Media production by an agent on a machine with an 8 GB GPU and commercial licensing, one skill per modality with rules that hold for any model: image and video generation, image retouching, transparent looping cut-outs for web pages, a mechanical check of a web page, voice-over at -16 LUFS, music and sound, stock media with per-item licenses, video composed as code, editing finished video, a 3D product turntable, scripted 3D scenes with characters in Blender, and sourcing rules for factual videos. Two hooks warn: where vipsthumbnail -o really writes, and when Kinocut runs before any skill is loaded.
13 skills · 1 MCP server · hooks
What fails silently when an agent works on a mobile app: device layer (adb, emulator, UI dumps, mirroring a phone) and bundler layer (Expo, Metro, prebuild). Narrow PreToolUse guards that warn and never block, and the Maestro MCP server for driving UI flows on a device.
5 skills · 1 MCP server · hooks
Software you maintain yourself on a Linux box, where the official docs stop: a Palworld dedicated server (the config it erases on shutdown), WireGuard through NetworkManager (the IPv6 leak that passes the IP test), and tarball installs (one versioned layout; self-updating apps stay out of /opt).
3 skills
Rules for multi-agent work: model routing per phase, fan-out sizing, closed output schemas and citation checks, with CLI workers for search and extract and hooks that load the rules and check routing at dispatch.
1 skill · hooks
Another vendor's agent CLI as an isolated subagent, from any host: ask a read-only question or hold a multi-turn conversation (agy, opencode, Codex, Grok, Hermes, Copilot, Cursor, Muse, Kimi, or Claude Code from another host), get a second opinion on a diff, or delegate one coding slice to agy, opencode, Copilot, Codex, Grok, Cursor or Claude Code under the worker CLI's own hooks and a mechanical gate. Tiers (cheap, mid, top) map to a model and effort per backend. The verdict never reads the CLI's exit 0 as success.
1 skill · hooks
Pins the installed game's version before the agent answers from training memory: a UserPromptSubmit hook reads Steam appmanifests and injects appid, buildid, update date and the patch-notes URL, and a skill routes the question and weighs sources against the last patch.
1 skill · hooks
Writes, trims and audits a repository's AGENTS.md: what belongs in it, what to cut, size limits per tool, and a stdlib checker for missing commands, dead paths, oversize files and a CLAUDE.md that hides AGENTS.md.
1 skill
Audits a repository of agent skills, plugins and hooks against Anthropic's authoring guidance and current prompting research: a sourced checklist for the judgment calls and a stdlib linter for the mechanical ones.
1 skill