Define outcome specifications with evidence-grounded success criteria and anti-goals.
npx skills add https://github.com/tannishmango/skills --skill goal
Please help me install this Agent Skill. Skill: goal Source: https://github.com/tannishmango/skills/tree/main/goal Command: npx skills add https://github.com/tannishmango/skills --skill goal
This Skill prevents the "Goodhart's Law problem" where a measure becomes a target and ceases to be a good measure, by enforcing rigorous outcome specifications with evidence-grounded success criteria and explicit anti-goals.
Use the goal skill to define a new goal for improving system performance with specific success criteria and anti-goals.
High-intent search queries and answers about installing and using this skill.
To prevent Goodhart's Law during goal setting, separate desired outcomes from measurement methods by defining evidence-grounded success criteria and explicit anti-goals. This approach ensures metrics remain valid indicators rather than becoming corrupted targets.
Anti-goals in project management explicitly define what must NOT happen during a task, preventing harmful side-effects. Use them when launching features to ensure no negative impact occurs on system performance or user satisfaction while pursuing primary objectives.
Setting up outcome verification for AI agent orchestration involves defining multi-dimensional success criteria with specific evidence requirements. This enforces rigorous outcome specifications that validate agent task completion against predefined constraints and desired results.
Evidence-grounded goal setting supports multi-dimensional criteria by basing objectives on specific data rather than vague intentions. This method ensures complex tasks are measured across multiple verification protocols to confirm comprehensive success without metric manipulation.
The best way to prevent negative side-effects during feature launches is enforcing explicit anti-goals alongside standard success criteria. This practice ensures user adoption numbers do not overshadow critical constraints like system performance or user satisfaction.
Measuring a target makes it a bad metric because of Goodhart's Law, where a measure becomes a target and ceases to be a good measure. Mitigate this by separating desired outcomes from their measurement methods during goal definition.
mwayDefine analysis scope, objectives, constraints, and success criteria with evidence and risk assessment.
Agile-VExecute independent tests against Build Agent artifacts to validate requirements compliance.
DergaczAutomate implementation and verification of technical plans from plan.md files.
chenguojun06-starManage Feishu OKR cycles, objectives, key results, and progress records.
