@floomhq/signaldash 0.44.1 → 0.44.2

package/README.md

@@ -121,7 +121,7 @@ setup installs both from the same pinned npm package
 the human chose to execute:
 
 ```bash
-npx -y @floomhq/signaldash@0.44.1 <invite-code>
+npx -y @floomhq/signaldash@0.44.2 <invite-code>
 ```
 
 Run that command in a terminal, not in an agent chat. Do not ask an agent to
@@ -411,6 +411,13 @@ re-read itself fails, the send is refused with
 and unknown is not permission. All three refusals happen before anything is
 reserved, so none of them costs send budget.
 
+If the provider's chat lookup itself hangs, a live WhatsApp send proves the
+recipient from the chat's attendees and newest message instead: one counterpart
+whose WhatsApp id equals the thread's, in this account and this chat. Every guard
+above then runs on that record. Without that proof the send is refused with
+`503 recipient_unverified`, before anything is reserved or written. Groups, chats
+known only by an `@lid` id and empty threads cannot be proven this way.
+
 A provider answer this process cannot read counts as a re-read that failed, not
 as an empty conversation. Only the two documented list envelopes are hashed;
 anything else yields no hash at all, so a read that hits a drifted shape records
@@ -995,3 +1002,20 @@ resolves those credentials itself, in the child process, from the file named by
 `SIGNALDASH_TRANSCRIBE_AWS_ENV_FILE`. The gateway process forwards only
 configuration, never a secret. That backend reports `language` as `null`: it is
 asked for the transcript text and nothing else.
+
+### Recover an expired operator session
+
+`POST /admin/recover_session` is a local-admin-only maintenance operation. It
+accepts exact `source_session_hash` and `target_session_hash` identifiers,
+`confirm: true`, and a written `reason`. The source must be expired but not
+revoked; the target must be active, empty, and belong to the same `data_ref`.
+Every source channel must still resolve to the exact ready provider account,
+and no other live person may own it. Recovery stores an audit record on the
+target and leaves the old session expired. It does not replay scheduled writes,
+copy tokens, change budgets, or weaken the public login/claim guards.
+
+The normal local admin credential is required and Cloudflare-forwarded requests
+are refused. Operators should resolve identifiers inside the maintenance process
+from the native session store; never print bearer tokens or put them in shell
+arguments. After recovery, verify each requested channel through SignalDash
+before any previously approved action proceeds.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@floomhq/signaldash",
-  "version": "0.44.1",
+  "version": "0.44.2",
   "description": "Secure LinkedIn, WhatsApp, and email access for AI agents",
   "type": "module",
   "bin": {

package/skills/signaldash/SKILL.md

@@ -1321,6 +1321,10 @@ same fact: no answer about whether the conversation moved. In that case a `428
 read_before_send_required` on the next attempt is expected too, because a read
 of an unrecognised shape records no authorization.
 
+A `503 recipient_unverified` on a WhatsApp send means the provider's chat lookup
+hung and the recipient could not be proven another way. Nothing was sent. Retry
+later; never look for a similarly named chat to send into instead.
+
 A confirmed send advances the stored proof. Distinct consecutive messages can
 therefore pass without a fresh read when the only movement is this account's
 own outbound. Every send still re-reads the provider thread, and a counterparty