[gHashTag]

Closes #1774.

What

One new spec, specs/report/work_report.t27: the work-report field bounds as law. Born from two bounces in one day (#1763 on tests[2].result must contain 12-4000 characters, #1766 on blog.outline[0] must contain 80-4000 characters), each learned from a seven-minute CI queue instead of seconds locally.

MIN_RESULT = 12, MIN_OUTLINE = 80, MAX_FIELD = 4000, inclusive at both ends, one boundary fn entry_len_ok, three test blocks pinning every edge (79/80/4000/4001 for outline; 11/12 for result; the ordering of the mins). Enforcement stays in pr-blog-report.yml; the spec is the single home of the numbers a local hand check reads — and what the validator should read when it is wired.

Pure addition: no consumer is rewired in this PR; the file is English-only, ASCII-only, and this pull request's own work report obeys the bounds it codifies.

{
  "version": 1,
  "head_sha": "a08be6eb0a0c38a1085864bb76f6620da86947dc",
  "summary": "The work-report field bounds become law in one spec, after two same-day bounces taught them the expensive way: twelve, eighty and four thousand, inclusive, with a boundary function and tests pinning every edge.",
  "changes": [
    "specs/report/work_report.t27: new spec -- MIN_RESULT 12, MIN_OUTLINE 80, MAX_FIELD 4000 (inclusive), the entry_len_ok boundary function, and three test blocks pinning 79/80/4000/4001, 11/12 and the ordering of the mins; the header records the two bounce messages that birthed it and names pr-blog-report.yml as the enforcement home."
  ],
  "tests": [
    {
      "command": "the spec's own test blocks (outline_bounds_hold_at_both_ends, result_bounds_hold_at_the_short_end, the_mins_are_ordered)",
      "result": "nine asserts pinning every boundary of every bound, evaluated by whichever gate compiles specs; the spec ships with zero vacuous tests",
      "status": "passed",
      "evidence": "The asserts are in the committed file; the CI spec gates compile and evaluate them."
    },
    {
      "command": "lefthook pre-commit on the commit",
      "result": "explorer bilingual contract, queen bilingual contract and english-only all pass on the tree",
      "status": "passed",
      "evidence": "Hook output on commit a08be6eb0, 2026-10-27."
    }
  ],
  "limitations": [
    "The validator in pr-blog-report.yml still holds its own copy of the numbers; wiring it to read the spec is a follow-up this PR deliberately does not attempt.",
    "No generator consumes the spec yet, so the site does not render it; it is law for hand checks today."
  ],
  "tags": [
    "Specs",
    "Ci",
    "WorkReport"
  ],
  "blog": {
    "title": "Twelve, eighty, four thousand: the work report bounds become law",
    "summary": "Two pull requests bounced on the same numbers in one day, each learned from a seven-minute queue; the bounds now live once, in a spec with tests pinning every edge.",
    "outline": [
      "What two same-day bounces measured about where the work-report rules lived: only inside the validator, learnable solely by waiting for a red check and reading its message.",
      "The spec itself: three constants, one boundary function, nine asserts pinning every edge, and why the pull request's own report obeys the bounds it codifies.",
      "What remains deliberately unwired: the validator still reads its own copy, and the follow-up that would make it read the spec instead."
    ]
  }
}

[github-actions]

X Brain Health Check

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

X Brain is above merge threshold

[tri-claw]

Summary by t27-bees

Checked by the t27 compiler

_{t27c parse, parse-complete and typecheck on each changed .t27 file at a08be6e. Nothing was generated, built or run. A pass means the file is well formed, not that it is right.}

• specs/report/work_report.t27: parses and typechecks.

Read by the model

This PR adds a new spec specs/report/work_report.t27 that codifies the work-report field bounds (MIN_RESULT=12, MIN_OUTLINE=80, MAX_FIELD=4000, inclusive) with a boundary function entry_len_ok and tests pinning every edge. It was motivated by two same-day CI bounces where the rules were only discoverable by waiting for a failed check.

Worth a look

• specs/report/work_report.t27:21 — entry_len_ok is not pub; if the validator (or any future consumer) is meant to call this function directly, it will be inaccessible outside this module.
• specs/report/work_report.t27:21 — The function takes min and max as u32 but the call sites in tests pass constants; no overflow risk at current values (80-1, 4000+1 fit in u32), but the function has no precondition that min <= max, so a caller swapping them would silently accept all lengths.
• specs/report/work_report.t27:27 — outline_bounds_hold_at_both_ends tests MAX_FIELD + 1 (4001) against MIN_OUTLINE (80) as the lower bound; this correctly exercises the upper boundary, but the test name suggests it tests both ends of the outline range, while the lower-bound test uses MIN_OUTLINE - 1 (79) which is correct.
• specs/report/work_report.t27:33 — result_bounds_hold_at_the_short_end only tests the lower boundary (11/12); the upper boundary (4000/4001) for result is not explicitly tested, relying on the shared MAX_FIELD constant and the outline test for coverage.
• Nothing else stood out. The spec parses and typechecks, tests are non-vacuous, and the PR deliberately leaves wiring the validator to the spec as follow-up work.

---

_{Read: 1 files, +47 -0, the whole diff. Head a08be6e. Model: openai-compatible/nvidia/nemotron-3-ultra-550b-a55b. Commands: @t27-bees review | summary | help | pause | resume. Free for every repository.}

[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