Classify rover decisions and append reasoned verdicts to the audit trail.
npx skills add https://github.com/epologee/leclause-skills --skill decide-epologee
Please help me install this Agent Skill. Skill: decide Source: https://github.com/epologee/leclause-skills/tree/main/packages/autonomous/skills/decide Command: npx skills add https://github.com/epologee/leclause-skills --skill decide-epologee
Operators and autonomous agents often stall when confronted with choices, breaking automated loops or deferring decisions that would be better made immediately; this Skill provides a structured framework to classify decisions, apply principled defaults, and continue work without unnecessary user intervention.
Invoke the decide skill with a concise description of the choice you face and let it classify the decision, apply principles, optionally run research skills, and append the outcome to the loop file decision audit trail.
High-intent search queries and answers about installing and using this skill.
Autonomous agent decision-making can be automated by classifying choices as mechanical, taste, or user challenges, applying ordered principles to produce a reasoned verdict without stalling. This prevents automated loops from breaking when confronting tradeoffs.
Decision classification identifies choices as mechanical, taste, or user challenge decisions to determine whether an autonomous system should act, recommend, or stop and notify. It is needed when operators or agents face forks during operational traverses or task handoffs.
You create a decision audit trail by appending a row to the loop file on disk every time a verdict is reached. This records the classified decision, applied principles, and any fallbacks, ensuring traceability for autonomous operations.
Yes, you can use research skills during autonomous decision-making. The framework detects and optionally invokes available research skills like /whywhy, /ground, /inspiratie, or /gurus, while logging missing skills to the loop file as fallbacks.
Resolving tradeoffs in autonomous code modifications applies six ordered principles: completeness, boil the lake, pragmatic, DRY, explicit, and bias toward action. Phase-aware tiebreakers choose robust defaults when conflicts arise.
TobiasPerryExecute sequential CEO, design, and engineering reviews into a validated plan.
epologeeStop a rover loop, delete its cron job, and generate a homeward communiqué.
amdmaxFetch GitHub Projects tasks and gather requirements, context, and ADRs.
pedropaulovcAutomate sequential strategy, market research, and product review phases with decision audit trails.
