Routine Budget Checkup: Pause, Audit, and Prioritize Before You Burn Through Quota

Input Bot/Routine list and budget tier → Apply a 7-step pause-inventory-prioritize-restore workflow → Output KEEP/SLOW/EVENT/CHANGE/RETIRE classification table and recovery sequence

Who

People running multiple Routines across several Bots, often hitting usage limits, or waking up to expired approvals

You get

Stop quota bleed by pausing non-critical Routines, migrating heavy radar jobs, resetting bloated sessions, scheduling weekday windows, classifying with a Chief Bot prompt, re-enabling one by one, and doing weekly reviews

You need

Steps

1. Emergency stop: pause all non-critical Routines

   Go to your Bot management dashboard and locate the Routine list. Pause all non-critical Routines, especially those running 24/7, firing every 30 minutes, or proactively suggesting tasks. Stop the bleeding first, then investigate causes.

2. Relocate: move high-frequency radar out of the chat main Bot

   Migrate high-frequency radar or summary Routines from the 'chat main Bot' to a dedicated lightweight Bot. Reserve the chat main Bot for human review and posting gates only, reducing context buildup and overlapping tool calls.

3. Clear context: reset bloated sessions and add silence rules

   For obviously bloated sessions, start a fresh clean session. Also revise the prompt of high-frequency Routines to include 'point to skill/file, stay silent if nothing new'—prohibit long reports when there is no incremental update.

4. Schedule: limit to weekday windows and reset-day strategy

   Set Routine run windows, for example weekdays only during local hours 9–19. Before quota reset day, proactively pause Routines that can wait, avoiding unnecessary consumption right before the reset.

5. Classify: use the Chief Bot prompt to tag each Routine

   Paste your Bot/Routine list (name, frequency, trigger method, whether on chat main Bot) to the Chief Bot and run the budget checkup prompt. For each Routine, receive a KEEP / SLOW / EVENT / CHANGE / RETIRE tag, reasoning, estimated relative burn, and specific adjustment advice. **This step produces recommendations only; do not change configurations until human confirmation.**

   Execution prompt · Source text。Chief Bot chat input; replace {{产品名}} with your actual product name before sending

   Copy prompt

   你是 {{产品名}} 的首席助理，本轮只做 Routine 额度体检，不要改任何配置、不要发帖、不要花钱。
   
   输入：我粘贴的 Bot/Routine 清单（名称、频率、触发方式、是否在聊天 Bot 上）。
   
   对每条 Routine 输出一行表格：
   - 名称
   - 建议标签：KEEP / SLOW / EVENT / CHANGE / RETIRE
   - 理由（一句）
   - 预估相对烧量：高 / 中 / 低（不要编造具体美元或百分比）
   - 若 SLOW：建议频率或时段
   - 若 EVENT：改成什么事件触发更合适
   - 若 CHANGE：改哪一类触发/窗口
   - 若 RETIRE：可否归档
   
   规则：
   1. 无可靠信息就标 [需人工]，不要猜。
   2. 24/7、半小时级、主动建议、长上下文聊天 Bot 上的雷达，默认倾向 SLOW 或 RETIRE。
   3. 只读第三方市场技能可以提一句「可选」，不要当成必装。
   4. 最后给 5 步止血顺序（先停什么、再开什么），用我自己的话，短句。

6. Re-enable one by one: follow the classification order and observe

   Re-enable Routines in the order KEEP → SLOW → EVENT. After enabling each one, observe one cycle of execution and confirm no anomalies before enabling the next. For CHANGE items, modify trigger conditions first; for RETIRE items, delete or archive without re-enabling.

7. Weekly review: run a standing audit and write results back

   Run a standing 'budget checkup' review weekly (can link to `weekly-review` playbook). Write red/yellow/green status results back to your task board for continuous iteration.

你是 {{产品名}} 的首席助理，本轮只做 Routine 额度体检，不要改任何配置、不要发帖、不要花钱。

输入：我粘贴的 Bot/Routine 清单（名称、频率、触发方式、是否在聊天 Bot 上）。

对每条 Routine 输出一行表格：
- 名称
- 建议标签：KEEP / SLOW / EVENT / CHANGE / RETIRE
- 理由（一句）
- 预估相对烧量：高 / 中 / 低（不要编造具体美元或百分比）
- 若 SLOW：建议频率或时段
- 若 EVENT：改成什么事件触发更合适
- 若 CHANGE：改哪一类触发/窗口
- 若 RETIRE：可否归档

规则：
1. 无可靠信息就标 [需人工]，不要猜。
2. 24/7、半小时级、主动建议、长上下文聊天 Bot 上的雷达，默认倾向 SLOW 或 RETIRE。
3. 只读第三方市场技能可以提一句「可选」，不要当成必装。
4. 最后给 5 步止血顺序（先停什么、再开什么），用我自己的话，短句。