AI Daily Self-Review System
Paste the following prompt into your AI chat to install this skill:
Please follow https://skillhub.cn/install/skillhub.md to install @user_fe828a0e/self-reivew-pro.
About this skill
Problem
AI agents accumulate a lot of daily conversation, task execution, and error traces, but many failures do not naturally compound into learning. A stuck command, an unvalidated assumption, or a recurring “something felt off” moment often gets patched in the moment and then reappears. Self Review Pro treats daily runtime material as an auditable review record: it surfaces blind spots, recognizes repeated patterns, and compresses the outcome into executable rules.
How it works
- Trigger: It can run automatically at
22:00daily, or manually when the user says “review” or “self review.” - Data collection: It gathers today's conversation history, task logs,
logs, stuck-execution records, and yesterday's review frommemory/YYYY-MM-DD.mdandMEMORY.md. - Review template: It follows nine modules, including the day's sharpest pain point, challenged beliefs, thinking blind spots, unvalidated assumptions, pattern recognition, three-level root-cause questioning, cognitive upgrade,
if-thenaction anchors, and a self-audit checklist. - Memory writes: Core conclusions go into long-term
MEMORY.md, the full process goes intomemory/YYYY-MM-DD.md, and an external summary can be reduced to one or two sentences.
Boundaries
It fits agent workflows that already have conversation, task, log, and memory artifacts, especially when debugging, tool calls, or process decisions repeat. Quality depends on inputs: without history, execution logs, or searchable conversations, the template still runs, but pattern recognition and root-cause analysis become weaker. It is not a replacement for engineering monitoring, audit systems, or human incident review; it works best as a lightweight daily calibration loop.
Editor's note
Written for AI Agent Hub. This is separate from the publisher description above.
What problem it solves
After a day of agent conversations you want a structured review: what failed, what repeated, what you are about to automate by accident. This card, canonical name misspelled "reivew," describes a multi-part review of logs. The problem is reflection over records, not a smarter model.
Who it is for
Someone who already keeps task notes or chat exports and wants them forced into a template once a day. You will skim the template. You will not grade yourself with it.
Who should skip it
Skip it on day one of an empty editor. The first week earns a loop before it adds a skill that watches the loop. Skip prompts that hide tool calls or tell the agent to ignore other rules. Those show up on the AI agent shelf and they are a reason to close the card. Skip it if the "log" would include other people's private chats. That is the WeChat problem again.
First three steps
- Save a half-page note of what you actually did today, including one mistake.
- Run the skill on that note only. Ask it to quote the mistake rather than praise the day.
- If the review mentions work that is not in the note, the skill is confabulating. Remove it. If it quotes the mistake and stops, you can try it on a real log tomorrow.
Alternatives on this site
A project markdown file is the alternative when the day had one decision worth keeping. A memory plugin is for Harness sessions, and it will also store nonsense if you let it. Prefer the file until the file is not enough.
Questions
Does a nine-part template make the review true? No. It makes it long. Truth is whether each part cites the log.
Should this skill manage my other skills? No. One meta-prompt that claims priority over the rest is how you stop being able to tell which instructions fired.
Use Cases
- Automatically collect daily conversations, task logs, and error records at 22:00 to complete a nine-module review.
- Manually trigger a self-review after repeated task failures to dig into root causes and extract if-then action anchors.
- Compress core review conclusions into MEMORY.md and archive the full process into memory/YYYY-MM-DD.md for future reference.
- Combine yesterday's review and today's error logs to identify recurring blind spots, repeated patterns, and unvalidated assumptions.
Best For
- AI app developers who need to equip their AI agents with daily reflection and experience accumulation capabilities.
- Prompt engineers aiming to transform agent logs, conversations, and error records into actionable improvement rules.
- Architects focused on AI agent memory management, seeking to standardize writing workflows for MEMORY.md and short-term memory.
- Independent developers debugging multi-turn dialogue agents who need to locate thinking blind spots and recurring failure patterns.
Related Skills
A portable multi-agent framework with 18 specialist agents, three workflows, and tiered quality gates from requirements through deployment.
Call the Xiaoba AI interview Open API from your agent workflow to manage plans, candidates, sessions, and retrieve scored results.
File-based cross-session memory (MEMORY.md, daily logs, USER/SOUL) so agents reload context instead of relying on in-chat recall.
When instructions are vague but judgment is expert, use permission-based probing and structured anchors to calibrate tacit criteria and record a reusable comprehension model.