cobibean

Build a throwaway prototype to flesh out a design — a runnable terminal app for state/business-logic questions, or several radically different UI variations toggleable from one route.

Use when an agent or orchestrator needs to discover, inspect, retrieve, or install reusable skills from its configured Shared Skills Registry (SSR), while distinguishing registry access from caller-local installation.

Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes

Test-driven development. Use when the user wants to build features or fix bugs test-first, mentions "red-green-refactor", or wants integration tests.

Example-only skill that demonstrates how Shared Skills Registry MCP lists, previews, retrieves, and locally installs a small bundle.

Use when an agent needs to read, search, create, or edit Markdown notes inside an Obsidian vault. Resolve the vault path, keep writes inside the vault root, and use filesystem-first note operations.

Use when the user says “prep for reset”, “prep for a context window”, “prepare for a context window”, or invokes /prepforreset. Create curated Obsidian Daily Log and Session Log notes, then return a lean copy/paste handoff.

Use when the user invokes /wikijanitor or when a scheduled janitor reviews recent sessions. Reconcile Obsidian Daily Logs, Session Logs, and Janitor Reports over a lookback window; make conservative routine updates and notify only for review/uncertainty.

Use when grading, reviewing, rewriting, or approving a Hermes Agent SOUL.md. Uses the SOUL.md field-guide research artifacts as the only normative source for what makes a good SOUL.md.

Diagnose and safely repair Hermes Desktop, Hermes Dashboard, and remote Hermes agent fleet issues. Use when Hermes Desktop is stuck connecting, a remote gateway URL or session token may be wrong, profiles route to the wrong backend or workspace, dashboard services need verification, `/api/status` works but profile/session routes fail, Electron cache/state may be stale, or a remote Hermes profile works in one surface such as Telegram but fails in Desktop or Dashboard.

Use this skill when starting or finishing substantial project work that would benefit from durable continuity across Codex, Claude, or other agent sessions. It teaches agents to read recent project memory before work and write concise Markdown session memory after meaningful work under docs/memory/YYYY-MM-DD/descriptive-slug-memory-YYYY-MM-DD.md, including what happened, what was learned, decisions made, gotchas, changed files, verification, constraints, open decisions, and useful next context while avoiding secrets and sensitive details.

Step-by-step guide for deploying a Hermes Agent (by Nous Research) to a DigitalOcean droplet so it runs 24/7 with messaging-platform access (Telegram, Discord, Slack, etc.). Covers provisioning via doctl, Python 3.11 bootstrap on Ubuntu 24.04, non-interactive Hermes install, profile setup, credential migration from a laptop, OAuth caveats, and gateway-as-systemd-service. Load this skill when a user wants to move a Hermes agent off their laptop onto a cheap always-on cloud host.

Manage the local Codex Read Aloud plugin that speaks Codex assistant responses on macOS. Use when the user asks to install, configure, test, disable, uninstall, debug, or explain Codex text-to-speech/read-aloud behavior.

Write, review, or relay instructions that another agent, subagent, worker, Codex task, reviewer, or future session will act on. Use before spawning or delegating to subagents; calling spawn_agent, send_message, or followup_task; asking one agent to prepare a prompt for another; creating implementation, research, review, orchestration, or reset handoffs; returning a prompt for the user to copy and paste manually; or revising an incoming agent-authored prompt. Preserves user intent, decision provenance, appropriate agent autonomy, constraints, and task-specific proof without over-structuring capable models.

Run and synthesize multiple parallel subagents for research, review, option comparison, or independent verification. Use when the user asks to send N subagents, give subagents the same instructions, compare agent answers, create a mixture of agents, vary subagent reasoning levels or models, collect independent perspectives, or synthesize several subagent replies into one best answer.

Build temporary in-product pickers for comparing visual or UI alternatives, then help choose and lock the winning option. Use when the user says /compicker, $compicker, compicker, compare/pick visual variants, make a review picker, compare in a design-lab route/page, compare backgrounds, compare hero images, compare component designs, compare copy/layout/card treatments, or wants to audition multiple UI directions inside the actual app before committing.

Use when the user wants a long-running implementation run with phase gates, durable progress tracking, subagent execution, review checkpoints, or explicit wording like "$10ximplement", "10x implement", "run this plan", "phase-gate this build", "orchestrate builders", "keep implementation notes", "track decisions", or "continue this implementation". Combines a goal ledger, PM phase gates, subagent-driven plan execution, and executing-plans fallback into one workflow.

Keep your agent's cloud host in sync with your laptop by auto-pulling the repo when you git push. Python 3 stdlib webhook listener + user-systemd service + HMAC-SHA256. No CI, no ngrok, no pip install. Built for vibe coders and agent builders who run one bot/agent/service on one droplet and just want push-to-deploy.