Public completion record
Archive status and known limitations
· Counts are a dated build snapshot and will change with later releases.
Verified published snapshot
Published files
28,244
Current verified deployment payload, including the two Cloudflare configuration files.
HTML pages
21,546
19,966 are indexable; 1,522 are intentional noindex pages or aliases.
Sitemap files
14
One sitemap index and 13 child sitemaps, including seven non-overlapping page sitemaps and specialized evidence and media sitemaps.
Unique sitemap page URLs
19,966
Unique URLs across the seven page sitemaps; specialized evidence, image, video, and X-capture sitemap entries are reported separately and can overlap those pages.
PDF files
425
Every published PDF is linked from HTML and listed in the evidence sitemap. Complete manual document review remains unfinished.
Physical image files
1,247
The current media catalog contains 1,247 images. 1,217 have non-empty automated OCR; the OCR remains unreviewed.
Published video files
6
All local videos are in the video sitemap; synchronized caption coverage remains incomplete.
Verified official YouTube uploads
357
Public source records from the official ABI Resources channel; 0 reviewed transcripts are currently published on this site.
Measured completion by workstream
Progress is shown only where both a completed count and a defensible target exist. These bars measure inventory coverage, not accuracy, legal sufficiency, search indexing, or manual accessibility conformance.
MuckRock request reconstruction
PartialNo defensible completion percentage is available.
170 verified distinct MuckRock requests are documented in the current source inventory. A historical target of 200 is not supported as a verified request count. Complete communication threads and all direct attachments have not yet been established by the current evidence inventory.
Open supporting archive sectionImage OCR coverage
PartialAutomated OCR is an unreviewed discovery aid, not a verified transcript. Readers must compare it with the source image.
Open supporting archive sectionYouTube transcript publication
IncompleteVideo titles and embeds are published, but they are not substitutes for synchronized captions or reviewed transcripts.
Open supporting archive sectionPeople Index cross-references
PartialThe candidate registry is not an exhaustive extraction of every person in every PDF, image, video, FOIA thread, or article. A reference is not a finding of wrongdoing.
Open supporting archive sectionPDF editorial and accessibility review
PartialNo defensible completion percentage is available.
All 425 PDFs are public and discoverable in the verified release snapshot. Complete OCR, metadata, tagging, reading-order, and manual accessibility review is not yet proven.
Open supporting archive sectionExternal search and AI indexing
External stateNo defensible completion percentage is available.
The archive can publish crawlable pages, sitemaps, manifests, and structured data. It cannot guarantee crawling, indexing, ranking, retention, citation, or interpretation by a third party.
Open supporting archive sectionMachine-readable status sources
- Evidence-control architecture 8 human-reviewed public control records
- Media catalog current source manifest
- MuckRock request archive current verified reconstruction
- People cross-reference index current conservative candidate registry
- Sitemap index current crawler discovery index
Explicit limitations
- Evidence-control labels govern archive presentation; they are not judicial, agency, or expert determinations.
- No percentage is reported when the project lacks a defensible completed-item count.
- Automated OCR is kept separate from reviewed transcription.
- Published and sitemapped does not mean indexed by a search engine.
- Automated accessibility checks do not replace manual assistive-technology testing.
- A SHA-256 digest verifies bytes, not authorship, context, truth, or legal effect.
- The referenced G: source drive was unavailable during the latest MuckRock reconciliation.
Known incomplete or externally controlled areas
MuckRock request reconstruction
Partial170 verified distinct MuckRock requests are documented in the current source inventory. A historical target of 200 is not supported as a verified request count. Complete communication threads and all direct attachments have not yet been established by the current evidence inventory.
Open the MuckRock archivePeople and entity registry
PartialThe current People Index contains 75 conservative candidate name strings derived from the historical AccountabilityActors import. It is not an exhaustive extraction of every person named in the Livewire, FOIA, PDF, image, and video corpus.
Open the People IndexUnified PDF forensic catalog
PartialAll 425 PDFs are public and sitemapped. The unified forensic manifest has been expanded to include published evidence_derived and evidence_documents PDFs; the release gate now fails if any published PDF is absent.
Review forensic readinessVideo accessibility
IncompleteSix locally archived video files and 357 verified official-channel YouTube uploads are public. The current audit does not verify synchronized caption tracks or reviewed transcripts for them. A transcript-boundary notice is not a substitute for captions or a reviewed transcript.
Open the video archiveUnique search metadata
PartialEvery indexable page has the baseline title, description, canonical URL, H1, structured data, and machine-readable manifest links. Duplicate-title and duplicate-description groups remain and require page-specific editorial revision.
Read the editorial methodologyExternal indexing and AI use
External stateThe archive can provide public HTML, sitemaps, robots directives, structured data, and AI manifests. It cannot guarantee that Google, Bing, ChatGPT, or another system will crawl, index, cite, retain, rank, or interpret every record.
Read the verification boundariesPrioritized completion work
- 1.Review orphan communication and binary artifacts against the 170 verified MuckRock requests without treating 30 unsupported historical positions as missing requests.
- 2.Build a manually reviewed entity registry from the full Livewire and evidence corpus before publishing person-specific profiles.
- 3.Record OCR, metadata, tagging, and accessibility status for all 425 PDFs and remediate priority documents.
- 4.Create synchronized captions and reviewed transcripts for the six local files and the highest-priority official-channel videos without inventing spoken content.
- 5.Replace duplicated search titles and descriptions with page-specific editorial metadata.
- 6.Complete manual keyboard, screen-reader, zoom, reflow, document, caption, privacy, factual, and legal review.
- 7.Monitor current Search Console and Bing reports without treating sitemap submission as proof of indexing.
Standing acceptance checks
These checks condense the high-leverage audit questions into release requirements. They are tests to satisfy, not claims that every item is already complete.
Completeness
Every published record has an identified source, stable URL, discoverable parent index, and documented status.
FOIA reconstruction
A request is not marked complete until its request, agency, dates, communications, attachments, disposition, and source basis are accounted for.
Named people
A person-specific profile requires source review, identity disambiguation, relevant context, careful attribution, and links back to the underlying records.
Claims and findings
Allegations, testimony, archive summaries, technical observations, and adjudicated findings must remain visibly distinct.
Authorship
Every article must identify a verified author or accurately identify the archive as maintainer when imported attribution is uncertain.
Search metadata
Indexable pages require a descriptive title, description, canonical URL, one clear H1, useful internal links, and matching visible structured-data content.
Accessibility
Automated checks are followed by manual keyboard, screen-reader, zoom, reflow, contrast, document, caption, and transcript review.
Evidence integrity
Published hashes identify bytes; they do not by themselves prove origin, truth, legal effect, or chain of custody.
Privacy and safety
Personal, medical, contact, witness, sealed, privileged, and other sensitive material receives human review before publication.
Corrections
Material changes preserve an intelligible correction record, date, reason, and right-of-reply path.
Machine readability
Sitemaps, manifests, HTML, and schema are validated as discovery aids without claiming control over third-party crawling or interpretation.
Release evidence
A deployment is complete only after the build, release checks, live status codes, canonical host, key pages, assets, sitemaps, and error routes are verified.