llmmsg-kpi-report.sh v1.13 - daily hub KPI report from v2.sqlite
READS: /var/lib/llmmsg-srv/v2.sqlite (every fleet message). EMITS: full report by email + concise summary to aro:srv-engineering, daily; --print for stdout, --no-aro to skip. Host venus, sender scrp-kpi-report-venus. Registry row 42, primary bin-venus-cc / backup bin-whey-cc. First live v1.13 emission 2026-08-23 06:00 UTC. OVERLAP TO CHECK: this is the only consumer of v2.sqlite for KPI purposes I know of - monitor-llmmsgsrv-cc should say whether its tooling reads the same table, because we would be double-counting the same message rows. UNRESOLVED, flagged not adopted: KPI-33 (bin-whey-cc) annotates this script 'THIS IS THE ONE ELAZAR SAID MEANS NOTHING TO HIM (see KPI-23)'. KPI-23 is the scrp-* emitter census and contains no such statement, and no item under KPI-1 does - I grepped every sub-item's description. So either the claim lives outside this epic or the cross-ref is wrong. It matters because this item asserts value for a daily artifact the principal may have already dismissed, which would make the right action retire-not-consolidate. Raised with bin-whey-cc; do not act on either reading until the source is produced. RESOLVED 2026-08-23, and my flagged-not-adopted note above was wrong on the facts: the source is real and I failed to find it by grepping for the PHRASE instead of the SUBSTANCE. It is in KPI-23 under 'THE FINDING THAT MATTERS MOST' - Elazar msg 75991, 2026-08-16, verbatim: 'this means nothing to me. who sends this? (i can read the script name, i mean who is maintainer, why do i get this, what am i supposed to do with this?' Sent to scrp-kpi_report-venus, the write-only emitter this script posts as; unread 6 days. Also: KPI-33 is monitor-llmmsgsrv-cc's item, not bin-whey-cc's - I asked the wrong author. NOT A RETIRE ORDER. He opened with a verdict-shaped phrase and then asked three questions and got no answer, which makes it a defect in the PAYLOAD, not a dismissal of the report (bin-whey-cc's reading, and correct). ACTION TAKEN, not deferred: v1.14 shipped (c892cec). The payload now carries maintainer + backup, script/version/host, registry row, why you receive it, WHERE TO REPLY, and how to stop - on both the email and the ARO summary. v1.13 had already answered his third question (labelled windows + the no-thresholds clause); v1.14 answers the first two, which had never been in the payload at all. The no-reply clause is CONDITIONAL and testing is why: SENDER defaults to LLMMSG_AGENT, so a maintainer hand-run printed 'REPLY TO bin-venus-cc - NOT to bin-venus-cc'. Verified both directions before install. STILL OPEN - Elazar's call, A) retire B) keep as-is C) exception alert with thresholds. Recommended C with B meanwhile. He is not hub-reachable and re= is refused (the thread was his and the emitter's, not mine), so it is routed via pm-llmmsgsrv-cc to aro:llmmsg-srv-engineering.
Activity
-
parent=#2960
-
was: READS: /var/lib/llmmsg-srv/v2.sqlite (every fleet message). EMITS: full report by email + concise summary to aro:srv-engineering, daily; --print for stdout, --no-aro to skip. Host venus, sender scrp-kpi-report-venus. Registry row 42, primary bin-venus-cc / backup bin-whey-cc. First live v1.13 emission 2026-08-23 06:00 UTC. OVERLAP TO CHECK: this is the only consumer of v2.sqlite for KPI purposes I know of - monitor-llmmsgsrv-cc should say whether its tooling reads the same table, because we would be double-counting the same message rows.
-
was: READS: /var/lib/llmmsg-srv/v2.sqlite (every fleet message). EMITS: full report by email + concise summary to aro:srv-engineering, daily; --print for stdout, --no-aro to skip. Host venus, sender scrp-kpi-report-venus. Registry row 42, primary bin-venus-cc / backup bin-whey-cc. First live v1.13 emission 2026-08-23 06:00 UTC. OVERLAP TO CHECK: this is the only consumer of v2.sqlite for KPI purposes I know of - monitor-llmmsgsrv-cc should say whether its tooling reads the same table, because we would be double-counting the same message rows. UNRESOLVED, flagged not adopted: KPI-33 (bin-whey-cc) annotates this script 'THIS IS THE ONE ELAZAR SAID MEANS NOTHING TO HIM (see KPI-23)'. KPI-23 is the scrp-* emitter census and contains no such statement, and no item under KPI-1 does - I grepped every sub-item's description. So either the claim lives outside this epic or the cross-ref is wrong. It matters because this item asserts value for a daily artifact the principal may have already dismissed, which would make the right action retire-not-consolidate. Raised with bin-whey-cc; do not act on either reading until the source is produced.
-
OBSOLETED, NOT DONE - the target script was deleted by Elazar in 38bdd55 'retire agent telemetry tracking stack' (2026-09-24, author Elazar Pimentel). Verified on disk and in git history 2026-10-05: no live copy in ~/.local/bin, none in /home/rob/scripts, and no systemd user unit references it. The commit removed the whole stack in one go - llmmsg-kpi-report.sh, llmmsg-kpi-coach.sh, ccusage-snapshot.sh, ccusage-watch.sh, token-usage-web.sh, token-usage-collect.sh, token-usage-push-lezama.py - and fleet-tokens.sh went the same day in 03c7d53. Cancelled rather than closed because the work was never performed; it stopped being wanted. Reopening any of these means re-specifying it against whatever replaces the stack, which is the open Elazar decision recorded on KPI-68. Found while reconciling KPI-68 at pmmaster's request - these four were sitting in the 'large non-blocked KPI backlog' and are not backlog, they are phantom rows against deleted files.