Free cognitive-bias Agent Skill
Use Subadditivity Effect as a focused cognitive-bias lens. Explain the pattern, check whether it plausibly fits the situation, surface an alternative non-bias explanation, and return a practical counter-check without diagnosing people or claiming causation.
What the agent should do
The subadditivity effect occurs when separate estimates for parts of an event add up to more than the estimate for the same event described as one category. Unpacking changes the judgement, not the underlying probability. The effect has boundary conditions and can change or reverse with the description of the parts.
Install
Hermes Agent
mkdir -p ~/.hermes/skills/bias-probability-bias-subadditivity-effect && curl -fsSL https://cognitive-biases.github.io/agent-skills/bias-probability-bias-subadditivity-effect/SKILL.md -o ~/.hermes/skills/bias-probability-bias-subadditivity-effect/SKILL.mdOpenClaw
mkdir -p ~/.openclaw/workspace/skills/bias-probability-bias-subadditivity-effect && curl -fsSL https://cognitive-biases.github.io/agent-skills/bias-probability-bias-subadditivity-effect/SKILL.md -o ~/.openclaw/workspace/skills/bias-probability-bias-subadditivity-effect/SKILL.mdOther Agent Skills-compatible tools
Download SKILL.md and put it in the skill directory or import surface supported by your agent.
Boundary
The skill must preserve uncertainty, surface a non-bias explanation and avoid turning a cognitive-bias label into a claim about a person's motives, intelligence or mental health.