Make agent time detail match reality

Confirm the native source before deriving intervals

First check whether the documented agent_stats_export already supplies the required totals for the exact agent, campaign grouping, timezone and report window. Use the installed Non-Agent API reference and the protected API setup; this fixture does not certify that export’s internal formula.

vicidial_agent_log contains duration/epoch fields; sorting event_time or treating sub_status as a generic READY/PAUSED state stream is not a verified adapter. A report owner must map the installed *_epoch/*_sec semantics and the actual login/session source before a native calculation.

Ask for one minimized synthetic row plus its verified session/login boundaries. Required adapter output is session identity, state, interval start,end and source-row reference. If the owner cannot establish that mapping, stop at the offline fixture and hand off those missing fields; do not invent logout or a preceding boundary state.

SHOW COLUMNS FROM vicidial_agent_logWHERE Field IN ('event_time','pause_epoch','pause_sec','wait_epoch','wait_sec','talk_epoch','talk_sec','dispo_epoch','dispo_sec','dead_epoch','dead_sec','sub_status','pause_type');

You should seeUse documented export totals or a reviewed native adapter. The next exercise is explicitly fictional.

Prepare a protected API read →

Prepare the read-only client →

Treat state as intervals, not labels

Fast answer: use the following clipping model to investigate a report mismatch. The available documentation does not establish that agent_stats_export implements this exact formula internally. A state change has a start but often no explicit end; derive its end from the next event, then intersect (clip) that interval with the agent's actual login window and report boundary.

An agent who never logs out cleanly — a crashed browser tab, a lost network connection — leaves a final state change with no next event to derive an end from at all. Treat that last open interval as ending at either the report boundary or a documented forced-logout time, whichever is earlier and actually knowable, and say plainly in the report which one you used; silently picking one without disclosing it is how two people reconciling the same day arrive at two different totals.

Rows outside the window must not become visible simply because their state label matches the current filter.

See VICIdial terminology for complete beginners: users, phones, campaigns and leads for the rest of this glossary.

visible_start = max(event_start, login_start, report_start)visible_end   = min(next_event_start, logout_time, report_end)visible_secs  = max(0, visible_end - visible_start)

Run the complete fictional duration-row fixture

This fixture assumes the adapter has already established start_epoch plus duration_seconds for each interval and one UTC login session. It does not read a database. The fictional raw row has PAUSED09:00–09:12, WAIT09:12–09:20 and TALK09:20–09:40; login09:05–09:30 clips the first and last.

The same three item identities must produce420,480 and600 seconds. The partial PAUSED interval is retained even though its raw start predates login. An actual report needs the reviewed adapter above, including any event that begins before report_start.

python3 - <<'PYMODEL'from datetime import datetime,timezoneT=lambda text:int(datetime.fromisoformat("2026-11-07T"+text+"+00:00").timestamp())# Fictional duration-row adapter, not a captured database export.raw=[("PAUSED",T("09:00:00"),720),("WAIT",T("09:12:00"),480),("TALK",T("09:20:00"),1200)]login,logout,report_start,report_end=map(T,["09:05:00","09:30:00","09:00:00","10:00:00"])visible=[]for state,start,duration in raw:    end=start+duration    left=max(start,login,report_start); right=min(end,logout,report_end)    seconds=max(0,right-left); visible.append(seconds)    print(state,seconds,"seconds",("partly clipped" if seconds and seconds<duration else "whole or excluded"))assert visible==[420,480,600] and sum(visible)==logout-login==1500print("PASS total 1500 seconds = 25 minutes; partial overlap retained")print("Fictional UTC interval fixture only; no native report calculation certified.")PYMODEL

View a still screenshot

Observed local fictional model

Open to enlarge

You should seeYou have reproduced partial-overlap clipping and can explain what the native adapter still must establish.

Exclude intervals by overlap, not by their starting label

Keep the raw interval unchanged and derive the visible overlap separately. A raw start before login/report start does not make the whole row invisible: retain the part inside the verified bounds.

Classify a fully excluded row by its end/start relative to the boundaries. Policy-hidden time is a separate permission/report rule, not a timing shortcut.

You should seeThe exclusion rule agrees with the 420-second partial interval and the 1,500-second total.

• Fully before: event_end is at or before max(login_start,report_start). Visible duration is0.
• Fully after: event_start is at or after min(logout_time,report_end). Visible duration is0.
• Partial overlap: visible_end exceeds visible_start, even if the raw start is earlier or raw end later. Retain only the positive clipped part.
• Hidden by policy: explain the independent access rule; preserve raw audit evidence and do not count it as a clipping error.

Reconcile daily totals

For a controlled agent day, sum clipped intervals and compare them with login duration, pause detail and call timestamps. Gaps and overlaps should be explainable.

Test midnight boundaries and an agent with multiple login sessions. Those cases reveal most interval mistakes quickly.

Allow a small, named rounding tolerance rather than demanding an exact match to the second — first identify the timestamp precision and rounding in both calculations; a difference can also indicate a real mismatch. State the tolerance you are using explicitly, though: an undisclosed tolerance is indistinguishable from a report that simply does not reconcile.

Stop on an unexplained second and reverse the query change

Do not release a revised report while synthetic interval totals exceed the login window, cross session boundaries or differ between repeated runs at the same as-of time. Record the smallest failing fixture and preserve the old query/version.

Rollback means restoring the prior report calculation or view without altering event history. Fix the interval derivation in a test copy, repeat midnight and multi-login cases, then ask an independent reviewer to reproduce the arithmetic.

• Old and proposed query definitions remain versioned and reversible.
• Synthetic normal, midnight and multiple-login sessions reconcile independently.
• Visible and hidden durations remain separate from raw audit evidence.