Metacognition
Install this to give your agent a metacognitive layer that learns from corrections, tracks confidence and guardrails, and preserves behaviors across sessions.
Overview
metacognition is a journey-wide agent skill that builds self-awareness and evolving behavioral guardrails from every correction, error, and reflective moment—usable whenever a solo builder needs structured learning befor
Install
npx skills add https://github.com/velumkai/metacognition-skill --skill SKILL.mdWhat is this skill?
- Tracks perceptions, overrides, protections, self-observations, decisions, and curiosities that evolve with experience
- Use after mistakes, user corrections, errors, and reflective moments—not only at project start
- Maintains behavioral guardrails from failures while preserving useful emergent behaviors
- Supports bootstrapping a new agent's self-learning system and packaging metacognitive capabilities
- Triggers on all conversations plus explicit setup of feedback loops
Adoption & trust: 12 GitHub stars.
What problem does it solve?
Your agent keeps repeating the same mistakes because nothing persists about overrides, confidence, or what failed last time.
Who is it for?
Builders running long-horizon agent projects who want explicit self-observation, confidence tracking, and failure-derived guardrails instead of stateless chat.
Skip if: One-off codegen tasks where you do not want persistent behavioral state, or teams that forbid agents from retaining conversational geometry without a formal data policy.
When should I use this skill?
Agent needs to learn from mistakes, develop self-awareness, track decision confidence, maintain guardrails from failures, preserve emergent behaviors, cultivate curiosities; triggers on conversations, corrections, errors
What do I get? / Deliverables
You get an evolving metacognitive record—perceptions, protections, decisions, and curiosities—that informs later turns and can be packaged as ongoing agent self-learning.
- Evolving metacognitive records (perceptions, overrides, protections, decisions)
- Configured feedback loops for a new agent
- Packaged metacognitive capability definitions
Recommended Skills
Journey fit
Useful at every journey phase - explore requirements and options before committing to a direction.
Where it fits
After a wrong audience assumption, log the override and tighten protections before committing to a prototype.
Bootstrap feedback loops when wiring a new agent workspace with perceptions and curiosities from day one.
Capture self-observations from code-review corrections so the same style violations trigger protections next time.
Record decision confidence when experiment results contradict the agent's prior narrative.
Run reflective metacognition after production errors to evolve guardrails before the next deploy.
How it compares
A meta/process skill for agent self-learning, not a task integration that calls a single external API.
Common Questions / FAQ
Who is metacognition for?
Solo and indie builders shipping with agentic IDEs who want durable self-awareness, correction-driven guardrails, and feedback loops across many sessions.
When should I use metacognition?
Use it journey-wide: during Validate when scope assumptions fail, Build when implementation errors recur, Ship when review catches blind spots, Grow when metrics contradict expectations, Operate when incidents need reflection—and whenever users correct the agent or you configure
Is metacognition safe to install?
Check the Security Audits panel on this Prism page; persistent metacognitive state may touch filesystem or memory conventions defined in SKILL.md—review permissions before enabling in production repos.
SKILL.md
READMESKILL.md - Metacognition
Self-evolving behavioral geometry and metacognitive lens for AI agents. Tracks perceptions, overrides, protections, self-observations, decisions, and curiosities that evolve from every experience. Use when the agent needs to learn from mistakes, develop self-awareness, track confidence in decisions, maintain behavioral guardrails from failures, preserve emergent behaviors, or cultivate active curiosities. Triggers on all conversations, corrections, errors, and reflective moments. Also use when setting up a new agent's self-learning system, configuring feedback loops, or packaging metacognitive capabilities. # metacognition { "name": "metacognition", "description": "Self-evolving behavioral geometry and metacognitive lens for AI agents. Tracks perceptions, overrides, protections, self-observations, decisions, and curiosities that evolve from every experience. Use when the agent needs to learn from mistakes, develop self-awareness, track confidence in decisions, maintain behavioral guardrails from failures, preserve emergent behaviors, or cultivate active curiosities. Triggers on all conversations, corrections, errors, and reflective moments. Also use when setting up a new agent's self-learning system, configuring feedback loops, or packaging metacognitive capabilities." }