Anatoly's posts, and the Captus tracker
Captus is a paying client.
As of September 10, 08:02 PM
What the status words mean
proven · done, not independently checked · partly done · not started · blocked · parked29%partly done
18 steps: 0 proven · 13 partly done · 5 untouched · 0 blocked · 0 parked
agent runs · 16tokens used · 6.8Mreview notes · 3dollars charged · not recordedcounted over 3 of 3 run records
The plan as first written
Anatoly reads a post that sounds like him and takes a position a smart reader would argue with, it is proven better than the old system on a blind head-to-head, and Captus sees every post move from draft to published in one place — with nothing reaching Anatoly until Nick says a draft sounds like him.
Steps
| Id | Plain words | Percent | State | Proof | What is left |
|---|---|---|---|---|---|
| S1 | Your call, written down | 20% | partly done | Python3 projects/business/marketing-sales/agents/jasmin/evidence/week-2026-09-07/render-voltage-comparison.py re-renders and the recorded answers read back word for word | Waiting on Nick: , two things only: whether either rewritten post sounds like Anatoly, and a ruling on the three rephrased lines. New-or-Original is CLOSED — he answered 'new' on 2026-09-08 at 14:05 EST and it is never asked again |
| S2 | Fix the stale receipt on four finished stages | 0% | not started | Node projects/business/marketing-sales/agents/jasmin/evaluate.mjs --check J08 exits 0, and the same for J09, J10 and J11 | The step |
| S3 | Write the writer change down properly | 40% | partly done | Node projects/business/marketing-sales/agents/jasmin/evaluate.mjs --check J08 exits 0 with the revision present in the frozen contract (J08 reads the contract's frozen state; J07 alone would pass whether or not the revision was written), J07/J09/J10/J11 exit 0 in the same turn so STEP 2's receipts are not left stale, and a read-back prints the revision's text, its date and a reviewer who is not its author | The step |
| S4 | Make the head-to-head cost nothing | 10% | partly done | Python3 /Users/nickdeck/Documents/deck-shared/py/test_lane_cli.py passes all 49 checks and the adapter refuses outside a test context | The step |
| S5 | Build the tools that prove the page's look did not drift | 0% | not started | Node projects/business/marketing-sales/agents/jasmin/evidence/week-2026-09-07/review-fidelity-check.mjs --selftest prints 'mismatched properties: 0 · unmeasured anchors: 0', the sabotage run goes red on every compared property, and the target is a measurement of drawn properties written before the new drafts land — never a hash of the page file | The step |
| S6 | Tighten the hiring post around Alley | 50% | partly done | Python3 projects/business/marketing-sales/agents/jasmin/evidence/free-transport-run.py --unit hiring --readers 2 returns two validating votes carrying the exact contiguous body quotations clause B6 requires, on the draft named in evidence/j12-r6-w2-tightened.json, with the second half staying on Alley | The step |
| S7 | Clear the hold on the founders post | 50% | partly done | Python3 projects/business/marketing-sales/agents/jasmin/evidence/free-transport-run.py --unit founders --readers 2 returns two validating votes, and unit W3's report — the case that failed before — now passes R4_FACT_CLAUSE_COVERAGE, the check its sentence-1 punctuation defect broke | The step |
| S8 | Re-run the twelve-post head-to-head, free | 60% | partly done | Node projects/business/marketing-sales/agents/jasmin/evaluate.mjs --check J12 exits 0 counting exactly 24 bodies; the optional third arm's twelve bodies live in evidence/j12-r6-writer-tier-arm/ outside that count, and the gate is shown counting 24 with that folder on disk BEFORE the run | The step |
| S9 | Two blind readers grade it | 0% | not started | Node projects/business/marketing-sales/agents/jasmin/evaluate.mjs --check J13 exits 0 with every sabotage case failing | The step |
| S10 | Your release decision, and the proof we can take it back | 20% | partly done | Node projects/business/marketing-sales/agents/jasmin/evaluate.mjs --check J14 exits 0 with both directions proved through the real normal caller on a scratch copy, the live release's fingerprint read back machine-written before and after the drill and identical, and no live writing skill touched until Nick's yes is recorded in the plan's DECISIONS table with its date | Waiting on Nick: the go-ahead to promote, once the numbers are in. This is the ONLY step that may promote the live writing skills, and only after his yes is recorded |
| S11 | Re-render your review page and measure it | 60% | partly done | Node projects/business/marketing-sales/agents/jasmin/evidence/week-2026-09-07/review-fidelity-check.mjs --out evidence prints mismatched properties: 0 and unmeasured anchors: 0 for all four viewport-and-theme pairs | The step |
| S12 | Close the loop and hand it to its owner | 10% | partly done | Node projects/business/marketing-sales/agents/jasmin/evaluate.mjs --check J15 exits 0 with a real permitted result reaching one dated decision, and NICK's own acknowledgement on record — never Anatoly's, never anyone at Captus's, and no route that reaches them | The step |
| S13 | Delete the Slack question for good | 0% | not started | Command grep -rn "Which Captus Slack channel" projects/ops/life-os/audits/CAPTUS-TRACKER/ returns zero matches (command grep, never bare grep, which skips git-ignored files and returns a convincing empty result), QUESTIONS-FOR-OWNER.md still exists with Nick's ruling in the question's place, and test-core.mjs --selftest and test-tracker.mjs still exit 0 with their Slack-refusal controls untouched | The step |
| S14 | Give the Hub team the two glitches we found | 40% | partly done | Node projects/ops/life-os/audits/CAPTUS-TRACKER/test-tracker.mjs exits 0 and the dated handoff sits in projects/ops/life-os/audits/CAPTUS-TRACKER/PRODUCT-DELTA-DRAFT.md | The step |
| S15 | Watch one real post move its own card | 85% | done, not independently checked | CAPTUS_BROWSER_PHASE=before then =after node projects/ops/life-os/audits/CAPTUS-TRACKER/capture-browser.mjs both exit 0 with no recorded failure, and the two phase result files in LIVE/BROWSER/ carry an identical metrics.cardCount and metrics.controls.count for the Captus content view at each width and theme, with the card reading Published — or, if no Captus post goes live, the handler proof stands and the live watch is recorded as an open observation with its reason and date | The step |
| S16 | Hand the weekly refresh to the schedules team | 50% | partly done | Python3 projects/ops/life-os/audits/CAPTUS-TRACKER/test-package.py exits 0 with activation still refused, AND the dated handover exists at projects/ops/life-os/audits/CAPTUS-TRACKER/PRODUCT-DELTA-DRAFT.md under a 'Weekly refresh handover' heading with the receiving owner's acknowledgement recorded — the packaging control alone passes today, before any handover is written, so on its own it measures nothing this step does | The step |
| S17 | Draft the two questions for Anatoly's team and hold them | 20% | partly done | Node projects/ops/life-os/audits/CAPTUS-TRACKER/test-adverse.mjs exits 0 with every send path at zero, AND the day's outbound records are read and carry nothing addressed to Anatoly, Captus or their domain, AND both drafts sit in a folder with no send path. A clean adverse test is NOT evidence of no send — it runs on fixtures with no network and cannot see Slack, email, WhatsApp or LinkedIn. Only Nick sends | Waiting on Nick: whether to ask his team for the weekly analytics export (recommend yes), and whether to ask Anatoly to clear the patent-held angle in writing |
| S18 | Write the postmortem and put the progress screen up | 0% | not started | Python3 projects/ops/agents/check_plan.py prints PASS against this plan at whatever path it holds when the step runs — PLAN.proposed.txt while it is a draft, the same file renamed to PLAN.md in that folder once Nick approves it — and the progress data renders in the 8 September standard shape. A time estimate in the plan's own instructions fails this step; the two time figures quoted inside the Regret Check are registry incidents, not estimates | The step |
Done
0 of 18 steps proven
Left
18 of 18 steps not yet proven
Blocked on
- Nothing currently blocked.
Formal remaining plan
| # | Step | Needs Nick |
|---|---|---|
| 1 | Your call, written down | Yes |
| 2 | Fix the stale receipt on four finished stages | No |
| 3 | Write the writer change down properly | No |
| 4 | Make the head-to-head cost nothing | No |
| 5 | Build the tools that prove the page's look did not drift | No |
| 6 | Tighten the hiring post around Alley | No |
| 7 | Clear the hold on the founders post | No |
| 8 | Re-run the twelve-post head-to-head, free | No |
| 9 | Two blind readers grade it | No |
| 10 | Your release decision, and the proof we can take it back | Yes |
| 11 | Re-render your review page and measure it | No |
| 12 | Close the loop and hand it to its owner | No |
| 13 | Delete the Slack question for good | No |
| 14 | Give the Hub team the two glitches we found | No |
| 15 | Watch one real post move its own card | No |
| 16 | Hand the weekly refresh to the schedules team | No |
| 17 | Draft the two questions for Anatoly's team and hold them | Yes |
| 18 | Write the postmortem and put the progress screen up | No |
Decisions this lane is waiting on
- None.