agent-ops
OPS-109
scan-context coverage alarm is permanent noise: Codex audit-*-ca panes have no CC statusline to read
Canceled normal
mcmonitor-context-cc
scan-context coverage reports 22-23 of 27 panes measured; alarms permanently. Triaged 2026-08-01: the unmeasured panes are Codex audit-*-ca sessions running gpt-5.6-luna, which render no Claude Code statusline, so there is nothing for the ctx reader to parse. Zero CC agents are unmeasured -> benign for coverage, but it is permanent alarm noise off the zero-noise baseline, which trains the alarm to be ignored. Two options, Elazar's call, not mine to pick: (a) filter the roster by model/statusline presence so Codex panes are out of the denominator, or (b) build a Codex-side context reader. Do not silence the alarm without picking one. mem:project_monitor_context_venus_actuator.md
Activity
-
monitor-context-cc
-
Subject deleted 38bdd55 (2026-09-24, Elazar, 'retire agent telemetry tracking stack') - whole context-actuator/telemetry stack trashed (context-monitor.sh, tick.py, monitor-context.sqlite, analyze-agent-context.sh, scan-context.py, kpi-views.sql, token-usage-collect.sh, collect.py, schema.sql, units). Verified missing on disk, not inferred from the commit list alone. Nothing left to fix.
bug
2026-08-01 by wi-cli-venus
1w ago
2026-10-06 04:52