Repository navigation
feat(source): support verified generated release assets - #85
vincentkoc wants to merge 1 commit into
Conversation
|
🦞👀 Pull request received. I will update this pull request when review starts. |
|
@clawsweeper review |
|
🦞👀 Re-review progress:
|
|
Codex review: needs maintainer review before merge. What this changesThis PR lets source releases carry one caller-generated file, optionally built from an exact artifact in the same workflow run. Example: A caller requests fixture-source.txt and supplies same-run input artifact 12192733980.
Review scores
ProductKind: Feature · Worth it: Yes · Fix scope: Complete Merge readiness✅ Ready for maintainer review Keep this maintainer-authored draft open: it adds a distinct capability absent from main, preserves existing defaults, and introduces no concrete correctness defect found in review. Priority: P2 Before mergeNone. FindingsNone. Tests
Agent review detailsHow this fits togetherThe reusable source-release workflow accepts caller inputs and a frozen protected commit, builds without write credentials, assembles immutable artifacts, verifies their inventory on two runners, and publishes only attested bytes. flowchart LR
A[Protected release target] --> C[Read only build]
B[Same run input artifact] --> C
C --> D[Immutable raw artifact]
D --> E[Assembly and inventory]
E --> F[Two independent verifiers]
F --> G[Attested release publication]
Technical reviewBest possible solution: Retain the opt-in transport in the existing source-release owners, with package qualification enforced by each caller. Do we have a high-confidence way to reproduce the issue? This is a feature addition; source inspection establishes the missing baseline capability, and the supplied hosted run exercises its generated-byte transport. Is this the best way to solve the issue? The change fits the existing source archetype: it adds opaque transport while retaining artifact-ID binding, verifier ownership and publication checks instead of creating a second release path. AGENTS.md: not found in the target repository. Codex review notes: model internal, reasoning medium; reviewed against fc49859568b3. Provenance checked
TestingProof path: in-process harness. SecurityNone. EvidenceWhat I checked:
Likely related people:
Review metrics
LabelsLabel changes:
Label justifications:
Rating scale6/6 🦀 challenger crab · 5/6 🦞 diamond lobster · 4/6 🐚 platinum hermit · 3/6 🦐 gold shrimp · 2/6 🦪 silver shellfish · 1/6 🧂 unranked krab. Overall follows the weaker of proof and patch quality; ✨ marks media proof (a screenshot, video, or linked artifact) that directly shows the changed behavior. WorkflowClawSweeper edits this one comment on every review. Comment Reviewed October 19, 2026, 10:23 AM ET / 14:23 UTC. |
Summary
Add opt-in transport for one caller-generated release asset while preserving the source-only default.
generated-assetnames one bounded regular file written to the workflow-ownedRELEASE_ASSET_PATH.generated-input-artifact-iddownloads one exact artifact from the current workflow run into a separateRELEASE_ASSET_INPUT_DIR. No cross-run/repository selector or new credential is accepted.This is opaque byte transport. A package caller must qualify the exact input bytes, validate its own source/tree/hash/lock contract against the frozen release target, and depend on its producer and platform consumers before publication. A fresh rebuild at the same commit is not equivalent proof. No ClawSweeper caller is enabled here; its package name, version and publication decisions remain separate. The existing source-archive Homebrew handoff is unchanged.
Base:
fc49859568b3781af7663bee48c003fd987dc08b. Signed head:5c3ee0fd825b520603872c1d2d0693c203a15ab5.Validation
mapfile(unavailable in system Bash 3.2) with a large heredoc that stalls in installed Bash 5.3. The owned stalled test was stopped; no fixture assertions or host tooling were changed.67185d3ec421fd5e775de65eb9bff9f8ed8491e8.12192733980(digest0926172745e5715deb2f70b7c0bf71f8801b72132ee9f33ea421d16a2ff187a9), ran the production builder, uploaded/downloaded raw artifact12192748802(digest74ecbb09c99c776aa7d92656317501865ab604ecb1a99711ed210521fe023647), then passed production assembly and both verifier blocks. Both blocks ran on Ubuntu; this does not claim an independent ARM job or a complete consumer release rehearsal.Production workflow delta: +91/-2 lines (net +89), implementing the optional input/output boundary. Test/CI proof delta: net +137; docs/example/changelog: net +13.
No tag, release,
v1movement, installation, new permissions or publication workflow dispatch is part of this change.