opencode-drewgent

{{AGENT_NAME}} brain signal architecture — event bus, signal processor, awareness reporter, brain monitor

TDD-PDCA plan for building a self-replicating branching agent in {{AGENT_NAME}}. 단일 에이전트가 스스로 서브에이전트를 분기하고, 결과를 수렴시키는 구조.

Integrate {{AGENT_NAME}} with ACP (Agent Client Protocol) agents. Reverse-engineered from goose's Rust ACP provider implementation.

Maintain .{{AGENT_NAME_LOWER}} as an Obsidian vault — graph connectivity overhaul (P-layer mesh, skill clusters, SEO article web), broken link detection, backlink density verification via Obsidian CLI, filename deduplication, bidirectional linking, auto-generated orphan management, cron output graph bloat prevention, and Obsidian compatibility.

AI Trend Collection & Filtering — collect GitHub trending repos, score via 5-axis philosophy filter, evaluate, apply, and retire.

RSS 피드 모니터링 → SEO 기사 수집·분석·트렌드 리포트

Authoring and using agent profiles — pre-defined subagent roles that set model, provider, toolsets, and instructions in one task() call. Covers profile format, the 14 standard roles, pipeline patterns, cost-tiered model assignment, and the ESCALATE mechanism for capability routing.

npm으로 설치된 n8n 셀프호스트 상태 진단 + launchd 관리 스킬

Add a "Thinking..." spinner card to an ACP server's tool-call stack so ACP clients (Zed Agent Panel, openCode TUI, JetBrains/VS Code ACP UI) show a working indicator during the LLM API call phase. Triggers when working on an Agent Client Protocol (ACP) server and the user wants LLM-thinking visibility in the client's tool-call stack.

CMO-style autonomous content agent. Observes user's work activity (sessions, git, kanban), curates narrative-worthy material, produces multi-format drafts (blog + X thread), maintains narrative arc continuity, and generates visual assets (SVG cover, Mermaid, Excalidraw PNG). session: "2026-06-14 content-manager-cmo-agent" decision: "Agent profile + cron trigger + narrative_arc.md tracking file (not a multi-stage pipeline). SVG over paid image API (user: '비용 발생안시키고 싶어'). Excalidraw→PNG via Puppeteer for architecture diagrams."

Complete guide to using and extending {{AGENT_NAME}} Agent — CLI usage, setup, configuration, spawning additional agents, gateway platforms, skills, voice, tools, profiles, and a concise contributor reference. Load this skill when helping users configure {{AGENT_NAME}}, troubleshoot issues, spawn agent instances, or make code contributions.

{{AGENT_NAME}} 코어 시스템 (AIAgent, signal, kanban, dispatcher, worker) 의 기본기를 점검하는 6-Phase 절차. docs와 reality mismatch 발견 시 methodology.

{{AGENT_NAME}} kanban dispatcher의 logfile redirect + dead worker watchdog 적용 skill. PIPE deadlock 회피, dead worker 즉시 reclaim.

Diagnose and recover stuck kanban in_progress tasks (worker dead, claim_expires expired, dispatcher down). Includes SQL reset script, root cause analysis, and preventive monitoring.

Diagnose launchd service state when launchctl list shows stale exit codes but process is actually running

Clean up {{AGENT_NAME}}'s persistent memory file (MEMORY.md) when it nears the 8K char cap. Identify resolved/one-time entries, preserve operational facts, verify after.

Filename uniqueness, wikilink resolution safety, P-layer identity in naming, and Obsidian exclusion boundaries. Prevents duplicate-name ambiguity and graph pollution.

"CMO-style content agent — observes work activity, produces multi-format drafts with SVG/Mermaid/Excalidraw visuals, pushes to WordPress via MCP. User-facing: 'CMO 에이전트가 내 작업을 보고 콘텐트로 만든다'."

{{AGENT_NAME}} content pipeline - editorial topic selection, draft writing, Korean humanization, review-ready publishing

Diagnose and recover cron jobs in jobs.json that stopped running (next_run_at=null, no recent execution). Distinguishes scheduler bug vs script quality issue. Applies 60-second recovery patch.

"{{AGENT_NAME}} 에이전트 상태를 Cloudflare Workers + KV로 호스팅. 4-tab insight-driven dashboard (Overview/System/Brain/Usage) + live agent activity + 23 collectors"

Add a script field to jobs.json + scheduler.py branch to bypass LLM for simple shell cron jobs (cost + reliability). Reusable pattern for any 'Run: python3 xxx.py' style cron.

Decomposition playbook + anti-temptation rules for an orchestrator profile routing work through Kanban. The "don't do the work yourself" rule and the basic lifecycle are auto-injected into every kanban worker's system prompt; this skill is the deeper playbook when you're specifically playing the orchestrator role.

Pitfalls, examples, and edge cases for Hermes Kanban workers. The lifecycle itself is auto-injected into every worker's system prompt as KANBAN_GUIDANCE (from agent/prompt_builder.py); this skill is what you load when you want deeper detail on specific scenarios.

Check MiniMax Token Plan usage, current 5h interval, and weekly remaining — terminal-friendly output (no browser/console). Use when the user asks about "Token Plan 남은거", "minimax 사용량", "리셋까지 얼마", "주간 한도", or wants a quick status check without opening platform.minimax.io/console.

WordPress CMS setup on Docker + Blocksy theme + MCP integration for autonomous content publishing

"WordPress on Docker (colima) with Blocksy theme, custom fonts, MCP server for AI agent integration."

Cloudflare Pages에 wrangler로 deploy할 때 발생하는 'More than one account available but unable to select one in non-interactive mode' 또는 403 Authentication error 해결 절차. OAuth token에 여러 CF 계정이 연결돼 있을 때 (personal + org 흔함) 발생하는 multi-account selection 문제 + wrangler.toml이 Pages 형식과 안 맞는 문제. YOUR_DOMAIN, Quartz publishing pipeline, fswatch 자동 deploy 등 영향.

memory/vault의 "Done" path claim이 실제 filesystem에 존재하는지 검증

{{AGENT_NAME}} kanban board UI — Flask dashboard server for visual task management

kanban dispatcher cron이 안 돌거나 dead worker reclaim 실패를 4가지 fail mode로 진단

PortOne V2 SDK + D1 payment database integration for Korean PG (KG이니시스/카카오페이/토스페이먼츠) in a Cloudflare Workers SPA. Covers: SDK setup, per-entity fingerprint purchase tracking, D1 schema, mobile redirect, and multi-channel config.

⚠️ DEPRECATED — gbrain replaced by SQLite FTS5 + Ollama embeddings (2026-07-05). Use recall()/remember() tools instead.

Built-in MCP (Model Context Protocol) client that connects to external MCP servers, discovers their tools, and registers them as native {{AGENT_NAME}} Agent tools. Supports stdio and HTTP transports with automatic reconnection, security filtering, and zero-config tool injection.

Serverless GPU cloud platform for running ML workloads. Use when you need on-demand GPU access without infrastructure management, deploying ML models as APIs, or running batch jobs with automatic scaling.

GGUF format and llama.cpp quantization for efficient CPU/GPU inference. Use when deploying models on consumer hardware, Apple Silicon, or when needing flexible quantization from 2-8 bit without GPU requirements.

Control LLM output with regex and grammars, guarantee valid JSON/XML/code generation, enforce structured formats, and build multi-step workflows with Guidance - Microsoft Research's constrained generation framework

Remove refusal behaviors from open-weight LLMs using OBLITERATUS — mechanistic interpretability techniques (diff-in-means, SVD, whitened SVD, LEACE, SAE decomposition, etc.) to excise guardrails while preserving reasoning. 9 CLI methods, 28 analysis modules, 116 model presets across 5 compute tiers, tournament evaluation, and telemetry-driven recommendations. Use when a user wants to uncensor, abliterate, or remove refusal from an LLM.

Guarantee valid JSON/XML/code structure during generation, use Pydantic models for type-safe outputs, support local models (Transformers, vLLM), and maximize inference speed with Outlines - dottxt.ai's structured generation library

OpenAI's model connecting vision and language. Enables zero-shot image classification, image-text matching, and cross-modal retrieval. Trained on 400M image-text pairs. Use for image search, content moderation, or vision-language tasks without fine-tuning. Best for general-purpose image understanding.