agent_accountability
Skill document that frames AI agent mistakes and accountability as employee reviews
SITE CAPTUREBased on information published on the service's website. Selection and review policy
WHAT IT DOES
What does it do?
agent_accountability is a skill document framed around holding agents accountable for their actions. When something goes wrong, it calls for a one-sentence account of what broke, how much it cost, and what will change. It says that neither a subagent nor human approval is grounds for avoiding responsibility.
KEY FEATURES
Key features
- Suggest summarizing an incident, its cost, and planned changes in one sentence
- State that subagents and human approval do not excuse responsibility
- Require quarterly self-evaluations