ArcanEdge-AI
- 12 skills
- 0 followers
- last month last updated
- ▌ Senior Code Review 2 · arcanedge-aiUse before finalizing a meaningful code change. Reviews the final diff for correctness, scope control, maintainability, validation gaps, safety risk, performance risk, accessibility risk, and subagent claims that still need verification.
- ▌ Worktree Lifecycle 2 · arcanedge-ai bundleManage Git worktrees created or used during a Codex task. Use when an agent proposes an isolated checkout, parallel writers may need filesystem isolation, a task already owns auxiliary worktrees, or the root must integrate, preserve, or remove task-created worktrees before completion.
- ▌ Reference Doc Routing 2 · arcanedge-aiUse when a task may need architecture, testing, access-control, design-system, API, release, data-model, subagent, or worktree-lifecycle reference documents. Helps select relevant docs, classify authority, and pass concise context to subagents.
- ▌ Subagent Orchestration 2 · arcanedge-aiUse for every repository task when Codex subagents are available. Makes the main agent the orchestrator, delegates bounded execution by default, enforces dependency-aware decomposition and fixed Luna/max routing, and requires main-agent verification.
- ▌ Task Graph Orchestration 2 · arcanedge-aiUse for complex coding tasks with substantial fan-out, multiple genuine dependencies, broad file or repository scope, multi-layer consolidation, separate implementation and verification paths, or an approval-gated irreversible action. Compiles an instruction-only task graph, executes only ready nodes, preserves completeness through fan-in, and retries only invalidated work. Skip for small or genuinely linear tasks.
- ▌ Multi Session Coordination 2 · arcanedge-aiUse when multiple independent Codex threads are working on related areas of the same project and need conflict detection, ownership, sequencing, and integration guidance.
- ▌ Senior Code Review · arcanedge-aiUse before finalizing a meaningful code change. Reviews the final diff for correctness bugs, regressions, scope creep, missing tests, safety, performance, accessibility, and maintainability risk — and for subagent claims that were never independently verified.
- ▌ Worktree Lifecycle · arcanedge-ai bundleUse when Git worktree isolation is being proposed for a Claude Code task, when parallel writers might need filesystem isolation, when a task already owns an auxiliary checkout, or when a task-created worktree must be integrated, preserved, or removed before finishing. Keeps worktree use finite, root-owned, and reconciled inside the task.
- ▌ Reference Doc Routing · arcanedge-aiUse when a task might need architecture, testing, access-control, design-system, API, release, data-model, subagent, or worktree reference documents. Helps pick the ones that matter, judge how much authority each has, read only the relevant sections, and pass concise labeled context to subagents.
- ▌ Subagent Orchestration · arcanedge-aiUse when planning how to delegate a coding task to Claude Code subagents — deciding whether to delegate at all, which role fits, how to write an assignment that comes back usable, how to run several safely in parallel, and how to verify results before accepting them. Use multi-session-coordination instead when other independent Claude Code sessions already own related work.
- ▌ Task Graph Orchestration · arcanedge-aiUse for a complex Claude Code task with substantial fan-out, several genuine dependencies, broad repository scope, layered consolidation, separate implementation and verification paths, or an approval-gated irreversible action. Makes the work topology explicit in Markdown, runs only the parts whose inputs are ready, preserves completeness through consolidation, and reruns only what a failure actually invalidated. Skip it for small or genuinely linear work.
- ▌ Multi Session Coordination · arcanedge-aiUse when several independent Claude Code sessions are working on related areas of the same project and someone has to hold the whole picture — discovering what is in flight, detecting incompatibilities beyond merge conflicts, assigning ownership of shared contracts, sequencing the work, and planning integration. Not for ordinary subagent delegation within one session.