ur-agent 1.61.0 → 1.61.2
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 +1,34 @@
|
|
|
1
1
|
# Changelog
|
|
2
2
|
|
|
3
|
+
## 1.61.2
|
|
4
|
+
|
|
5
|
+
- Fixed `ur agent-inspect --costs` reporting nothing, always. It resolved the
|
|
6
|
+
*live* session, but every `ur` invocation mints a new session id, so bare it
|
|
7
|
+
pointed at a session created milliseconds earlier that had by definition
|
|
8
|
+
spawned nothing — two consecutive runs in the same directory produced two
|
|
9
|
+
different empty session ids and no data either time. It now prefers the live
|
|
10
|
+
session when it has transcripts (the `/agent-inspect` case, inside a session
|
|
11
|
+
that already fanned out) and otherwise falls back to the most recent session
|
|
12
|
+
in the project that does.
|
|
13
|
+
- The transcript writer was never at fault: Agent-tool subagents do write
|
|
14
|
+
`agent-*.jsonl`. Only the reader's choice of session was wrong.
|
|
15
|
+
|
|
16
|
+
## 1.61.1
|
|
17
|
+
|
|
18
|
+
- Fixed secondary model queries failing on any Ollama setup whose session model
|
|
19
|
+
is not `qwen2.5-coder:7b`. `getSmallFastModel()` fell back to the compiled
|
|
20
|
+
default when auto-routing was off or no model list had been discovered, so a
|
|
21
|
+
user running `kimi-k2.7-code:cloud` had every WebFetch summarisation and
|
|
22
|
+
classifier call rejected with "Model qwen2.5-coder:7b is not available for
|
|
23
|
+
provider ollama". A fallback to a model the user may never have pulled is a
|
|
24
|
+
guaranteed failure; it now falls back to the session model, which is
|
|
25
|
+
guaranteed to exist.
|
|
26
|
+
- WebFetch no longer returns a failed summarisation as if it were the page.
|
|
27
|
+
The error text was passed through as the fetch result, so the evidence ledger
|
|
28
|
+
recorded "API Error: Provider ollama ..." as the content of example.com and a
|
|
29
|
+
`--check` against real page text correctly found nothing. It now throws, and
|
|
30
|
+
the failure is visible as a failed tool call.
|
|
31
|
+
|
|
3
32
|
## 1.61.0
|
|
4
33
|
|
|
5
34
|
- Added `ur selftest`, end-to-end drills for the gap that produced every
|