claude_storage 1.0.0

CLI tool for exploring Claude Code filesystem storage
Documentation
# CLI Output Formats

Output format catalog for `claude_storage` CLI. Documents each named rendering mode produced by the `format::` parameter on `.export`.

### Taxonomy

The CLI produces two categories of output:

1. **Export formats** — controlled by the `format::` parameter on `.export`. Each format writes a complete session transcript to a file in a distinct structure. These are the formats documented in this catalog.
2. **Verbosity-driven output** — controlled by `verbosity::` across multiple commands. These are not named formats but progressive detail levels on the same structural output. Documented in [commands.md]../commands.md per-command.

Export formats are mutually exclusive (one per invocation) and produce structurally different files.

### Catalog

| # | Format | Category | Trigger | File | Extension | Machine-Parseable |
|---|--------|----------|---------|------|-----------|-------------------|
| 1 | markdown | Export | `format::markdown` (default when omitted) | [markdown.md]markdown.md | `.md` | No |
| 2 | json | Export | `format::json` | [json.md]json.md | `.json` | Yes |
| 3 | text | Export | `format::text` | [text.md]text.md | `.txt` | No |

**Parameter:** [`format::` in params.md](../params.md#parameter--5-format) | **Type:** [`ExportFormat` in types.md](../types.md#exportformat)

### Rendering Conventions

All three formats share these conventions:

- **Header section** — session ID, storage path, and entry count appear at the top of every export
- **Entry ordering** — entries appear in chronological order (same as JSONL file order)
- **Timestamps** — ISO 8601 strings as stored in the JSONL source (no reformatting)
- **File creation** — output file is created (or overwritten) atomically with `fsync` after write
- **Streaming** — markdown and text formats stream entries through a buffered writer; JSON streams raw JSONL lines. Memory usage scales with single-entry size, not session size