workflow-db-setup

name	workflow-db-setup
description	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.
argument-hint	["--force"]
disable-model-invocation	true

Workflow Toolkit Setup

Verify and initialize the workflow-toolkit environment for the current project.

Instructions

Run the following checks in order. Report each as PASS, FAIL, or SKIP. At the end, show a summary.

If $ARGUMENTS contains --force, re-initialize the database even if it already exists.

1. Check SQLite availability

sqlite3 --version

If this fails, report FAIL and stop — SQLite is required for all features.

2. Check project directory

echo "Project dir: $CLAUDE_PROJECT_DIR"
test -d "$CLAUDE_PROJECT_DIR" && echo "EXISTS" || echo "MISSING"

If $CLAUDE_PROJECT_DIR is empty or doesn't exist, report FAIL and stop.

3. Create directory structure

mkdir -p "$CLAUDE_PROJECT_DIR/.claude/journal/entries"
mkdir -p "$CLAUDE_PROJECT_DIR/.claude/tasks"

Report PASS when directories are created or already exist.

4. Initialize the journal database

DB_PATH="$CLAUDE_PROJECT_DIR/.claude/journal/journal.db"

• If the database already exists and --force was NOT passed, report PASS (already initialized) and show table count:

  sqlite3 "$DB_PATH" "SELECT count(*) FROM sqlite_master WHERE type='table';"
  

• If the database does not exist OR --force was passed, run the init script:

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/init-journal-db.sh"
  

  Then verify tables were created:

  sqlite3 "$DB_PATH" "SELECT name FROM sqlite_master WHERE type='table' ORDER BY name;"
  

  Expected tables: compaction_snapshots, entries, github_issues, sessions, tasks, tool_usage

5. Verify database schema

DB_PATH="$CLAUDE_PROJECT_DIR/.claude/journal/journal.db"
sqlite3 "$DB_PATH" "SELECT name FROM sqlite_master WHERE type='table' ORDER BY name;"

Check all 6 tables exist: compaction_snapshots, entries, github_issues, sessions, tasks, tool_usage. Report PASS if all present, WARN if some missing.

6. Check .gitignore

Check if .claude/journal/journal.db or .claude/ is in the project's .gitignore:

grep -q "\.claude" "$CLAUDE_PROJECT_DIR/.gitignore" 2>/dev/null && echo "COVERED" || echo "NOT COVERED"

If not covered, report WARN and suggest adding .claude/journal/ to .gitignore to avoid committing the database.

7. Check gh CLI (optional)

gh --version 2>/dev/null

If gh is not installed, report SKIP — GitHub issue features won't work but everything else will.

If installed, check auth:

gh auth status 2>&1

Report PASS if authenticated, WARN if not (with suggestion to run gh auth login).

8. Check Node.js (optional)

node --version 2>/dev/null

If not installed, report SKIP — issue sync uses Node.js but it's not required for core features.

Summary

After all checks, show a formatted summary:

Workflow Toolkit Setup
======================
SQLite:          PASS (version X.X.X)
Project dir:     PASS
Directory setup: PASS
Journal DB:      PASS (6 tables, initialized)
Schema:          PASS (all tables present)
.gitignore:      PASS | WARN
gh CLI:          PASS | SKIP | WARN
Node.js:         PASS | SKIP

Status: Ready to use | Action required

If everything is PASS or SKIP, show: "Setup complete. Run /workflow-toolkit:help for usage guide."

If any FAIL, show what needs to be fixed before the plugin will work.