Commands overview

Every subcommand is listed here with a one-line summary. Click through for flags, examples, and the JSON shape.

Global flag: --color auto|always|never (respects NO_COLOR).

Reports span Claude Code, OpenAI Codex, GitHub Copilot (CLI and VS Code), Pi, and OpenClaw by default. Every report accepts the shared filter flags — --provider, --model, --since/--until, --on-disk-only. --project is a separate per-command filter (it has its own column in the matrix below).

Read-only reports

Command	What it does
summary	Dashboard — sessions, cost, top projects/tools, model mix.
sessions	List sessions grouped by project (all providers).
session	Drill into one session: spend, files, tools, PRs, turns.
cost	Token usage and approximate cost per project (or per session).
search	Full-text search across session messages (FTS5), with JSON hits.
tools	Tool-usage frequency, optionally per session.
models	Call counts, token usage, and cost per model.
turns	Per-turn timing (avg / p50 / p95 / max).
prs	Sessions linked to pull requests.
files	Most frequently modified files across sessions.
providers	Provider roots, sync status, retention, and parse diagnostics.
timeline	Daily or weekly usage trend.
budget	Monthly budget burn and projection.
activity	Recent sessions, PRs, files, and slow projects.

Flag support matrix

Not every report accepts every flag. Consult the per-command page for exact usage; the matrix below is the quick overview.

Command	filters	--project	--limit	--json	--no-index
summary	✓	—	—	✓	✓
sessions	✓	✓	✓	✓	✓
session	—	✓	—	✓	✓
cost	✓	✓	✓	✓	✓
search	✓	✓	✓	✓	✓
tools	✓	✓	✓	✓	✓
models	✓	✓	—	✓	—
turns	✓	✓	✓	✓	—
prs	✓	✓	✓	✓	—
files	✓	✓	✓	✓	—
providers	✓	—	—	✓	—
timeline	✓	—	✓	✓	—
budget	✓	—	—	✓	—
activity	✓	—	✓	✓	—

Notes:

• filters = the shared --provider / --model / --since / --until / --on-disk-only set.
• search supports --json; case-sensitive queries still fall back to a file scan automatically.
• turns, prs, files, and models derive their data from the index only — there's no file-scan fallback path, so --no-index isn't accepted.
• summary is a whole-index dashboard spanning all providers; the shared filters apply, but there is no row limit.
• summary also accepts --plan <api|flat-monthly:USD> to reframe the cost section for flat-fee subscribers. See summary.

Actions

Command	What it does
export	Dump a session transcript as Markdown or JSON.
watch	Tail Claude Code's --debug-file log in real time.
index	Manage the SQLite index — force sync or full rebuild.
update	Self-update claudex, or print the right upgrade recipe for Nix / cargo / brew installs.
completions	Generate shell completion scripts.
skills	Generate or install the claudex agent skill for Claude Code, Codex, Pi, or OpenClaw.

Conventions

• Project filter. --project foo matches any session whose decoded project path contains foo. Worktree sessions roll up to their parent project.
• Session selector. Commands that take a session (currently session and export) match on session-ID prefix or project name.
• Limit default. Most commands default to --limit 20. Pass a higher number for more rows.
• Thousands separators. Token counts and message counts render as 326,297. Costs render as $12,345.67, falling back to $0.0042 for sub-cent values.

Quick alphabetical index

• activity
• budget
• completions
• cost
• export
• files
• index
• models
• providers
• prs
• search
• session
• sessions
• skills
• summary
• timeline
• tools
• turns
• update
• watch