DMs
birdclaw treats DMs as first-class content: full-text indexed, multi-account, and filterable by sender context. The web app's DMs lane uses the same query model you get from the CLI.
#List conversations
3 dms list 0 1 10 2
5 dms list 0 1 500 2 90 3 influence 4 dms list is the no-query view onto conversations and recent events. It is optimized for agent and operator filtering — you do not need a search query to walk the inbox.
Flags:
--refresh— refresh live DMs throughbirdbefore listing--cache-ttl <seconds>— tune freshness--participant <handle-or-id>--min-followers <n>/--max-followers <n>--min-influence-score <n>/--max-influence-score <n>--sort recent|influence--replied/--unreplied--account <name>--limit <n>
#Sync
Refresh live direct messages through bird and merge into the canonical conversation/message tables:
3 dms sync 0 50 1 2 Flags:
--account <account-id>--limit <n>--refresh— force a live fetch--cache-ttl <seconds>
Sync is idempotent — re-running merges new events without disturbing already-imported message bodies.
#Search
2 search dms 0 1
5 search dms 0 1 1000 2 120 3 influence 4
4 search dms 0 1 @someone 2 3 Same FTS5 backbone as tweet search, with the DM-specific filters layered on top. See Search for the full flag list.
#Influence score
Influence is a derived ranking signal that starts with follower count and folds in:
- verified status
- prior reply / DM history with the active account
- follower-to-following ratio
- account age
- block / mute history
It is intentionally simple. The goal is to bucket noisy inboxes ("strangers with no follow signal" vs "people you actually talk to"), not to produce a global ranking.
When triaging a quiet day, sort by influence to surface higher-context conversations first:
4 dms list 0 1 influence 2 20 3 When triaging a noisy day, hide low-influence senders entirely:
4 dms list 0 1 80 2 20 3 #Reply
1 compose dm dm_003 0 Replies use the active live transport (auto by default). Without a working transport, the command fails fast with exit code 4 rather than recording a half-state local row.
#DM mode on archive import
Twitter archives include full DM history but the JSON is awkward. By default, import archive runs --dm-mode full so message bodies land in SQLite and become FTS5-searchable. Pass --dm-mode metadata if you only want conversation skeletons (faster import, no body content).
2 import archive ~/Downloads/twitter-archive.zip 0 metadata 1
2 import archive ~/Downloads/twitter-archive.zip 0 full 1 #Web UI
The DMs lane uses a two-column layout:
- left: conversation list, filterable by participant, follower count, influence, reply state
- right: detail view with sender bio, follower count, and influence visible in the header so you do not have to dig
Theme follows the system / light / dark switcher with an animated transition.