ur-agent 1.13.3 → 1.13.5
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 +56 -0
- package/dist/cli.js +4172 -1328
- package/docs/VALIDATION.md +1 -1
- package/package.json +1 -1
package/CHANGELOG.md
CHANGED
|
@@ -1,5 +1,61 @@
|
|
|
1
1
|
# Changelog
|
|
2
2
|
|
|
3
|
+
## 1.13.5
|
|
4
|
+
|
|
5
|
+
### Added
|
|
6
|
+
- **Headless agent crews.** Added `ur crew` for lead/worker task boards that
|
|
7
|
+
split a goal into subtasks, let worker subagents claim work, and support
|
|
8
|
+
parallel dry-runs or live execution.
|
|
9
|
+
- **Long-horizon goals.** Added `ur goal` for persistent objectives with
|
|
10
|
+
progress notes, workflow or pattern links, and resumable execution.
|
|
11
|
+
- **Model routing.** Added `ur model-route` to recommend the best local Ollama
|
|
12
|
+
model for a task by capability fit.
|
|
13
|
+
- **Trigger bridge.** Added `ur trigger` for parsing GitHub, Slack, and generic
|
|
14
|
+
webhook payloads and optionally launching a headless UR run.
|
|
15
|
+
- **Embeddable SDK surface.** Added `ur sdk` plus `src/sdk` helpers so projects
|
|
16
|
+
can drive UR programmatically and scaffold TypeScript/Python examples.
|
|
17
|
+
|
|
18
|
+
### Changed
|
|
19
|
+
- **Automation and model diagnostics.** Expanded automation scheduling support,
|
|
20
|
+
model-doctor reporting, agent feature scaffolds, and trend coverage for the
|
|
21
|
+
new crew, goal, trigger, model-route, and SDK surfaces.
|
|
22
|
+
|
|
23
|
+
### Verified
|
|
24
|
+
- Rebuilt `dist/cli.js` at 1.13.5 and prepared npm package metadata for
|
|
25
|
+
`ur-agent@1.13.5`.
|
|
26
|
+
|
|
27
|
+
## 1.13.4
|
|
28
|
+
|
|
29
|
+
### Added
|
|
30
|
+
- **Parallel workflow execution.** The declarative workflow executor now runs
|
|
31
|
+
independent ready steps concurrently. `ur workflow run --concurrency <n>`
|
|
32
|
+
caps fan-out width (`1` forces sequential); approval gates and the
|
|
33
|
+
verification/review loop still run sequentially so existing semantics are
|
|
34
|
+
preserved exactly.
|
|
35
|
+
- **Live multi-agent dashboard.** `ur workflow run --live` streams a real-time
|
|
36
|
+
execution board (per-step state, verdicts, and parallel-wave grouping) driven
|
|
37
|
+
straight from executor events — the live counterpart to the post-hoc
|
|
38
|
+
`ur agent-inspect` timeline.
|
|
39
|
+
- **More orchestration patterns.** Added `concurrent` (parallel fan-out/fan-in),
|
|
40
|
+
`handoff` (triage → specialist), and `debate` (propose → critique → moderate
|
|
41
|
+
loop) alongside PEER and DOE, completing the
|
|
42
|
+
sequential/concurrent/handoff/group-chat/manager-loop taxonomy. Patterns can
|
|
43
|
+
now compile into true DAGs (stage `dependsOn`), and `ur route` recommends the
|
|
44
|
+
new patterns.
|
|
45
|
+
- **Public eval harness.** New `ur eval init|list|validate|run|report` runs
|
|
46
|
+
replayable, gradeable cases (contains / notContains / regex / verdict / length
|
|
47
|
+
checks) grouped by category, with deterministic offline grading and a
|
|
48
|
+
`--dry-run` mode. Seeds a starter suite under `.ur/evals/`.
|
|
49
|
+
- **A2A delegation tokens.** Signed (HMAC-SHA256), attenuated, expiring
|
|
50
|
+
capability tokens scoped to specific skills and bound to one audience.
|
|
51
|
+
`ur a2a token mint|verify`, enforced by the A2A task server (static bearer
|
|
52
|
+
still works), advertised in the Agent Card `securitySchemes`.
|
|
53
|
+
|
|
54
|
+
### Verified
|
|
55
|
+
- Full test suite green (workflow parallelism, patterns, live board, eval
|
|
56
|
+
grading, and delegation each covered); `tsc --noEmit` clean; `dist/cli.js`
|
|
57
|
+
rebuilt and smoke-tested through the CLI.
|
|
58
|
+
|
|
3
59
|
## 1.13.3
|
|
4
60
|
|
|
5
61
|
### Added
|