Repository navigation
chore(mint): initialise secret-types from rsr-template-repo c481073 - #17
Conversation
secret-types was created from rsr-template-repo 05bf1b9 (2026-09-24; the Initial commit's tree is that commit's tree) and never minted, so Canon lockstep still fetches the profile from hyperpolymath/rsr-template-repo/<this repo's sha> and can never pass (as #14 recorded), and openssf-compliance fails on unfilled tokens. The template has since retired the .a2ml descriptiles for clauses of the repo deed (D312) and reads the canon pin from the deed through github.repository. This commit lays the template's tree at c481073 over the repository and restores, unchanged, every file secret-types owns: everything added since the Initial commit (src/SecretTypes, tests/reject, tests/minimal-laws, scripts/check.sh, minimal-laws.sh, proof-flags.sh, grade-receipt.sh, dev-toolchain.sh, secret-types.agda-lib, proofs.yml, PROOF-STATUS.adoc, verification/receipts/agda-grade3.a2ml, docs/*), README.adoc and docs/EXPLAINME.adoc, and the six reusable-workflow callers #15 pinned to metadatastician/standards@46ee183 (governance, hypatia-scan, mirror, rust-ci, scorecard, secret-scanner): owner decision 2026-11-04, keep #15's callers. Licence files are untouched (doctrine 6): LICENSE, LICENSES/ and docs/legal/ are byte-identical to main, and the template's LICENSE.adoc is not added. It is not minted yet: the next commit runs repo-init on this tree. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BYbQRtNu216R13jtZLmoqs
Run `just repo-init` non-interactively on the previous commit's tree (PROJECT_NAME=secret-types, REPO=secret-types, OWNER=hyperpolymath, AUTHOR=Jonathan D.A. Jewell, ORCID 0000-0002-3078-6652, PROJECT_TYPE=library, description from README) with RSR_TEMPLATE_COMMIT=c481073: the root commit's tree is the older template 05bf1b9, and the resolver would otherwise pin that. Owner decisions applied by hand, as repo-init leaves them to a human: clade fv (Formal Verification & Proofs), secondary sc (Security), prefixed name fv-secret-types (decided 2026-10-27); :born 2026-09-27, the Initial commit; ecosystem :purpose from README. Corrections to what repo-init wrote: * the self-name pass rewrote rsr-template-repo in six of secret-types' own documents (EXPLAINME, the transition review, four reports) where the name means the real template; restored. * restored secret-types' own wiring of template files: `just test` runs scripts/check.sh and scripts/minimal-laws.sh; *.agdai ignored; the root allowlist names PROOF-STATUS.adoc and secret-types.agda-lib; docs/README.adoc links the transition review; proofs.just keeps #14's proof-receipt recipes (now byte-identical to main's). * LICENSE.adoc is optional in the root allowlist: renaming LICENSE is a licence edit, manual and owner-only (doctrine 6). * actions.lock: the template's lock with proofs.yml and #15's six callers (and their records) merged in. * docs/reports/audit/consistency-audit-2026-11-01.adoc quotes an unfilled token as prose; listed in .machine_readable/mint-verbatim.txt. * CLAUDE.md and the repository map regenerated from the minted deed. Checked locally: Canon lockstep's comparison against this deed passes (2.1.3, criteria 6a5aa885, gates 2f31f46d); DEED lint passes on the deed; no unfilled tokens; root shape, adoc renders, action pinning, repo-map determinism, self-name, deed-clause, mint-cleanup and non-interactive mint tests pass; `just test` passes both proof gates. Known red, owner-accepted or pending an owner decision: * check-no-a2ml / DEED lint: verification/receipts/agda-grade3.a2ml (#14's receipt, written by scripts/grade-receipt.sh). * check-lock-sync SELF-SHA REFS: #15's records for metadatastician/standards@46ee183 and hyperpolymath/proven-cicd-and-workflows@55556cf, @0da816c carry their own SHA as ref:, and no branch or tag points at those commits. GitHub runs them on main today (Governance, 2026-11-03). Left UNASSIGNED for the owner: ecosystem :not, :position-type, :pipeline-position, :chain; anchor :kind / :domain. Registration in gv-clade-index is the owner's act. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BYbQRtNu216R13jtZLmoqs
|
Important Review skippedToo many files! This PR contains 376 files, which is 276 over the limit of 100. To get a review, reduce the PR to 100 files or fewer by splitting it into smaller PRs or changing its base branch. Upgrade to a paid plan to raise the limit. Usage-priced reviews support at most 300 files.
|
| - Repeated harassment of others. In general, if someone asks you to stop, then stop. | ||
| - **When we disagree, try to understand why.** Disagreements, both social and technical, happen all the time and Hyperpolymathy Estates is no exception. It is important that we resolve disagreements and differing views constructively. Remember that we’re different. The strength of Hyperpolymathy Estates comes from its varied community, people from a wide range of backgrounds. Different people have different perspectives on issues. Being unable to understand why someone holds a viewpoint doesn’t mean that they’re wrong. Don’t forget that it is human to err and blaming each other doesn’t get us anywhere. Instead, focus on helping to resolve issues and learning from mistakes. | ||
|
|
||
| Original text courtesy of the [Speak Up! project](http://web.archive.org/web/20141109123859/http://speakup.io/coc.html). |
The mint's actions.lock was the template's lock with main's records merged in. hyperpolymath/proven-cicd-and-workflows@55556cf and @0da816c came with them, but on main they served main-estate-audit.yml's old reusable call; the template's main-estate-audit.yml checks cicd-suite out at a pinned commit instead and lists only actions/checkout. Nothing in this tree references either record. check-lock-sync now reports a single SELF-SHA record, metadatastician/standards@46ee183, the target of #15's six reusable-workflow callers. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BYbQRtNu216R13jtZLmoqs
Port hyperpolymath/rsr-template-repo#284 (check-lock-sync.sh clause 5 and its four controls) as it stands at bf04a1d: a dependencies: record that is only the target of job-level reusable-workflow calls may record its own SHA as ref:, as GitHub runs it on main today (Governance run 41678782554). That is metadatastician/standards@46ee183, the target of #15's six callers. The port no-ops once the template carries #284 and secret-types takes a template update. Prune nine records the mint's union merge carried over from main and nothing in this tree references (download-artifact@v8.0.2, fetch-metadata@v3.1.0, action-editorconfig-checker@v3.0.0, setup-beam@v1.24.1, deed-ecosystem@main, k9-ecosystem@main, smtp-notify-action@v0.5.1, setup-bun@v2.2.0, sonarqube-scan-action@v8.4.0). bash scripts/check-lock-sync.sh -> in sync and transitively closed bash tests/workflows/check_lock_sync_test.sh -> 132 PASS Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BYbQRtNu216R13jtZLmoqs
|
Update on known red #2 (
Locally: The one known red left is #1, Generated by Claude Code |
|
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
… ref (#284) ## Summary Clause 5 of `scripts/check-lock-sync.sh` (#271) refuses every `dependencies:` record whose `ref:` is a full SHA. #271's evidence was *action* records: Protoctist.jl#18's jobs failed while preparing their actions, and metadatastician/lintel#15 hit the same failure. A reusable workflow pinned by SHA has no tag or branch to name, and GitHub runs it with its own SHA as `ref:`. Evidence: hyperpolymath/secret-types `main` at `d1d843b`, Governance run 41678782554 (2026-11-03), where every job got past "Set up job" with `metadatastician/standards@46ee183` recorded as `ref: 46ee183…`. So the rule rejects a lock the runner accepts. Minting secret-types from this template (hyperpolymath/secret-types#17) turned its lock-sync gate red over a working pin. Owner decision 2026-11-05: fix it here, at source. ## Changes - `scripts/check-lock-sync.sh`: pass 2 records whether each workflow `uses:` ref is the target of a job-level reusable-workflow call (`…/.github/workflows/x.yml@`) or an action. A self-SHA record is exempt only while it is reusable-only: never a step action, and never named as a nested `uses:` by another record. The header's clause 5 states the exception and its evidence; the summary line names it. - `tests/workflows/check_lock_sync_test.sh`: four controls (below). `lock-sync-gate.yml` already runs this file. ## RSR Quality Checklist ### Required - [x] Tests pass: `bash tests/workflows/check_lock_sync_test.sh` → 132 PASS under mawk; `bash scripts/check-lock-sync.sh` on this repo passes. - [x] Code is formatted. - [x] Linter is clean: `bash -n` on both files. - [x] No banned language patterns. - [x] No `unsafe` blocks. - [x] No banned functions. - [x] SPDX headers present (unchanged). - [x] No secrets. ### As Applicable - [ ] Documentation: the script header carries the rule; no other doc names clause 5. - [ ] `CHANGELOG`: not touched. ## Testing New controls: - a reusable-only target with a self-SHA `ref:` → pass; - the same ref also used by a step as an action → `SELF-SHA REFS`; - the same ref named in another record's nested `uses:` → `SELF-SHA REFS`; - a job-level path outside `.github/workflows` → `SELF-SHA REFS`. Negative control: against the unmodified checker, the first control fails (`expected 0, got 1`), so it tests this change. The patched checker passes on this repo's own lock and on the secret-types mint tree. Scope note, so nothing is overclaimed: the evidence covers a *direct* job-level reusable target. Nested self-SHA records stay refused because no run has shown GitHub accepting one. --- _Generated by [Claude Code](https://claude.ai/code/session_01BYbQRtNu216R13jtZLmoqs)_ Co-authored-by: Claude <noreply@anthropic.com>
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: b719586ea3
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| - name: No .a2ml files | ||
| run: bash scripts/check-no-a2ml.sh . |
There was a problem hiding this comment.
Exempt or migrate the retained proof receipt
The newly enabled estate gate deterministically fails because verification/receipts/agda-grade3.a2ml is deliberately retained in this commit. Running this exact command reports that file and exits 1, so every estate-rules push/PR run stops here before reaching the later validation steps; migrate the receipt or explicitly exempt it before enabling the gate.
Useful? React with 👍 / 👎.
| wf_hypatia="$(sed -nE "s/.*HYPATIA_SHA: *'?([0-9a-f]{40})'?.*/\1/p" "$HYPATIA_WF" | head -n1)" | ||
| if [ -z "$wf_hypatia" ]; then | ||
| fail "hypatia-scan.yml declares no HYPATIA_SHA; the policy's scanner.commit cannot be verified" |
There was a problem hiding this comment.
Support the retained reusable Hypatia workflow
This validator assumes hypatia-scan.yml defines an inline HYPATIA_SHA, but the retained project workflow delegates directly to metadatastician/standards/.github/workflows/hypatia-scan-reusable.yml@46ee... and therefore has no such variable. On this tree bash scripts/check-hypatia-rule-order.sh . always exits 1; the newly wired edge and verdict tests make the same inline-workflow assumption about local artifact upload and path conversion, so the estate job has multiple guaranteed failures until these checks understand the reusable caller or the workflow is expanded locally.
Useful? React with 👍 / 👎.
| fi | ||
|
|
||
| echo " --- the badge URL and the orphan guarantee survive ---" | ||
| has "README still reads the stable raw URL" 'raw\.githubusercontent\.com/\{\{OWNER\}\}/\{\{REPO\}\}/governance-badges/badges/k9\.json' "$README" |
There was a problem hiding this comment.
Match the minted badge URL instead of template tokens
In a minted repository the {{OWNER}} and {{REPO}} tokens must be gone, so this assertion can never accept the actual stable badge URL (and the retained README currently has no governance badge at all). The estate workflow runs this suite directly, and its wiring phase consequently exits 1 with README still reads the stable raw URL; check the concrete hyperpolymath/secret-types URL or derive the expected owner/repository instead.
Useful? React with 👍 / 👎.
| - name: Hypatia exemption meta-rules (.hypatia-ignore) | ||
| run: | | ||
| set -euo pipefail | ||
| bash tests/workflows/hypatia_ignore_test.sh | ||
| bash scripts/check-hypatia-ignore.sh . |
There was a problem hiding this comment.
Remove the stale docs-seed Hypatia exemption
The mint deletes build/docs-seed/, but the retained .hypatia-ignore still exempts structural_drift/SD022:build/docs-seed/. The newly enabled command checks that every exemption names an existing path, so on this exact tree it reports the line-26 exemption as stale and exits 1; remove or retarget that exemption before wiring this gate into estate-rules.
Useful? React with 👍 / 👎.
| # The toolchain for this repository (idris2 zig). Rendered at mint by | ||
| # scripts/render-mise.sh from the languages the tree declares; the language→tool | ||
| # table is build/just/provision-lib.sh's, so this file and `just doctor` cannot | ||
| # disagree. Specs are `latest`; mise.lock records the concrete version and |
There was a problem hiding this comment.
Provision the Agda compiler used by the project tests
The minted toolchain declares only Idris2 and Zig and provides no Agda installation path, even though just test immediately invokes scripts/check.sh, whose first operation is agda --version and whose core sources are src/SecretTypes/*.agda. On a fresh machine, following the newly generated mise install / just setup instructions therefore leaves the primary test compiler absent, and the new generic just build delegation also never builds the Agda development; include Agda in the setup path (for example by wiring the existing scripts/dev-toolchain.sh agda) and describe it in the generated setup metadata.
Useful? React with 👍 / 👎.
…calls) (#19) ## Summary The mint (#17) carried the exception from hyperpolymath/rsr-template-repo#284 into `scripts/check-lock-sync.sh`. Codex review found two defects in that exception, and both were mine: - it never checked that a record's key is the same full SHA as its `ref:`; - it judged "reusable call" from the shape of the path, not from where the `uses:` sits in the YAML. They are fixed upstream in [hyperpolymath/rsr-template-repo#285](hyperpolymath/rsr-template-repo#285) (`c438af6`). This PR ports that fix as it stands. It becomes a no-op once a template update brings #285. ## Changes - `scripts/check-lock-sync.sh` and `tests/workflows/check_lock_sync_test.sh`: byte-identical to #285's. This repo's only exempt record, `metadatastician/standards@46ee183`, still passes. Its key and `ref:` are the same SHA, and all six callers are job-level. ## RSR Quality Checklist ### Required - [x] Tests pass: `bash scripts/check-lock-sync.sh` → in sync and transitively closed; `bash tests/workflows/check_lock_sync_test.sh` → 137 PASS. - [x] Code is formatted. - [x] Linter is clean. - [x] No banned language patterns. - [x] No `unsafe` blocks. - [x] No banned functions. - [x] SPDX headers present (unchanged). - [x] No secrets. ### As Applicable - [ ] `CHANGELOG`: not touched. ## Testing Local runs as above. #285 carries the negative controls: each new test fails against #284's checker. --- _Generated by [Claude Code](https://claude.ai/code/session_01BYbQRtNu216R13jtZLmoqs)_ Co-authored-by: Claude <noreply@anthropic.com>
…nside (#288) ## Summary Hypatia's `http_in_docs` rule (at `be98f74`, `lib/rules/cicd_rules.ex`) matches `\bhttp://` in `.adoc` prose. It flagged line 28 of `CODE_OF_CONDUCT.adoc` on hyperpolymath/secret-types#17 (code-scanning alert 42), which is the first repository minted from this template since that rule landed. Every future mint would inherit the same alert, so the fix belongs here at source. Owner decision 2026-11-05. ## Changes - `CODE_OF_CONDUCT.adoc` line 28: `http://web.archive.org/web/20141109123859/http://speakup.io/coc.html` → `https://web.archive.org/web/20141109123859/speakup.io/coc.html`. Switching only the outer scheme does not clear the alert, because `/http://speakup.io` inside the Wayback path still matches the rule. The Wayback Machine also accepts the archived original without a scheme, so this link names the same 2014-11-09 snapshot with no `http://` left in it. ## RSR Quality Checklist ### Required - [x] Tests pass: the rule's own regex, run against the three forms: the old line → flagged; an https-only rewrite → flagged; this form → clean. - [x] Code is formatted. - [x] Linter is clean. - [x] No banned language patterns. - [x] No `unsafe` blocks. - [x] No banned functions. - [x] SPDX headers present (unchanged). - [x] No secrets. ### As Applicable - [x] Documentation updated (this is the change). ## Testing Regex check as above. **Not verified:** web.archive.org is outside this session's network, so I could not confirm that the scheme-less URL redirects to the snapshot. Please click it once before merging. --- _Generated by [Claude Code](https://claude.ai/code/session_01BYbQRtNu216R13jtZLmoqs)_ Co-authored-by: Claude <noreply@anthropic.com>
…dge checker (#20) ## Summary Squabble's `check-workflow-edges` has been red on `main` since the mint (#17), for two reasons. **A real defect.** `hypatia-scan.yml` calls `metadatastician/standards/.github/workflows/hypatia-scan-reusable.yml@46ee183` without `hypatia-ref`. So the scan runs the reusable's default Hypatia (`adb2f41`), while `fork-governance.yml` and `verdict-grid-publish.yml` grade its findings as `--hypatia-ref be98f74`. The grid was naming a scanner that didn't run. **A checker blind spot.** The checker looked for the `hypatia-scan-findings` upload inside `hypatia-scan.yml` itself, but the reusable is what uploads it. That is fixed at source in [hyperpolymath/rsr-template-repo#290](hyperpolymath/rsr-template-repo#290), owner decision 2026-11-05. ## Changes - `.github/workflows/hypatia-scan.yml`: the call passes `with: hypatia-ref: be98f74ce93df200752f3798efa1305bfa3e0e10`, the grid's pin, with a comment saying why. - `scripts/check-workflow-edges.sh` and `tests/workflows/workflow_edges_test.sh`: ported from #290 as they stand (`b31c166`, `ff922d4`). The port becomes a no-op once a template update brings #290. ## RSR Quality Checklist ### Required - [x] Tests pass: `bash scripts/check-workflow-edges.sh .` → all workflow edges resolve; `bash tests/workflows/workflow_edges_test.sh` → 46 PASS; `check-lock-sync.sh` → in sync (the `uses:` line is unchanged). - [x] Code is formatted. - [x] Linter is clean. - [x] No banned language patterns. - [x] No `unsafe` blocks. - [x] No banned functions. - [x] SPDX headers present (unchanged). - [x] No secrets. ### As Applicable - [ ] `CHANGELOG`: not touched. ## Testing Local runs as above. The checker reads the reusable live at `46ee183` from raw.githubusercontent.com. Its offline controls live in #290. --- _Generated by [Claude Code](https://claude.ai/code/session_01BYbQRtNu216R13jtZLmoqs)_ Co-authored-by: Claude <noreply@anthropic.com>




Summary
secret-types was created from rsr-template-repo
05bf1b9(2026-09-24) and never minted. Two checks are red onmainbecause of that, and no change to the proofs can turn them green:hyperpolymath/rsr-template-repo/<this repo's sha>, which can never exist (proof(grade): mathesis Grade 3 receipt, canonical-flag attempt with a recorded waiver, and origin-record reconciliation #14 recorded this).{{…}}template tokens.Owner decisions: mint from the current template, clade fv/sc (2026-10-27), and mint now while keeping #15's callers (2026-11-04).
Two commits, so each can be reviewed separately:
fa62dd5lays the template's tree atc481073over the repository and restores, unchanged, every file secret-types owns (see Changes).ebca846runsjust repo-initnon-interactively, then applies the owner's decisions and the corrections listed in Changes.Changes
main:src/SecretTypes,tests/reject,tests/minimal-lawsscripts/check.sh,minimal-laws.sh,proof-flags.sh,grade-receipt.sh,dev-toolchain.shsecret-types.agda-lib,proofs.yml,PROOF-STATUS.adoc,README.adoc,docs/EXPLAINME.adocand the project docs and reportsmetadatastician/standards@46ee183)LICENSE,LICENSES/,docs/legal/): doctrine 6, so the template'sLICENSE.adocis not added, and the root allowlist marks it optional.secret-types_chora.deed:fv-secret-types;:born 2026-09-27;:purposefrom README.github.repository.RSR_TEMPLATE_COMMIT=c481073. The root commit's tree is the older template, and the resolver would otherwise have recorded05bf1b9.rsr-template-repoin six of secret-types' own docs, where the name means the real template; restored.just testruns both proof gates;*.agdaiignored; root allowlist entries; a docs link; proof(grade): mathesis Grade 3 receipt, canonical-flag attempt with a recorded waiver, and origin-record reconciliation #14'sproof-receiptrecipes.actions.lockis the template's lock withproofs.ymland fix(ci): canonical reusable callers, least-privilege Actions canary and lock sync #15's entries and records merged in..machine_readable/mint-verbatim.txt.CLAUDE.mdand the repository map are regenerated from the minted deed.Left UNASSIGNED for the owner: ecosystem
:not,:position-type,:pipeline-position,:chain; anchor:kind/:domain. Registration in gv-clade-index is the owner's act.RSR Quality Checklist
Required
just test→PASS: core proofs and all 3 expected rejections,PASS: minimal-laws certificate (3 laws, each needed)(Agda 2.6.4.3).unsafeblocks.As Applicable
CLAUDE.md, the repository map, and thedocs/status/ROADMAP.adocstub thatrepo-initinstalls.CHANGELOG: not touched.Testing
Run locally on the minted tree:
dogfood-gate.yml, against this deed and currentcanon.lock→This template implements the canon at HEAD.(2.1.3, criteria6a5aa885, gates2f31f46d).Known red, needing your call:
check-no-a2ml/ DEED lint fail onverification/receipts/agda-grade3.a2ml, proof(grade): mathesis Grade 3 receipt, canonical-flag attempt with a recorded waiver, and origin-record reconciliation #14's receipt written byscripts/grade-receipt.sh. You accepted this when choosing "mint now".check-lock-syncSELF-SHA REFS. The template's checker rejects anydependencies:record whoseref:is its own SHA. Three records carry that shape: fix(ci): canonical reusable callers, least-privilege Actions canary and lock sync #15'smetadatastician/standards@46ee183,hyperpolymath/cicd-suite@55556cfand@0da816c. No branch or tag points at any of those commits, so there is no truthful name to write. GitHub does run these callers onmaintoday: the 2026-11-03 Governance run got past "Set up job" on every job. So either the template's rule is stricter than the runner for reusable-workflow records, or fix(ci): canonical reusable callers, least-privilege Actions canary and lock sync #15 should pin to a tagged standards commit. Both fixes live upstream, not in this PR.Generated by Claude Code