Skip to content

fix(website): the catalog refresh runs the checks its counts must pass - #1217

Merged
gHashTag merged 3 commits into
mainfrom
fix/onboarding-counts-1737
Oct 1, 2026
Merged

gHashTag merged 3 commits into
mainfrom
fix/onboarding-counts-1737

Conversation

@gHashTag

@gHashTag gHashTag commented Oct 1, 2026 •

Copy link
Copy Markdown
Owner

The apex publisher (gHashTag/ghashtag.github.io publish-website.yml, run 36893486012, 2026-10-01 16:38Z) failed at Build: onboarding-from-spec: 9 problem(s). #1176 (t27 worlds: refresh the catalog) had moved public/t27/manifest.json to 1737 specs and 14 repositories, and specs/catalog/onboarding.t27 still said 1577 and 12. The same refresh left check:silicon red on two vendored gf16 ports.

#1212 has since written the counts and the two silicon exclusions; this branch merged it. What remains here is what #1212 did not do:

file what
.github/workflows/t27-world-scan.yml npm run check:onboarding and npm run check:silicon in Catalog gates. #1176 merged with Website checks red, because only T27 work report is required. The refresh never ran either check, so the publish was the first place the disagreement surfaced. A refresh that moves the counts now stops before it commits.
specs/catalog/onboarding.t27 MEASURED_AT 2026-10-01: #1212 kept 2026-09-24 beside the 2026-10-01 numbers
same prose the new numbers made false. It said "every remaining fail never parsed", but 175 − 169 = 6 specs now parse and lose only Verilog (queen-rehearsal, gen_w312, run_v51_multiple, health_check, gen_mock, trios-mesh/tools/deploy)
public/llms.txt, public/agents.t27 regenerated from the spec

Verified: npm run prebuild:ci exit 0; onboarding-from-spec --check up to date (10 tests, 54 asserts hold); check:silicon classifies every spec; Website checks run 36900947748 green on the pre-merge head.

🤖 Generated with Claude Code

{
  "version": 1,
  "head_sha": "342451a266b355d19ee549e5d106b5833741bf38",
  "summary": "The t27.ai publish failed because a catalog refresh (#1176) moved the manifest without the onboarding snapshot, and the refresh job never ran the check that compares them. #1212 has since fixed the numbers; this adds check:onboarding and check:silicon to the refresh's own gates, and corrects the date and the prose the new numbers falsified.",
  "changes": [
    ".github/workflows/t27-world-scan.yml: npm run check:onboarding and npm run check:silicon added to Catalog gates",
    "apps/website/specs/catalog/onboarding.t27: MEASURED_AT 2026-10-01; a paragraph recording the 2026-10-01 refresh; 'every remaining fail never parsed' replaced -- six specs now parse and fail only Verilog",
    "apps/website/public/llms.txt and public/agents.t27 regenerated by node scripts/onboarding-from-spec.mjs"
  ],
  "tests": [
    {
      "command": "cd apps/website && npm ci && npm run prebuild:ci",
      "status": "passed",
      "result": "exit 0 for the whole chain, onboarding-from-spec included",
      "evidence": "local run; failing main build before #1212: gHashTag/ghashtag.github.io run 36893486012"
    },
    {
      "command": "cd apps/website && node scripts/onboarding-from-spec.mjs --check && npm run check:silicon",
      "status": "passed",
      "result": "up to date, spec tests 10 asserts 54 all hold; every spec naming a verified format is classified",
      "evidence": "local run on 342451a26"
    },
    {
      "command": "Website checks (workflow_dispatch on the branch)",
      "status": "passed",
      "result": "success on 7ef402eeb, before merging main",
      "evidence": "https://github.com/gHashTag/trinity/actions/runs/36900947748"
    }
  ],
  "limitations": [
    "The world-scan gate is not exercised here; its next refresh that moves the counts will stop at check:onboarding until someone writes them, so the catalog can go stale where the site used to break.",
    "Website checks remains non-required on main, which is how #1176 merged red; this PR does not change branch protection."
  ],
  "tags": [
    "Engineering",
    "Verification"
  ],
  "blog": {
    "title": "The gate was on the publish path, not on the job that changed its input",
    "summary": "A bot refresh moved the numbers a spec states about the corpus, merged with the comparing check red, and the site's publish was the first job to refuse it.",
    "outline": [
      "The apex publish failed on a stale corpus snapshot that a bot refresh had invalidated.",
      "The counts were fixed in parallel; this change moves the comparing checks into the refresh itself and corrects the prose the new numbers falsified.",
      "Non-required checks let a red bot PR merge; whether Website checks should be required is left to the owner."
    ]
  }
}

… specs

trinity#1176 refreshed public/t27/manifest.json (1577 -> 1737 specs,
12 -> 14 repositories) and left specs/catalog/onboarding.t27 stating the
old counts. onboarding-from-spec refuses a snapshot that disagrees with
the manifest it ships beside, so `prebuild` on main exits 1 and the
apex publisher (ghashtag.github.io, run 36893486012) failed at Build.

- write the nine counts the generator names, and MEASURED_AT
- correct the prose the new numbers made false: "every remaining fail
  never parsed" -- six now parse and lose only Verilog
- regenerate /llms.txt and /agents.t27 from the spec
- t27-world-scan runs check:onboarding with the other catalog gates, so
  a refresh that moves the counts stops before it commits instead of
  after it reaches main

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@github-actions github-actions Bot added the status:in-progress 🔵 Agent working label Oct 1, 2026
gHashTag and others added 2 commits October 2, 2026 00:24
…check:silicon

check:silicon (Website checks) fails on main since #1176: gf16_dot4 and
gf16_matmul_top name GF16 and are in no table. Both consume the format
(a dot-product unit and the board top around it) and declare nothing, so
they join the use/convert/measure exclusions with reasons. The world scan
now runs check:silicon beside check:onboarding.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…MEASURED_AT, the corrected prose and the scan gate

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@gHashTag gHashTag changed the title fix(website): t27.ai stopped publishing when the catalog grew to 1737 specs fix(website): the catalog refresh runs the checks its counts must pass Oct 1, 2026
@gHashTag
gHashTag merged commit 2f868bb into main Oct 1, 2026
27 of 33 checks passed
@github-actions github-actions Bot added status:completed Done and removed status:in-progress 🔵 Agent working labels Oct 1, 2026
github-actions Bot added a commit that referenced this pull request Oct 1, 2026
fix(website): the catalog refresh runs the checks its counts must pass (#1217)

* fix(website): t27.ai stopped publishing when the catalog grew to 1737 specs

trinity#1176 refreshed public/t27/manifest.json (1577 -> 1737 specs,
12 -> 14 repositories) and left specs/catalog/onboarding.t27 stating the
old counts. onboarding-from-spec refuses a snapshot that disagrees with
the manifest it ships beside, so `prebuild` on main exits 1 and the
apex publisher (ghashtag.github.io, run 36893486012) failed at Build.

- write the nine counts the generator names, and MEASURED_AT
- correct the prose the new numbers made false: "every remaining fail
  never parsed" -- six now parse and lose only Verilog
- regenerate /llms.txt and /agents.t27 from the spec
- t27-world-scan runs check:onboarding with the other catalog gates, so
  a refresh that moves the counts stops before it commits instead of
  after it reaches main

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(website): classify the two gf16 ports #1176 vendored; scan gates check:silicon

check:silicon (Website checks) fails on main since #1176: gf16_dot4 and
gf16_matmul_top name GF16 and are in no table. Both consume the format
(a dot-product unit and the board top around it) and declare nothing, so
they join the use/convert/measure exclusions with reasons. The world scan
now runs check:silicon beside check:onboarding.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant