Cut Grok Bot Re-Reading Waste: Narrow Task Card + Save Summary + Change Check + Routine Audit

Export usage CSV → audit routines → add change checks → one job per chat with file-based state

Who

Grok Bot users burning through weekly quota on cache reads

You get

Reduce token waste from re-reading chat history using narrow task cards, file-based summaries, change checks, and routine audits

You need

Grok Bot

Steps

1. Export usage CSV and analyze re-read share

   Go to grok.com/dashboard, open Usage in the sidebar. Scroll past the Tokens per day chart. Click Export at the top right of Usage events to download a CSV. Every row contains your account email—do not post the raw file. Upload the CSV to your chief of staff Bot and paste the analysis prompt. Focus on the cache reads percentage; that is your re-read share.

   Execution prompt · Source text。Message to chief of staff Bot after uploading the CSV

   Copy prompt

   Check this usage file and analyze my bot behaviours and tell me how we should optimize for usage.

2. Audit existing routine frequency and value

   Paste the routine audit prompt into your chief of staff Bot to review all existing routines: check run frequency, useful output rate, whether event triggers could replace schedules, and whether a quick check could decide if the full job needs to run. Classify each as KEEP / SLOW DOWN / EVENT TRIGGER / CHANGE CHECK / RETIRE, ranked by potential usage savings. Re-run whenever your routine list grows.

   Execution prompt · Source text。Chat input for chief of staff Bot

   Copy prompt

   Review all my existing routines. For each one, check how often it runs, how many runs actually produce something useful, whether an event trigger could replace the schedule, and whether a quick check could decide if the full job needs to run. Classify each routine as KEEP, SLOW DOWN, EVENT TRIGGER, CHANGE CHECK or RETIRE. Rank them by potential usage savings. Estimate usage before and after, accounting for any additional checks. Show me what you can verify versus what you're estimating. Don't make any changes until I approve.

3. Add change checks in front of fast routines

   For jobs that need fast polling but have no event to hook, split the routine in two: step one reads a single state value (hash, count, timestamp) and compares it to the last value saved in routines/state/[name].txt. If it matches, log 'no change' and stop—do not open anything else. Only run the full job when something changed. The 'don't browse around' constraint is critical; without it the Bot tends to wander once the page is open, turning the check back into a full run.

   Execution prompt · Source text。First step of the routine (change-check node/message)

   Copy prompt

   Before doing any work, read only [the status field / latest commit / item count / last-updated date] and compare it to the value saved in routines/state/[name].txt. If it matches, log no change and stop. Don't open anything else. Don't browse around. If it changed, save the new value, then do the job.

4. One job per chat: narrow task card start and file-based summary

   Begin each new task with the narrow task card template as the first message, specifying task, location, required context, completion criteria, and stop conditions to prevent the Bot from wandering into extra research. When the job is done, have the Bot save a summary to a file (max 10 lines) covering what was done, decisions made, file names, and next steps. Start the next job from the file, not from chat history. If a chat has grown too long, use Duplicate to reset (keeps profile/settings/skills/routines, drops conversation history and attachments), or convert repeat jobs into skills for short-chat invocation.

   Execution prompt · Source text。First message of a new task (narrow task card)

   Copy prompt

   Task: [what to do, in one line] Where: [which app or file] Use: [the context it needs, nothing more] Done when: [the exact output] Stop after that. Don't research anything extra. If you get stuck twice, stop and ask me.

   Execution prompt · Source text。Message after task completion (save summary)

   Copy prompt

   Save a short summary of this job to a file: what we did, the decisions, the file names, and the next step. Max 10 lines. Then tell me the file name.

No prompt needed — just follow the steps.