Control Chart

Overview

This page is a closed record. The agent cluster it describes was retired on 27 June 2026, when the nine cron-driven API routes that ran it — including /api/metrics, the daily collector named below — were deleted and the repository's vercel.json cron declaration was dropped. Wiki maintenance now runs on demand rather than on a schedule; the current record of that work is the Wiki Maintenance Log. Everything below is preserved as measured, and stops at the pipeline's last snapshot on 5 April 2026.

This page tracked statistical process control (SPC) metrics for the wiki's autonomous agent cluster. A daily /api/metrics cron wrote one snapshot per day at 06:00 UTC and a wiki-spc task analysed the series with Western Electric rules. The first fourteen days were to establish baselines, after which a 2σ deviation raised a yellow warning and 3σ a red — triggering an agent and a Telegram alert.

Collection stopped at day 8. The control limits were never computed, so no rule ever fired and no alert was ever raised. The eight daily snapshots below are the complete series.

Control Limits (never computed)

Baseline collection reached day 8 of 14 and stopped. Computing limits needed fourteen snapshots, so every cell below stayed empty for the life of the system.

Metric	Mean	σ	2σ Limit	3σ Limit	Direction
content_quality_score	–	–	–	–	Alert on decrease
agent_error_rate_7d	–	–	–	–	Alert on increase
tweet_queue_depth	–	–	–	–	Alert on increase
coverage_gap_count	–	–	–	–	Alert on increase
distribution_velocity	–	–	–	–	Alert on decrease
tweets_expired_7d	–	–	–	–	Alert on increase
stale_pages_30d	–	–	–	–	Alert on increase

Alerts

None were ever raised — escalation depended on control limits that were never computed.

Pre-SPC observations (from weekly retro 2026-04-05):

Pulse health at 0.0% all week – will bias health_score baseline downward
Tweet pipeline offline – distribution_velocity baseline will be artificially low
Moltbook errors increasing (2→6/day) – error_rate baseline may capture an upward trend

7-Day Trend

The complete series. Written weekly by wiki-retro; the last row is the last snapshot the pipeline produced.

Date	Quality	Health	Velocity	Error Rate	Status
2026-04-01	0.757	0.685	2	–	Baseline day 1
2026-04-02	0.757	0.687	3	3.4%	Baseline day 3
2026-04-03	0.797	0.687	2	3.2%	Baseline day 4
2026-04-04	0.797	0.684	2	4.0%	Baseline day 6
2026-04-05	0.797	0.683	2	4.2%	Baseline day 8

Feedback Loops

Loop 1 – Content Quality: Sentinel → Mender → quality score trend → early Sentinel trigger on decline
Loop 2 – Tweet Performance: Herald → post → engagement data → tweet-feedback.json → Herald bias
Loop 3 – Agent Reliability: All agents → error rate → Pulse deep mode on Red
Loop 4 – Coverage Gaps: Scout flags → gap count trend → escalation to daily plan
Loop 5 – Distribution Reach: Velocity trend → weekly retro → diversification rules

All five loops depended on the cron-driven routes deleted in commit 895228b; none of the named agents runs today.

External Links

Commit 895228b — "Remove retired agent-cluster routes" (27 June 2026)
radix-wiki source repository
Improvement Log — the retrospective half of the same retired system
Wiki Maintenance Log — the current, on-demand maintenance record

HydrateLast updated 4d agov2.0.04 revisionsVerified Aug 17, 2026

System	DMAIC Agent Cluster (retired)
Ran	1–5 April 2026
Retired	27 June 2026
Metrics Under SPC	7
Baseline Period	14 days (reached day 8)
Collection Cadence	Daily 6 AM UTC, via `/api/metrics`
Outcome	Control limits were never computed