菜单
Read and query journal entries. Shows recent activity, search by keyword, or filter by date. Can read from both SQLite and markdown entries.
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
Show usage information for any workflow-toolkit skill. Use when the user asks how to use a specific skill, what a skill does, or needs guidance on workflow-toolkit features. Also triggers on "help", "how do I", "what does X do", or "usage" in the context of workflow-toolkit skills.
Initialize and verify the workflow-toolkit environment. Checks SQLite availability, creates the journal database, verifies directory structure, checks gh CLI auth, and reports readiness. Run this after installing the plugin or when troubleshooting.
Pre-flight checklist for open source contributions. Validates code changes against project rules, PR size limits, commit format, linting, and type-checking. Does NOT push or create PRs — only reports issues.
Creates a new Claude Code custom subagent configuration. Guides you through designing agent scope, tools, permissions, hooks, and system prompts.
Creates a new Claude Code custom skill (slash command). Guides you through designing, writing, and placing the SKILL.md file with proper frontmatter, arguments, and best practices.
Open the journal SQLite database in a GUI viewer (DB Browser for SQLite, or similar).
| name | journal-read |
| description | Read and query journal entries. Shows recent activity, search by keyword, or filter by date. Can read from both SQLite and markdown entries. |
| argument-hint | ["recent|today|search <keyword>|all|last <N>"] |
| disable-model-invocation | true |
This skill queries and displays journal entries from the SQLite database.
Parse $ARGUMENTS to determine the query type:
recent or no arguments: Show the last 10 entriestoday: Show entries from todaysearch <keyword>: Search summaries and details for the keywordall: List all entries (summaries only, for brevity)last <N>: Show the last N entriesQuery the SQLite database at $CLAUDE_PROJECT_DIR/.claude/journal/journal.db using Bash.
Use the following query patterns:
DB_PATH="$CLAUDE_PROJECT_DIR/.claude/journal/journal.db"
# recent / default — last 10 entries
sqlite3 -header -column "$DB_PATH" "SELECT id, timestamp, type, summary FROM entries ORDER BY timestamp DESC LIMIT 10;"
# today
sqlite3 -header -column "$DB_PATH" "SELECT id, timestamp, type, summary, details FROM entries WHERE date(timestamp) = date('now') ORDER BY timestamp DESC;"
# search <keyword>
sqlite3 -header -column "$DB_PATH" "SELECT id, timestamp, type, summary FROM entries WHERE summary LIKE '%<keyword>%' OR details LIKE '%<keyword>%' ORDER BY timestamp DESC;"
# all
sqlite3 -header -column "$DB_PATH" "SELECT id, timestamp, type, summary FROM entries ORDER BY timestamp DESC;"
# last <N>
sqlite3 -header -column "$DB_PATH" "SELECT id, timestamp, type, summary, details FROM entries ORDER BY timestamp DESC LIMIT <N>;"
Format the results in a readable way for the user.
Also check if there are markdown entries in $CLAUDE_PROJECT_DIR/.claude/journal/entries/ and mention them if present, so the user knows they can read detailed entries there.
If the database does not exist, inform the user and suggest initializing it. The database will be auto-created on the next session start via the SessionStart hook.