Skip to content

Latest commit

 

History

History
117 lines (102 loc) · 6.75 KB

File metadata and controls

117 lines (102 loc) · 6.75 KB

SourceMender submission-readiness source notes

Audit snapshot: August 22, 2026 at 17:35 UTC.

Reporting job

  • Question: Is SourceMender ready to hand in to Into the Scrape-Verse?
  • Audience: project owner preparing the official submission.
  • Decision: whether the technical package is complete and which remaining actions are still required for formal eligibility.
  • Scope: official rules, schedule, and live submission form; current private GitHub repository; checked-in live evidence; long-form and compliant short-demo media; local build/test results; and repository safety.
  • Baseline: the mandatory submission and participation requirements visible on the official event pages and submission form on August 22, 2026.
  • Success criterion: every technical requirement has direct evidence, any external or user-controlled gate is explicit, and no unsupported claim is presented as submission-ready.

Official sources reviewed

The overview and schedule say submissions close on August 23, 2026. The now-open submission form supplies the precise deadline: 8 p.m. BST on Sunday, August 23, which is 19:00 UTC. It requires a public GitHub repository and a YouTube video no longer than three minutes. It also asks for the project and Scraper Studio story, track selection, and required first-person product feedback.

Repository and evidence reviewed

  • Repository: https://github.com/michaeljamieson10/sourcemender
  • Audit base commit: 54bffd9e15a065cd1cc052c3684d9686ab777f6c
  • Repository created: August 17, 2026 at 15:28 UTC.
  • First commit: August 17, 2026 at 15:27 UTC.
  • Long-form 4K demo: https://www.youtube.com/watch?v=I7_fNpiN6uQ
  • Submission demo: https://www.youtube.com/watch?v=I-wdyLsXuuA, the 64.120-second 1080p Playwright-only Run → Approve → Verify walkthrough. The exact MP4, English WebVTT captions, source hashes, and provenance are checked in under apps/evidence-experience/public/media/.
  • Custom Scraper Studio collector: c_msxk8bqf29gzigb16j.
  • Live accepted document: evidence/live-bright-data/2026-08-17-competition-heal-proof-02/sourcemender/signal/runs/run-20260817-182340/document.json
  • Live verification receipt: evidence/live-bright-data/2026-08-17-competition-heal-proof-02/sourcemender/signal/runs/run-20260817-182340/verification.json
  • Corrected additive report variant: evidence/live-bright-data/2026-08-17-competition-heal-proof-02/sourcemender/signal/runs/run-20260817-182340/report.fixed-envelope.v2.html
  • Corrected report provenance: evidence/live-bright-data/2026-08-17-competition-heal-proof-02/sourcemender/signal/runs/run-20260817-182340/report.fixed-envelope.v2.provenance.json
  • Claim boundaries: evidence/live-bright-data/claims.json.

Verified facts

  • The live proof went from 5 of 9 required fields to 9 of 9, a recovery of 4 fields or 44.4 percentage points for one controlled public-page record.
  • The repaired result passed all 14 deterministic verification checks.
  • Three repeated strict live collections passed; all 27 required field slots were filled and the semantic payload was stable after excluding the volatile extraction timestamp.
  • One independent post-heal collection passed in 1.635 seconds.
  • The Google Forms registration receipt dated August 4 confirms Michael Jamieson registered as an individual entrant with team name Solo. The receipt remains private and is not copied into this repository.
  • On August 22, make submission-check passed: formatting, warning-denied Clippy, debug and release tests, Rust 1.88 MSRV tests and doctests, warning-free public and private docs, evidence recomputation, dependency policy, and current RustSec advisories.
  • On August 22, the React experience passed a clean install and production build. The Playwright suite reported 43 passed and one intentional mobile skip for the desktop-only section-navigation affordance. Production-only and full npm audits both reported zero vulnerabilities.
  • On August 22, Git object integrity and the full-history secret scan passed. No GitHub Actions workflow is present.
  • The repository remains private pending explicit owner authorization. GitHub metadata confirms visibility: private at the audit base commit.
  • The existing unlisted 4K film is 6 minutes 32 seconds and therefore exceeds the submission form's new three-minute limit.
  • The submission cut is 64.120 seconds, H.264 1920 by 1080 at 25 fps with AAC 48 kHz stereo. It passed full decode, measures -19.5 LUFS integrated with a -1.2 dBFS true peak, and has a reviewed 18-cue English WebVTT track. Its SHA-256 is bab540cb74274c31414ece76cfef02fa585708f907f7af892beb6010d8f1d5a1.
  • The short demo is live unlisted, signed-out accessible, embeddable, and fully processed. YouTube exposes a 1080p rendition and automatic English captions; the reviewed human-authored WebVTT remains checked in and active in the local interactive player.

Claim boundaries

The evidence supports one controlled live repair on one official public page. It does not establish universal repair success, production-scale reliability, or an always-heals claim. The official page was selected by the operator; it was not a discovery hit. The extracted document contains a normalized exact deadline time; that record remains separate from the submission deadline, whose exact 8 p.m. BST time now comes directly from the live organizer form.

Report construction notes

The requirement matrix remains the exact audit lookup. One simple categorical bar chart answers the aggregate question: how many of the eleven reviewed gates are ready versus externally pending? It uses the readiness_counts dataset, status as the category axis, and count as the quantitative axis. The chart uses the shared artifact reader's restrained single-root palette and axis labels, so color is not required to understand it. No other chart is warranted because the report contains neither a time trend nor a distribution.

Two static-HTML verifier failure screenshots produced while selecting the report surface remain preserved in the local checkout. They are excluded from Git because they document a shared report-runtime width defect, not SourceMender behavior, submission evidence, media provenance, or a reproducible project input. The canonical native report artifact validated and rendered successfully.