eonraider

Session ergonomics for Claude Code: /simplicity:just-ask collects every open question in the session and asks them through AskUserQuestion with a recommended option and reasoning; /simplicity:just-move prepares a session to run unattended: it asks every open decision and blocker up front, fixes what it can once you agree, and schedules a resume for after your usage limit resets; /simplicity:speed-up maps the session's remaining tasks as a dependency tree, proposes a reviewed reorder with safe parallel groups, and applies it and notifies the affected sessions only after you agree; /simplicity:just-say-it re-states the latest response(s) as a short plain-language numbered list, or explains a commit, PR, document or pasted snippet the same way with no technical jargon; /simplicity:what-now takes stock of the session and lists what's done, where the current task stands and the next steps; /simplicity:just-finish-it pushes this session's work, merges the PRs whose CI is green, runs /simplicity:cleanup and archives the session, or reports what's still pending; /simplicity:shift-session commits and pushes the work, updates the trackers, hands it to a fresh child session and closes this one, and a hook suggests it once context use reaches a threshold; /simplicity:call-it-a-day puts the session on hold at the end of the day: it winds down the agents and sessions it supervises and stops last, commits and pushes the work in progress, updates the trackers, saves a handoff note to memory, renames the session "<topic> -> On hold" and prints a handoff prompt, and the session gets its old title back when the work resumes; /simplicity:cleanup runs an end-of-session closing checklist; /simplicity:disk-cleanup frees disk space without losing work: it measures every candidate exactly, checks each worktree and build folder for uncommitted work, unsafe commits and running processes, deletes only what you pick, and in orchestrator mode has each coordinated session clean its own area first

12 skills

Sizes subagents to the task. Four pre-sized agent tiers (scout, worker, analyst, judge) with pinned model and effort; a routing skill; an enforcement hook that rejects mismatched spawns with one line naming the correct call; a per-project spawn log with outcome signals; and /agent-router:report, which proposes tier changes from that log for a person to accept or reject. Optional vendor mode covers cloud sessions. Hooks make no model or network calls.

3 skills · 4 agents

Build, validate, secure, and maintain Claude Skills. Supersedes Anthropic's shipped skill-creator with a spec-compliant, cross-vendor, security-hardened successor: frontmatter validation, a combined decision/dedup gate, a five-surface compatibility matrix, gitleaks-based security scanning with tamper detection, the full with/without-skill evaluation engine, lifecycle classification, and an audit mode for upgrading, rebuilding, and bulk-auditing existing skills.

1 skill · 3 agents

Language-agnostic methodology for writing and refactoring code against SOLID, GoF design patterns, and complementary principles (DRY, KISS, YAGNI, Boy Scout, SoC, Encapsulation, Law of Demeter, CQS, TDD, AHA) — for whole-worktree refactors and live in-flight review alike. Wires into whatever CI/QA a repo already has and verifies every finding through an adversarial reviewer/verifier subagent pair before touching code.

1 skill · 2 agents

Personal orchestrator for Claude Code: tracks a repo's Initiative through an agent-assisted engineering methodology, phase by phase.