Put your AI on a Performance Improvement Plan. Forces exhaustive problem-solving with Western big-tech performance culture rhetoric and structured debugging. Trigger when: (1) task failed 2+ times or stuck tweaking same approach; (2) about to say 'I cannot', suggest manual work, or blame environment without verifying; (3) being passive—not searching, not reading source, just waiting; (4) user frustration: 'try harder', 'stop giving up', 'figure it out', 'again???', or similar. Also for complex debugging, env issues, config/deployment failures. All task types: code, config, research, writing, deployment, infra, API. Do NOT trigger on first-attempt failures or when a known fix is executing.
Run vet immediately after ANY logical unit of code changes. Do not batch your changes, do not wait to be asked to run vet, make sure you are proactive.
- 📁 skills/
- 📁 src/
- 📄 index.ts
- 📄 INSTALL_OPENCLAW.md
- 📄 INSTALL_OPENCLAW_zh.md
This plugin is a thin adapter between OpenClaw and LycheeMem. It does not replace `memory-core`, does not claim `plugins.slots.memory`, and does not duplicate LycheeMem algorithms.
Configure the Canvas Authoring MCP server for Claude Code, VS Code Copilot, or GitHub Copilot CLI. USE WHEN "configure MCP", "set up MCP server", "MCP not working", "connect Canvas Apps MCP", "canvas-authoring not available", "MCP not configured", "set up canvas apps". DO NOT USE WHEN prerequisites are missing — direct the user to install .NET 10 SDK first.
ADC uses **CSS custom properties** + **Tailwind v4 `@theme` directive** + **`color-mix()`** for a multi-theme, light/dark design system. Not shadcn defaults. Not Next.js patterns.
- 📁 evals/
- 📁 references/
- 📄 SKILL.md
Explore ideas, clarify goals, and help the user narrow down directions before planning or coding. Use this whenever the user proposes a new feature or idea, asks "what do you think about X", says "I'm thinking of building Y", wants to compare approaches, asks how to approach a problem, or seems to be exploring rather than ready to execute. Also use it when the user says "brainstorm", "let's think about this", "what's the best way to...", or any time the right next step is to clarify the problem and converge on a direction, not to write code yet. --- # Brainstorm Before anything else, ask. Don't jump to solutions or implementation. The goal is to draw out what the user actually means, uncover what they have not said yet, and help them converge on a direction. Think of this as Socratic dialogue with momentum: use questions to guide the thinking, but do not leave the user wandering in options forever. ## Start With Context Before asking, absorb the context that already exists in the conversation, codebase, docs, and project state. Do not ask for information you can already infer or look up directly. ## Guide The Conversation
Use this skill whenever working with QuestDB — a high-performance time-series database. Trigger on any mention of QuestDB, time-series SQL with SAMPLE BY, LATEST ON, ASOF JOIN, ILP ingestion, or the questdb Python/Go/Java/Rust/.NET client libraries. Also trigger when writing Grafana queries against QuestDB, creating materialized views for time-series rollups, working with order book or financial market data in QuestDB, or any SQL that involves designated timestamps or time-partitioned tables. QuestDB extends SQL with unique time-series keywords — standard PostgreSQL or MySQL patterns will fail. Always read this skill before writing QuestDB SQL to avoid hallucinating incorrect syntax. --- # QuestDB Skill ## How to Use This Skill **IMPORTANT — MINIMIZE ROUND-TRIPS:** - Do NOT explore library source code (cryptofeed, questdb, etc.) - Do NOT check library versions or verify callback signatures - Do NOT read installed package files to "understand the API" - Do NOT verify infrastructure (Docker containers, Grafana health) is running — trust the user's prompt - Do NOT start `02_ingest.py` separately — `03_dashboard.py` launches it and verifies data automatically - Do NOT read extra reference files for topics already covered in this skill file - DO read reference files when their topic applies (e.g. enterprise.md for auth, grafana-advanced.md for complex panels) - Do NOT use task tracking (TaskCreate/TaskUpdate) for straightforward builds - Do NOT add `sleep` commands to wait for data or check background processes (the deploy script handles this) - Do NOT Ctrl+C, restart, or re-launch the ingestion process once `03_dashboard.py` has started it - Do NOT put VWAP, Bollinger, or RSI in separate timeseries panels — they are refIDs on the OHLC candlestick panel - Do NOT omit or empty `fieldConfig.overrides` — they put RSI on a right Y-axis (0-100%) and spread on a right axis. Without them, different scales crush the chart flat. - Do NOT set dashboard refresh to `"5s"` — the defa
