Repository navigation
The SDK e2e cleans up after itself, and the evidence page carries its recording - #73
Merged
Merged
Conversation
It runs against the live registry, so every run left a race called "Sterun SDK e2e <date>" on the public directory, and nothing could ever take it off: the organiser is a throwaway Friendbot account whose key dies with the process. Seven test races were cancelled by hand this morning to clear that board; recording the SDK run for the grant evidence would have put another one straight back. Cancelled rather than Completed, because the race did not happen. Saying a race ran when it did not is the claim this product exists to prevent, and the runners were throwaway accounts. It is the last step, so everything above still runs against a live Open event, and a failure there does not fail the run: by that point the e2e has proved everything it set out to prove, and turning a green run red over housekeeping would hide that. It prints the event id loudly instead. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01PkgGbz4eNSdzzzAPiqV1ZU
The material for the SOW's §6.2 checklist all exists, and none of it is findable. It is spread across a 162 KB deployments file, three rehearsal run directories, CI job summaries, npm and four live URLs. The SOW asks for evidence that is "easy to review by the Ambassador Chapter Lead with minimal technical expertise", and handing somebody a 162 KB file is the opposite of that. docs/EVIDENCE.md is the index: three deliverables, one row per piece of evidence, each row a link that opens with no install, no wallet and no access request, and a sentence saying what it proves. Every link in it was opened before committing. What is missing is in the table rather than left out: the recorded SDK run, the demo video and the screenshots, each with the date it is expected. So is what we do not claim — no escrow, every race in the demo is ours, testnet only, and two contract changes still waiting for review. The forwarded-screenshot section repeats the rehearsal's own wording rather than rounding it up: a screenshot goes stale in under a minute, and even a fresh one can only be used once. A reviewer who tests that claim will find it holds. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01PkgGbz4eNSdzzzAPiqV1ZU
The SOW's Deliverable 2 asks for a recording of the SDK creating an event, issuing a record and verifying it against testnet. It exists now, runs against the live contracts, and it is linked here rather than described. The rest of the page had drifted in three days: the rehearsal is on its fifth run rather than its third, the stale-screenshot step is in the evidence of every run since the 24th rather than waiting for one, the two contract changes it called "waiting for review" are live, and the demo races have posters and records now. Deliverable 2 is complete; Deliverable 3 waits on the video alone. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01PkgGbz4eNSdzzzAPiqV1ZU
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
This branch was successfully deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Three commits, all from recording the SOW's Deliverable 2 evidence this morning.
The e2e left a race on the live board
sdk/scripts/e2e.tsruns against the live registry, and every run left a race calledSterun SDK e2e <date>on the public directory for ever: its organiser is a throwaway Friendbot account whose key dies with the process, so nobody could ever take it off. Seven test races were cancelled by hand this morning to clear that board (STE-70); recording the run would have put another one straight back.It now cancels the event it created, as its last step.
Cancelledrather thanCompleted, because the race did not happen and saying it did is the claim this product exists to prevent. A failed cleanup does not fail the run: by then the e2e has proved everything it set out to prove, and turning a green run red over housekeeping would hide that. It prints the event id loudly instead.Verified on chain after the recording: event 46,
Sterun SDK e2e 2026-09-27,Cancelled, and the directory still opens on the three demo races.One page the Chapter Lead can read
docs/EVIDENCE.md: the SOW's §6.1 per deliverable, each row a link that opens with no install, no wallet and no access request, and every gap named. It was written on the 24th and sat on a branch; this brings it level and adds the recording.What the page does not do is round anything up: it still says there is no escrow, that every race in the demo is ours, that this is testnet, and it still repeats the rehearsal's own wording about a forwarded screenshot rather than claiming it is always refused.
🤖 Generated with Claude Code
https://claude.ai/code/session_01PkgGbz4eNSdzzzAPiqV1ZU