Prompt AI agents to self-correct using context checks and a guided decision loop.
npx skills add https://github.com/keshav55/writing_team --skill meta-keshav55
Please help me install this Agent Skill. Skill: meta Source: https://github.com/keshav55/writing_team/tree/main/atris/skills/meta Command: npx skills add https://github.com/keshav55/writing_team --skill meta-keshav55
The metacognition skill helps AI agents improve reliability by prompting self-awareness and self-correction before and during tasks.
Activate the meta skill at the start of a task by saying 'activate meta' or 'load meta'.
High-intent search queries and answers about installing and using this skill.
To add self-correction to an AI workflow, use a metacognition skill that enforces context checks and a guided decision loop to prompt self-awareness during tasks and prevent repeating errors.
The best way to improve AI reliability during complex tasks is applying a guided decision loop that enforces orientation, decision, action, checking, and learning steps to keep the agent focused and self-correct.
To activate metacognition for AI self-awareness, start your session by saying 'activate meta' or 'load meta' to enforce planning and context checks before the agent acts.
Apply a self-check loop at session start, during complex tasks, when output feels generic or incorrect, or after mistakes to trigger safe escalation and improve results.
Workflow planning with metacognition requires no specific dependencies, allowing you to integrate self-awareness and self-correction loops into your existing AI agent environment immediately.
An AI agent produces generic output when it lacks self-awareness; applying a self-check loop enforces context evaluation and safe escalation to correct the output and improve reliability.
GainInsightDevCreate a current-task.md file to initiate a work session.
maxorericLog execution details, decision rationales, and lessons learned during operational processes.
GarethManningGenerate calibrated metacognitive prompts for planning, monitoring, and evaluating tasks.
Snowflake-LabsA concise summary of your roleplay character: name,</name> personality, and/or behavior, formatted as a single string sentence. Typically不超过50字，此处省略
