birdclaw Files

Highlights

• Keep the full post. Preserve long-form X Note Tweets across sync, search, and backups, and expand them directly in the timeline.
• Jump from the archive to X. Open posts, quoted tweets, and conversations on X from their archive cards, including in read-only deployments.
• Read saved records from the CLI. Look up individual tweets, threads, and DM conversations in the selected account, with JSON output for scripts.
• Spend less time waiting. Start the CLI faster and reduce repeated profile and history work during sync and identity searches.

Upgrade notes

• Note Tweet storage adds SQLite migration 11. Prepare existing read-only archive snapshots with writable initialization before serving the updated application.
• Backups remain interoperable with schema-8 readers and writers. Older writers retain full Note Tweet text but omit the optional Note Tweet marker. (#134)

Fixes and improvements

• Preserve full Note Tweet text and entities through live ingestion, search, archive and backup merges, and expandable timeline rendering. (#134 — thanks @eferm)
• Add Open on X links to feed cards, parent and quoted tweets, and expanded conversations. Reposts open the original tweet when its ID is known. (#189)
• Show “no threads” in feed cards when no other archived replies are found, instead of opening an empty conversation panel. (#207)
• Add account-scoped show tweet, show thread, and show dm commands with JSON output, plus db vacuum to reclaim unused SQLite space. (#192)
• Honor --json for parser and uncaught runtime failures and server startup, expose global options in nested help, and validate the server port before startup. (#190)
• Validate numeric CLI options before account lookups or command work, preventing negative limits from requesting unlimited DM results and rejecting invalid score thresholds. (#192)

Performance

• Reduce CLI startup time by bundling the used Effect modules while keeping other direct dependencies external. (#193)
• Reconcile each tweet author once per ingested payload, avoiding repeated profile and history work for posts and included references by the same author. (#200)
• Bound profile-history reads in SQLite, omit unused raw payloads, and index historical follower membership, follow events, and list owners by profile to avoid repeated full-table scans. (#194, [#195])

Developer tools and maintenance

• Add opt-in numeric CLI timing summaries for elapsed time, process CPU, and database work without exposing queries or archive contents. (#191)
• Consolidate backup merge declarations, API data types, transport pagination, CLI streams, and input-error handling while preserving command output and archive compatibility. (#196, [#202], [#203])
• Give digest, discussion, and profile analysis one lifecycle for request construction, persistence, cache replay, and completion events; retain refresh, citation, and streaming behavior. (#206)
• Share parsing of profile URL entities and identity index insertion rules, and consolidate automatic backup setup and result handling while retaining freshness checks and sync behavior. (#204)
• Stabilize backup integration tests under slow filesystem I/O with focused fixtures and independent invalid-path cases, retaining real durability and publication checks. Exercise local JSON commands and failures through installed Node and Bun packages. (#190, [#205])
• Update Hono to 4.13.7 and the development/CI Node pin to 26.8.2; installed-package checks retain the Node 26.5.1 minimum. (#201)

Packages and verification

• npm 0.13.0
• Registry tarball
• Published: 2026-09-13T04:11:49.430Z
• Release CI: Bun, Node, installed package, and browser tests
• All changes since 0.12.6

Integrity: sha512-kNhvz+Ikl7/gP3bw7hLSARFjib/hiumzNcbgzdJ6+3AE+PlLRQuuXntunx9y15zcx+U8GJMC67TTOj1YyIVDSw==