A parallel six-inspector audit workflow for a Claude Code skills marketplace repo that verifies every serious finding before reporting it by priority.
Designs production-grade multi-agent workflows with the right pattern, explicit handoff contracts, failure handling, and cost/context budgets.
Decides which model, effort level, and cascade shape each subagent gets, routing on measured cost per completed task instead of per-token price.
A six-phase research pipeline that mines GitHub and skill registries for public SKILL.md files, classifies repos by archetype, and documents reusable processes and plugin ideas.
A skill for designing privacy-aware traces, spans, metrics, alerts, and investigation runbooks for AI agents.
Phase 1 of the prd-taskmaster pipeline: brainstorm-driven discovery that extracts constraints, classifies project scale, and hands off to spec generation.
A bounded multi-agent command protocol that splits genuinely independent work across 5–100+ isolated workers, verifies outputs independently, and integrates serially.
Redesigns a personal or team operating system as a supervised learning loop with Four-C diagnostics and one bounded, reversible experiment.
Fan one task out to N parallel agents, score them against a rubric, then pick a base and graft the best parts of the losers into a single synthesized result.
Enforces a strict output contract for claude-mem observer jobs on Grok Bot: either skip_summary or exactly one observation XML, never prose.
A coordination adapter skill that keeps multiple AI writers from colliding in one repo using messages, acknowledgements, and TTL-bounded advisory file reservations.
Delegates a task to the AGY (Antigravity) CLI only when the caller explicitly picks that runtime, with a disclosed permission posture and returned run evidence.
A maintainer-only workflow for qualitatively studying private short-drama project chains and promoting findings into public references only after de-identification, blind eval, and independent review.
Collects every operation event across a multi-agent system and turns it into a trace_id-linked evidence chain plus vector index for audit and replay.
Runs headless provider LLMs (DeepSeek, GLM, Kimi, Qwen, MiniMax) via the cc-fleet CLI for one-shot answers or lock-free parallel fan-out.
Teaches Claude to author and run multi-phase, resumable JS orchestration scripts over cc-fleet provider subagents.
Delegates one-shot or flat parallel batches of work to cc-fleet provider models (DeepSeek/GLM/Kimi/Qwen/MiniMax/Codex/Claude) via the headless `cc-fleet subagent` CLI.
Author a JS script that orchestrates multi-phase, resumable runs across many cc-fleet provider subagents, off the main context.
A cc-fleet skill that spawns long-lived third-party LLM teammates in tmux panes and coordinates them across turns via SendMessage.
Turns Claude Code into a coordinator that lets multiple agents message each other, share context, and hand off tasks through a single shared state file.