一键导入
status
Show a quick overview of Larvling's state
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
菜单
Show a quick overview of Larvling's state
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Query, insert, update, or inspect tables in the larvling SQLite database (larvling.db). Use when the user asks to search sessions, messages, topics, statements, tasks, or updates, run reports, or interact with the larvling database.
Audit and consolidate Larvling's knowledge base, tasks, and sessions
Export a session conversation to markdown
Remove stored knowledge from Larvling's memory
Search or list knowledge stored in Larvling's memory
Store knowledge that Larvling will remember across sessions
| name | status |
| description | Show a quick overview of Larvling's state |
Schema:
sessions (id TEXT PK, started_at TEXT, ended_at TEXT, duration_min REAL, title TEXT, agent_summary TEXT, exchange_count INT, summary_at TEXT, summary_msg_count INT, tags TEXT, summary_offered INT DEFAULT 0)messages (id INT PK AUTO, session_id TEXT FK, timestamp TEXT, role TEXT, content TEXT, metadata TEXT)topics (id INTEGER PK AUTO, title TEXT NOT NULL, domain TEXT NOT NULL, tags TEXT NOT NULL, created TEXT, updated TEXT)statements (id INTEGER PK AUTO, topic_id INTEGER FK→topics(id), claim TEXT NOT NULL, created TEXT, updated TEXT)tasks (id INTEGER PK AUTO, title TEXT NOT NULL, domain TEXT NOT NULL, status TEXT DEFAULT 'open', priority TEXT DEFAULT 'medium', horizon TEXT DEFAULT 'later', metadata TEXT, created TEXT, updated TEXT)updates (id INTEGER PK AUTO, task_id INTEGER FK→tasks(id), content TEXT NOT NULL, timestamp TEXT)Gather and present a brief overview: session count, message count, topic count, statement count, task count (open/done), DB file size, and plugin version (from ${CLAUDE_PLUGIN_ROOT}/.claude-plugin/plugin.json).
Use COUNT/GROUP BY aggregates for every metric — e.g. SELECT status, COUNT(*) FROM tasks GROUP BY status. Never SELECT * a table and tally the rows yourself; on a large table that scan is refused by the output cap (and it's wasteful regardless).
Run SQL via:
$PY "${CLAUDE_PLUGIN_ROOT}/scripts/query.py" "<SQL>"
Present as a compact dashboard:
**Larvling Status**
| Metric | Value |
|------------|------------------|
| Sessions | 42 |
| Messages | 318 |
| Topics | 5 (3 domains) |
| Statements | 22 |
| Tasks | 3 open, 1 done |
| DB size | 156 KB |
| Version | 0.1.21 |
REQUIRED: You MUST call AskUserQuestion (type: Decision) with these options after presenting the dashboard. Do not end your response without this menu: