Analyze structured observation data to evaluate pattern effectiveness and update knowledge bases.
npx skills add https://github.com/kleisli-io/kli --skill kli-reflection
Please help me install this Agent Skill. Skill: kli-reflection Source: https://github.com/kleisli-io/kli/tree/main/plugin/skills/kli-reflection Command: npx skills add https://github.com/kleisli-io/kli --skill kli-reflection
This Skill addresses the challenge of systematically learning from AI task execution by providing a structured framework for evaluating pattern effectiveness and updating knowledge bases based on concrete evidence.
Use the kli-reflection skill to analyze the observations from the last task and update pattern effectiveness ratings.
High-intent search queries and answers about installing and using this skill.
You can evaluate AI agent performance by applying structured retrospective analysis to observation data from task execution. This evidence-based learning approach classifies pattern effectiveness as helpful, harmful, or neutral to objectively assess outcomes.
Evidence-based learning for AI pattern evaluation is a framework that grounds all assessments of decision-making heuristics in concrete, observable task data. It moves beyond subjective opinions by requiring structured observation data to classify outcomes.
You can update your AI knowledge base from task outcomes by analyzing structured observations to determine if applied patterns saved time, introduced errors, or were ineffective. This process refines decision-making heuristics and operational playbooks continuously.
Harm signal tiers in AI operations define distinct levels of operational harm with corresponding response actions, ranging from auto-correction to flagging for human review. They provide clear criteria for classifying AI patterns as helpful, harmful, or neutral.
Yes, you need structured observation data to evaluate AI patterns effectively. Evidence-based pattern evaluation requires concrete task data and defined harm signal tiers to objectively assess whether an AI approach was helpful, harmful, or neutral.
The best way to discover new AI patterns during task execution is by analyzing structured observation data to identify and document novel, reusable approaches. This retrospective analysis ensures new patterns are grounded in concrete evidence before updating the knowledge base.
heymishyRecord a metric signal and evidence in pipeline-state.json.
KeyValueSoftwareSystemsDistill completed Maestro workflow runs into structured lessons and documentation updates.
Human-Frontier-Labs-IncDeploy AI agent swarms across multiple topologies for event-driven cloud workflows.
johnlindquistCapture and organize coding lessons from sessions and reviews into reusable patterns.
