메뉴
Capture notes, learnings, and context into ChromaDB for long-term memory. Use when saving something to remember, recording a discovery or decision, storing context for future sessions, or building the knowledge base.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Launch a Conclave workspace on Runpod. Use when the user wants to start a new GPU pod with the full Conclave environment (Matrix, Planka, Ollama, browser, coding agents, etc).
Connect to the Chromium browser via CDP (Chrome DevTools Protocol) for automation, screenshots, and cookie management. Use when you need to interact with web pages, capture screenshots, or automate browser tasks.
Build, test, and iterate on the Conclave container. Use when making changes to Conclave services, configs, scripts, skills, or extensions and need to rebuild the container image and verify everything works.
Manage tasks and projects in Planka kanban board — create cards, move between columns, track progress. Use when creating tasks, updating card status, checking project boards, managing the kanban workflow, or organizing work items.
Push metrics to Prometheus Pushgateway for Grafana dashboards. Use when recording operational metrics, pushing gauge values to Prometheus, or updating Grafana dashboard data points.
Send messages, read channel history, upload files, manage reactions, and list rooms on Matrix. Use for Matrix communication — posting messages, sharing files, reading conversations, checking reactions, or monitoring rooms.
| name | take-note |
| description | Capture notes, learnings, and context into ChromaDB for long-term memory. Use when saving something to remember, recording a discovery or decision, storing context for future sessions, or building the knowledge base. |
Quickly capture notes, learnings, observations, and context. Builds a searchable knowledge base over time.
pip install chromadb
python3 {baseDir}/note.py "Learned that Flux models respond better to natural language prompts"
Auto-generates ID with timestamp: note-2026-02-10-213045-abc123
python3 {baseDir}/note.py "User prefers dark fantasy aesthetics with purple and green color schemes" \
--category user-preferences \
--tags "color,style,fantasy"
python3 {baseDir}/note.py "Important discovery about training parameters" \
--id important-training-discovery
python3 {baseDir}/note.py --search "training parameters" --limit 5
python3 {baseDir}/note.py --list [--limit 10]
Each note includes:
note-YYYY-MM-DD-HHMMSS-hash or customSuggested categories (use what makes sense):
learning — Technical discoveries, how-tosuser-preferences — User likes/dislikes, style preferencescontext — Background info, project contextdecision — Decisions made and whyidea — Ideas for future explorationobservation — Patterns noticed over timegeneral — Default for uncategorized notesCHROMADB_HOST: ChromaDB server host (default: localhost)CHROMADB_PORT: ChromaDB server port (default: 8000)# Quick learning capture
python3 {baseDir}/note.py "The scratching LoRA works best at weight 0.8-1.0 for sketch effects"
# User preference with tags
python3 {baseDir}/note.py "User avoids overly bright colors, prefers muted/dark palettes" \
--category user-preferences \
--tags "color,palette,style"
# Technical decision with category
python3 {baseDir}/note.py "Switched to cosine scheduler for all future training - smoother results" \
--category decision \
--tags "training,scheduler"
# Search for relevant context
python3 {baseDir}/note.py --search "user color preferences"
# List recent notes
python3 {baseDir}/note.py --list --limit 5
notes collection alongside any imported datachromadb skill for advanced semantic search