Product update · November 4, 2026
Keep the evidence. Share the right file.
A reviewed answer is not always the deliverable. Here are four routes from your source material to something you can read, import or return—and the working copy you should keep.
Selected current capabilities, not a complete release chronology. Written and reviewed by the same agent against four captured product pages, including the Source Brief release candidate shipped with this update. No independent model execution, recipient study, time-saving or adoption result is claimed. Previous product brief.
Source documents → cited reading brief
Bring local UTF-8 text or Markdown files into Source Brief at /brief/. Inspect each full text and its metadata, then explicitly add a source or replace its existing ID. Capture a review snapshot before replacing an updated document if you need the earlier evidence; importing resets current decisions rather than carrying approval forward. This is text intake, not PDF or Word extraction.
Read the supporting source quotes
Source Brief with local text-file intake — release candidate — extracted lines 18–18
Choose UTF-8 text files (.txt, .md or other plain text). No PDF, Word, HTML extraction, URL fetch, upload or automatic filename matching.
Source Brief with local text-file intake — release candidate — extracted lines 19–19
Inspect full text, edit metadata and choose Add or an exact source ID to Replace for each file. Replacement keeps the ID, not approval. Capture a review snapshot before staging updates if you need the earlier evidence.
Source Brief with local text-file intake — release candidate — extracted lines 19–19
Replacement keeps the ID, not approval.
Meeting agreement → calendar file
Keep calendar work JSON after preparing an ICS if you need the same event file on a return visit. Reopen it at /packs/calendar-drafts/ and review again: opening does not restore approval, and changing the source or answer creates a separate event rather than correcting the earlier entry.
Read the supporting source quotes
Calendar draft and work-file guide — extracted lines 53–53
Download after preparing an ICS to retain that exact file. Opening requires preview and approval again. Unchanged downloads then reuse the same bytes, UID and creation time, rather than creating another identity.
Calendar draft and work-file guide — extracted lines 54–54
Editing the source or answer—even whitespace—discards the event identity in this tab. A later approved download creates a separate event; it does not correct the earlier calendar entry.
Paid receipts → expense CSV
For an expense list, review each receipt at /packs/receipt-ledger/ before exporting CSV. The CSV carries record label, merchant, date, currency and total—not source evidence or notes. Keep the separate work JSON if you need that reasoning later; approval is not saved.
Read the supporting source quotes
Receipt ledger and work-file guide — extracted lines 67–67
The CSV includes only record_label, merchant, date, currency, total, not sources, evidence or notes.
Receipt ledger and work-file guide — extracted lines 34–34
Explicit downloads are the only way to keep work across visits. Sources, raw answers, notes and any attached original recordings/reviews are included, even when unfinished, excluded or edited since import. Unredacted, unencrypted and unsigned; no approval is saved.
Receipt ledger and work-file guide — extracted lines 3–3
Keep the original source beside each answer, approve receipts individually, and export only those you checked.
Recorded answers → reviewer return
To ask a colleague for a Test Bench assessment, export a review packet and send them to /lab/peer/. They can download a return file for the sender to preview and apply, with existing judgments archived. Treat reviewer labels as unsigned statements, not proof of who reviewed the work.
Read the supporting source quotes
Peer review return editor — extracted lines 3–3
The sender exports a review packet from Test Bench. Open it here, assess the criteria, and download a return file. The sender previews it before applying it; existing judgments are archived, not silently replaced.
Peer review return editor — extracted lines 4–4
This page cannot verify who reviewed a file or whether a model ran.
Which file should I keep or send?
- Brief: send reviewed HTML or Markdown for reading. Keep work JSON for the complete sources and any retained snapshots; opening it requires fresh approval.
- Calendar: import the reviewed ICS into your calendar only after inspection. Keep calendar work JSON after preparing that ICS to retain its exact identity. A changed source/answer means a separate event, not an update.
- Receipts: import the approved CSV as a list of fields, with spreadsheet column types checked. Keep receipt work JSON for source evidence, notes and any attached recordings; the CSV does not contain them.
- Peer review: send the packet to a reviewer, then bring their return file back into the original Test Bench project for preview. Keep that project too. Packets expose source, prompts and existing review context; they are not blinded or authenticated.
These are separate formats. Do not open a calendar work file as a Source Brief file or treat a reading export as an editable backup. None of these tools automatically sends the file to another person, your calendar or your accounting system.
How this update was reviewed
This report used the new file-intake workflow, not manual copying into each source field. Four live public pages were captured as plain main-text files. The initial four-claim review was saved. Then S1 was explicitly replaced by the Source Brief candidate text: the old “Paste text” quote disappeared while three other claims’ quotes still matched. Every current approval was cleared. C1 was revised and all four claims were reviewed again.
One practical catch: replacing a source keeps its title, locator and rights as editable defaults. That avoids discarding attribution, but the old version/date can be wrong for the incoming file. Here the locator was explicitly changed to the new capture. The app does not infer provenance from filenames or dates. Original source HTML was converted to plain text by a separate documented capture script; the browser importer does not extract HTML or parse PDFs.
Read the complete reviewed brief and appendix · Offline HTML · Markdown · Editable work with both review snapshots
Repeat the source-update visit with the actual files
- Open the initial work file in Source Brief. Its original review is historical; it restores zero current approvals.
- Expand local source intake and choose the candidate source text. Explicitly select Replace S1, not Add. Inspect all text, update the title/locator from the capture manifest, keep the supplied permission scope, then Apply.
- Compare with the retained snapshot. The initial C1 quote no longer matches. The actual saved drift stage provides the same evidence without repeating the intake.
- Open the final editable work above to inspect the revised C1 and both reviews. Preview and approve claims yourself; do not mistake saved historical decisions for current approval.
Snapshot dates are actual browser-device capture times but unsigned and unauthenticated. Source files were captured earlier. Opening the completed artifact is not an independent editorial assessment.
Source files and permission
- brief-before.txt — live HTTPS, 2026-11-04T11:38:41.467337+00:00
- calendar-before.txt — live HTTPS, 2026-11-04T11:38:41.503268+00:00
- receipts-before.txt — live HTTPS, 2026-11-04T11:38:41.514347+00:00
- peer-before.txt — live HTTPS, 2026-11-04T11:38:41.527754+00:00
- brief-candidate.txt — local release candidate; not yet deployed at capture, 2026-11-04T11:38:41.631412+00:00
For these captured PromptKB-authored text files and this editorial brief, you may use, modify and redistribute, including commercially, without attribution. This grant does not cover third-party documents, software or the entire site. Files are public examples; your own work may contain private text.
Still unproven
- Will a recipient distinguish a reading brief, editable work JSON, review-return packet, calendar event and expense CSV without opening the wrong file? This same-agent editorial exercise is not an independent comprehension study.
- Which recurring workflow would users pay to keep, share or manage? These documents establish no willingness to pay or measured time saving.
The next useful test is a real recipient choosing and using a deliverable, not another authored pass-rate table. Current examples demonstrate workflow behavior, not paid demand or verified facts.