Orchestrate multi-step agent tasks under agf.md governance rules.
npx skills add https://github.com/cassioalexandre/agf --skill main-skill
Please help me install this Agent Skill. Skill: main-skill Source: https://github.com/cassioalexandre/agf/tree/main/.agents/skills/main-skill Command: npx skills add https://github.com/cassioalexandre/agf --skill main-skill
It prevents agent actions from violating the project’s governance rules by centralizing decision-making, risk classification, validation, and controlled delegation to subordinate skills.
Tell the orchestrator what you want to change in your repo and include the keyword "borabill" so it can initialize safely and propose a plan when needed.
High-intent search queries and answers about installing and using this skill.
You coordinate safe governance across agent skills by using an orchestrator that applies an inviolable rule source to manage delegation, classify risks, and enforce strict safety stops. It validates environment outcomes to prevent rule violations and hallucinations.
Pre-flight validation is the process of reading governance rules and memory sources before executing writes or commands. It verifies actual environment outcomes rather than assuming success, preventing hallucinations and ensuring actions do not contradict defined constraints.
You classify task risks into TURBO, CAUTION, PLANNING, or SECURITY STOP modes based on reversibility and impact. This risk-mode selection determines how delegation is controlled and includes a 3-attempt failure policy for failed executions.
Start a governed orchestration session by telling the orchestrator what you want to change in your repository and including the initialization keyword. It safely initializes, proposes a structured plan when needed, and waits for explicit approval before applying changes.
Agent skill execution fails after multiple attempts due to a 3-attempt failure policy triggered by risk classification. When tasks exceed this limit, the orchestrator enforces a SECURITY STOP to prevent uncontrolled or repeated damaging actions.
HypoxanthineOvOManage Hypo-Workflow rule severities, hooks, and rule packs via YAML.
jellydnResume work from saved handoff sessions in .claude/handoffs/.
bostonaholicStandardize multi-step agent procedures with a todo-first ledger convention.
latestaiagentsDesign supervisor-based multi-agent systems with routing and state management.
