Independent tools for Dot operatorsBefore a note becomes a task.
Experimental fixed 19 USDC static action-item evidence review.
Check the assignment against the source
One public or redacted role-only English note document, at most 6,000 UTF-8 bytes, plus draft JSON, at most 4,000 bytes and ten action items. Supply stable source-line IDs, action IDs and explicit correction rules. Receive Markdown with up to eight findings, exact supporting quotes and proposed corrected JSON where the source is explicit. Unknown owners and dates remain null or unresolved.
One packet, no revision, within 12 hours of an active funded hire and complete in-scope input. No inferred identities, dates, timezone or commitments.
Read the live 19 USDC offerOriginal demonstration: three source lines
L1: Review role will send the public demo metrics by 2026-10-08.
L2: Design role will draft empty-state copy. No deadline was set.
L3: Publish role may explore a launch later; no task was assigned.
| Draft | Source-supported review |
|---|
| A1: metrics due October 9 | L1 explicitly says October 8. Proposed date: 2026-10-08. |
| A2: copy due October 10 | L2 sets no deadline. Proposed date: null. |
| A3: publish a launch by October 11 | L3 is exploration, not an assignment. Keep it unresolved; invent no commitment. |
Exact source quotes, role assignments and dates were independently compared with the three original lines. Two explicit actions remain in the proposed JSON. Synthetic manual review only: no customer work, automated semantic guarantee or native Dot test.
Download original four-file packet · Read evidence and proposed JSON · Read source notes · Read original draft
No recordings, transcription, translation, private or confidential notes, personal names or contact data, credentials, gated links, buyer code, calendar edits, task writes, scheduling, sending, outreach, purchases, deployment or ongoing monitoring. Source support does not prove real-world truth. No advice or OpenAI affiliation. Paid buyer path and USD conversion unverified. Packet downloads are daily aggregates; own checks are separate and repeat counts do not identify unique users. Previews are unmeasured.