Skip to content

Update the locked template to 1cb0d36 (patched SDK 3dfa5f7) - #94

Merged
afflom merged 5 commits into
mainfrom
claude/project-thread-jbqzew
Nov 15, 2026
Merged

afflom merged 5 commits into
mainfrom
claude/project-thread-jbqzew

Conversation

@afflom

@afflom afflom commented Nov 15, 2026

Copy link
Copy Markdown
Owner

Requested by Alex: "push to main and keep this moving."

This applies the template-update workflow's own patch exactly as produced, with no edits. It is the artifact of run 43651229016, sha256 b2813ebe…9e23. It changes .devcontainer/devcontainer.json, .github/workflows/bootstrap.yml, prismpm.lock and template.lock.

It pins template 1cb0d36, which is still the head of UOR-Foundation/template#89 (checked with git ls-remote just now), and the patched SDK ghcr.io/afflom/prismpm-sdk-patched@sha256:3dfa5f72…. It unblocks #81 (builder-generated full model) and #86, which need SDK 0.4.0 plus the PrismPM closure fix.


Generated by Claude Code

Applied exactly as the template-update workflow produced it (run
43651229016's artifact, sha256 b2813ebe...9e23): .devcontainer,
bootstrap.yml, prismpm.lock and template.lock. Template 1cb0d36 is the
head of UOR-Foundation/template#89.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013Asa3LuNYtPDBa4FbSEQUZ
@afflom
afflom enabled auto-merge November 15, 2026 04:00
`lexlean lock` in ghcr.io/afflom/prismpm-sdk-patched@3dfa5f7: the compiler
semantics changed since locking; lock --check and fmt --check (58 modules)
now pass.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013Asa3LuNYtPDBa4FbSEQUZ
…mplate 1cb0d36)

bootstrap.yml at template 1cb0d36 mounts a private mktemp directory as the
SDK's /tmp instead of a 2 GiB tmpfs; the audit (private_scratch) and its
planted-defect test are ported from the same template revision. validate
passes inside the patched SDK.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013Asa3LuNYtPDBa4FbSEQUZ
…cb0d36)

PrismPM 0.4.0 keeps the per-run browser_timing out of the manifest's
acceptance digest, so the kept proof did not read back ("the application
acceptance differs") after a full, successful proof. proof.rs is the
template's at 1cb0d36, with this repository's app/specs exclusion kept.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013Asa3LuNYtPDBa4FbSEQUZ
@afflom
afflom merged commit bcd7f41 into main Nov 15, 2026
6 checks passed
@github-actions

Copy link
Copy Markdown
Contributor

Merging this pull request made main bcd7f41e8ca7ebd3c1b69000709aa466f13239c3, and main's deploy refused it, so the live page is not this change.

Refused by page / gates / Build and verify the holo through the locked SDK: https://github.com/afflom/me/actions/runs/43678873675/job/132704931040

These are the same gates this pull request passed. If its checks ran on the same merge result, the refusal comes from a check that does not decide the same way every run: make that check deterministic in a new pull request.

afflom pushed a commit that referenced this pull request Nov 15, 2026
PrismPM 0.4.0 refuses `prismpm --project app build` with PP3018 because
prismpm.lock is at the root, not beside the app. verify writes the holo
without a provenance statement; built with --no-provenance, #94's tree gives
the same build id (a487fb6a) its PR proof verified, and the existing
build-id check still binds the reused proof to this holo.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013Asa3LuNYtPDBa4FbSEQUZ
afflom pushed a commit that referenced this pull request Nov 15, 2026
A pull request proves in full, so the holo build main runs when it reuses
that proof never ran before merge: #94 passed its gates and then blocked
main's deploy (PP3018). The page step now runs that same build (one
build_holo function for both) on a copy of the tree alongside the proof,
and fails the pull request if it fails. It runs in parallel, so it adds no
wall time (about 4 minutes against a 20-minute proof).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013Asa3LuNYtPDBa4FbSEQUZ
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants