ur-agent 1.68.3 → 1.68.7

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/CHANGELOG.md CHANGED
@@ -1,5 +1,69 @@
1
1
  # Changelog
2
2
 
3
+ ## 1.68.7
4
+
5
+ - Corrected the ur.ai -> ur.com replacement, which had rewritten identifiers as
6
+ well as URLs. `'ur.ai'` served two roles in this codebase: a domain in links
7
+ such as `https://ur.ai/settings/billing`, and a discriminant value in
8
+ `authTokenSource === 'ur.ai'`, `authMethod = 'ur.ai'` and
9
+ `source: 'ur.ai' as const`. A blanket replace changed both, which typechecked
10
+ and looked internally consistent but altered command availability — one
11
+ command that should have been hidden became visible, caught by
12
+ `commandRegistryIntegrity`. The 39 URLs now point at ur.com; the 11 auth
13
+ discriminants are back to their original values, which also keeps any
14
+ `"ur.ai"` already persisted in a user's auth state matching.
15
+
16
+ ## 1.68.6
17
+
18
+ - An oversized Ollama request now recovers instead of only explaining itself.
19
+ 1.68.3 reported "this request was 19.6 MB" clearly and then left the user to
20
+ run /compact by hand. UR already had both halves of the fix — `isMediaSizeError`
21
+ and `stripImagesFromMessages` — but both were wired to reactive compact's
22
+ retry, and REACTIVE_COMPACT is not compiled into any shipped build, so neither
23
+ was reachable. `isMediaSizeError` and `isMediaSizeErrorMessage` had no callers
24
+ at all outside their own file.
25
+ - On a body-size rejection the request is retried once with images from earlier
26
+ turns removed, keeping the most recent one. Stale attachments are the usual
27
+ bulk and the least valuable part of it: only the newest is normally still
28
+ under discussion.
29
+ - The retry is announced with both sizes and says to re-attach an earlier image
30
+ if it is needed. Dropping a user's attachments silently would be the same
31
+ invisible behaviour being fixed everywhere else.
32
+ - No retry is attempted when there is at most one image-bearing message, since
33
+ it would resend identical bytes and fail identically.
34
+
35
+ ## 1.68.5
36
+
37
+ - Replaced every `ur.ai` reference in `src/` with `ur.com` (202 sites, 0 left).
38
+ Note what these are: most are upstream URL structures carried into the fork —
39
+ `/settings/billing`, `/settings/connectors`, `/upgrade/max`, `/chrome`,
40
+ `/chrome/reconnect`, `/admin-settings/usage`, and desktop auto-update
41
+ redirects like `/api/desktop/darwin/universal/dmg/latest/redirect`. Changing
42
+ the domain does not make those endpoints exist; it moves them to a domain UR
43
+ will control. The features behind them still need a backend or removal.
44
+ - `test/tipsAreReal.test.ts` still forbids `ur.ai` and `ur.com` in tips, since
45
+ the domain is not serving yet. Relax that deliberately once it is.
46
+ - Corrected a comment in that test which claimed both domains "have no DNS
47
+ records". That came from a lookup run where `github.com` and `example.com`
48
+ fail identically — the environment had no DNS — so it was never evidence.
49
+
50
+ ## 1.68.4
51
+
52
+ - The status line now reports subagents running in the current turn:
53
+ `agents: 2 running`, separate from the existing background `tasks:` count.
54
+ `isBackgroundTask()` excludes foreground entries on purpose — it was narrowed
55
+ to stop stale ratios like `tasks: 0/4 active` outliving the work — but nothing
56
+ counted them instead, so while subagents ran the bar said nothing at all,
57
+ which is the one moment the number matters.
58
+ - Counted conservatively on purpose: a pending agent is not reported as running,
59
+ a backgrounded agent is not counted twice across both numbers, a foreground
60
+ shell is not labelled an agent, and zero renders nothing rather than
61
+ `agents: 0`. A wrong number in a status line is worse than a missing one.
62
+ - No tool count was added. A `toolCount` field was drafted and then removed
63
+ rather than shipped unpopulated; a meaningful count of in-flight tool calls
64
+ needs hooks into tool execution, and a static "tools registered" total is
65
+ noise.
66
+
3
67
  ## 1.68.3
4
68
 
5
69
  - `Ollama request failed (400): http: request body too large` now explains