vcd rule while the page's ?v= stamp hashes it, so a clean checkout reported the page stale. Negations scoped to public/widgets/ now ship widget data (*.vcd, and *.log for synth-smells' samples); other .vcd/.log files stay ignored.

{
  "version": 1,
  "head_sha": "630beba2f7aa5e40d00675ccabfe0b0f4bc2c0b8",
  "summary": "Thirteen widget tools that run in the reader's browser, each generated from its own specs/widgets/<id>.t27 with its words, numbers and tests in the spec, listed in the Queen's WIDGETS tab as kind tool.",
  "changes": [
    "scripts/widget-pages-from-spec.mjs (new) compiles each widget-tool spec with the wasm compiler, runs its test blocks, refuses a spec missing TITLE/HOOK/DATA_SOURCES/SENDS_NOTHING, and writes public/widgets/<id>/index.html with og and twitter card tags, a t27:hook meta and the spec copy.",
    "13 specs under specs/widgets/ with 67 test blocks, 323 asserts; tool.js, tool.css, card.png and data files under public/widgets/<id>/; the scripts that produced the data under scripts/widget-data/.",
    "specs/widgets/gallery.t27 gains the kind tool and the shared page words; 31 widgets in three category runs, a tool's hook left empty so its page holds the sentence; seven new ideas (gatle, bit-autopsy, synth-smells, fasm-skyline, vcd-wrapped, stat-card, fits-on) replace the seven now built.",
    "src/data/siliconHistory.ts excludes dmitrii-f-t27/trinity-memory/t27/rtl/gf16_attn.t27 (a Q16.16 attention controller whose name says gf16), which turned main red on check:silicon after #1371.",
    "public/widgets/widget.css: shared frame, buttons and drop zones; every frame link is a 32px tap target.",
    "package.json gets check:widget-pages and test:widget-pages and the generator joins prebuild; website-checks runs both."
  ],
  "tests": [
    {
      "command": "node scripts/widget-pages-from-spec.mjs --check",
      "result": "13 tool pages up to date; spec tests 67, asserts 323, all hold",
      "status": "passed",
      "evidence": "Local run on this branch before the commit (tri widget check)."
    },
    {
      "command": "node --test scripts/widget-pages-from-spec.test.mjs",
      "result": "3 pass, 0 fail",
      "status": "passed",
      "evidence": "Local run on this branch."
    },
    {
      "command": "node scripts/widgets-from-spec.mjs --check && npm run test:widgets-spec",
      "result": "31 widgets in 3 categories, 7 ideas not built; spec tests 7, asserts 22, all hold; 5 pass, 0 fail",
      "status": "passed",
      "evidence": "Local run on this branch."
    },
    {
      "command": "npx tsc --noEmit -p apps/website",
      "result": "exit 0, no type errors in the website project",
      "status": "passed",
      "evidence": "Local run on this branch."
    },
    {
      "command": "each of the 13 pages loaded at ?embed=1 in a 640x460 iframe",
      "result": "every tool mounted into #widget, html.embed set, no horizontal overflow, no console errors",
      "status": "passed",
      "evidence": "In-app browser against a static server of public/."
    }
  ],
  "limitations": [
    "Widget data are snapshots taken by the scripts under scripts/widget-data/ on the dates their specs name; nothing re-runs them on deploy.",
    "lut-explain reads an AX7203 (xc7a200tfbg484) build in /tmp/x7board, not the Wukong; the page names the board.",
    "test-touch classifies diffs by a heuristic with known false positives, listed on its page.",
    "gatle, bit-autopsy, synth-smells and fasm-skyline are still being built and are listed as ideas, not widgets.",
    "Nothing is posted anywhere; cards are made for the owner to post."
  ],
  "tags": [
    "t27",
    "FPGA",
    "Widgets",
    "QueenBoard"
  ],
  "blog": {
    "title": "Thirteen FPGA widgets that run in your browser, each from a t27 spec",
    "summary": "Bitstream diff, IDCODE decoder, pin map, placement map, timing waterfall, LUT explainer, build receipt and more: each one a static page whose every word and number lives in a .t27 spec with its own tests.",
    "outline": [
      "Why tools, not recordings: a reader can drop their own file and get their own card.",
      "One spec per widget: the page generator refuses a tool without a hook, data sources or a no-network promise.",
      "Where the numbers come from: yosys, nextpnr-xilinx and git, each named on the page.",
      "What the competitor sweep found, and the next ideas waiting on the gallery shelf, each with its source."
    ]
  }
}

🤖 Generated with Claude Code

[github-actions]

X Brain Health Check

Score: 100.0/100
Status: 🟢 HEALTHY
Threshold: 80/100

X Brain is above merge threshold

[github-actions]

🚨 CRITICAL: Brain Health Below Threshold

The brain health CI gate has failed. This PR cannot be merged until:

1. All brain region unit tests pass
2. Integration tests pass (zig build test-brain)
3. Every stress test passes with no leaks (zig build test-brain-stress)
4. Brain health score >= 80/100, and every probe passes in CLI Smoke

Please review the failed job logs and fix the issues.

[github-actions]

🚨 CRITICAL: Brain Health Below Threshold

The brain health CI gate has failed. This PR cannot be merged until:

1. All brain region unit tests pass
2. Integration tests pass (zig build test-brain)
3. Every stress test passes with no leaks (zig build test-brain-stress)
4. Brain health score >= 80/100, and every probe passes in CLI Smoke

Please review the failed job logs and fix the issues.

[github-actions]

🚨 CRITICAL: Brain Health Below Threshold

The brain health CI gate has failed. This PR cannot be merged until:

1. All brain region unit tests pass
2. Integration tests pass (zig build test-brain)
3. Every stress test passes with no leaks (zig build test-brain-stress)
4. Brain health score >= 80/100, and every probe passes in CLI Smoke

Please review the failed job logs and fix the issues.

[github-actions]

🧪 Brain Stress Test Results

Tests: 261/261 passed, 0 leaked
Status: 🟢 PASS
Gate: every test in zig build test-brain-stress passes and none leaks