← Back to Docs

Export

Polylogue does not have a separate export command. Export is query selection plus a read projection plus a renderer. Use read for one selected session and read --all when the query result set itself is the export target.

Single Session

Read one session by ID with a root filter that narrows the archive to a single match:

polylogue --id claude-ai:abc123 read

Specify format:

polylogue --id claude-ai:abc123 read --format json
polylogue --id claude-ai:abc123 read --format yaml

Query-Set Reads

Read every session matching a filter chain with read --all:

polylogue --origin claude-code-session --since 2026-01 find 'origin:claude-code-session' then read --all
polylogue --tag important find 'tag:important' then read --all --format markdown
polylogue --has-tool-use find "refactor" then read --all --format ndjson

Piping

read --all --format ndjson emits one JSON line per session, suitable for piping:

polylogue --origin claude-code-session find 'origin:claude-code-session' then read --all --format ndjson | jq '.title'
polylogue --since "last month" find 'since:"last month"' then read --all --format ndjson | wc -l

Export Formats

These are the formats supported by the standard query-set export path. Default output varies by view and cardinality, so specify a format in automation. Other views have their own contracts; use polylogue read --views --format json to inspect the current view-specific formats before relying on one.

Set format with --format / -f:

polylogue --id <id> read --format json
polylogue --since yesterday find 'since:yesterday' then read --all --format yaml

Content Blocks

Reads include all selected content blocks: text, thinking blocks, tool use, tool results, images, code blocks, and document references. Use query-unit expressions or explicit read views to narrow what is selected before rendering.

Repeatable Local Packages

For repeatable local demo/export bundles, describe ordinary read artifacts in a JSON/YAML package and render it through devtools workspace read-package. Artifact-level projection policy lives under projection, and renderer/layout policy lives under render. Both map to ordinary polylogue read flags:

{
  "version": 1,
  "artifacts": [
    {
      "name": "dialogue-json",
      "view": "dialogue",
      "format": "json",
      "path": "dialogue.json",
      "projection": {
        "max_tokens": 120
      }
    },
    {
      "name": "spec-json",
      "view": "temporal,chronicle",
      "format": "json",
      "path": "spec.json",
      "spec": true,
      "render": {
        "fields": "selection,projection,render"
      }
    }
  ]
}

This is still composition over polylogue read: the package renderer plans normal read commands such as --view dialogue --format json --max-tokens 120 or --spec --fields selection,projection,render.

Sharing Considerations

• Rendered output can contain full message content including tool inputs/outputs and thinking blocks. Review before sharing.
• Cost estimates are approximate; API-reported token counts are exact where available.
• Attachment references are preserved but binary blobs are not rendered.
• Use query-unit filters or a narrower read view to produce cleaner output for non-technical audiences.