ur-agent 1.58.0 → 1.59.0

package/docs/USAGE.md

@@ -339,6 +339,15 @@ UR includes slash commands and CLI subcommands for common workflows:
   raw video/trace requires selector redaction to be disabled
 - `ur trigger ...` to parse GitHub/Slack webhook payloads and optionally launch a headless run
 - `ur agent-templates ...`, `ur agent-task ...`, `ur agent-inspect`, `ur agent-features`, and `ur agent-trends` for agent template, PR handoff, timeline, and coverage utilities
+- `ur sources` lists every untrusted block that entered the session (web fetches,
+  MCP results) with its source, size, digest and any injection signals.
+  `ur sources --check "<span>"` reports which source contains a span, or states
+  that it was not grounded in anything fetched. The ledger is per-process and
+  in-memory by design.
+- `ur grade-trajectory --file <transcript.jsonl> --min-score 70` grades a run on
+  tool choice, verification, instruction compliance, safety and efficiency, and
+  exits non-zero below the threshold so it can gate CI. Rules are deterministic;
+  no model judges another model.
 - `ur agent-inspect --costs [subagentsDir]` breaks a fan-out down per agent.
   Subagent turns are never in the parent transcript — they are written to
   `{sessionId}/subagents/agent-{agentId}.jsonl` — so without this a fan-out

package/docs/providers.md

@@ -47,7 +47,14 @@ multimodal input, external CLI boundary, and sandbox scope:
 | Antigravity | subscription | subscription-cli | yes | no | no | no† | UR-run tools/output only† | `subscription-cli:antigravity` | official Antigravity CLI login, where supported |
 
 \* Ollama forwards images only to models that advertise vision support;
-unsupported models get a text placeholder instead of the image. This applies to
+a model whose advertised capabilities omit `vision` gets a text placeholder
+instead. A model that advertises nothing at all is treated as *unknown*, not
+unsupported: the image is still sent, and the note says support could not be
+confirmed rather than asserting the model is blind. That distinction matters —
+`/api/show` returns no capabilities for several cloud-suffixed models, and
+reporting that as "no vision support" sends you to change models for no
+reason. Resolution lives in `src/utils/model/visionCapability.ts` and is shared
+by the adapter, `ur model-doctor` and the model router. This applies to
 images returned by tools as well as images you paste: a `Computer` screenshot or
 any other image-bearing tool result is extracted from the tool message and sent
 as `images` on the following user message, because Ollama renders `images`

package/documentation/index.html

@@ -45,7 +45,7 @@
     <main id="content" class="content">
       <header class="topbar">
         <div>
-          <p class="eyebrow">Version 1.58.0</p>
+          <p class="eyebrow">Version 1.59.0</p>
           <h1>UR-Nexus Documentation</h1>
           <p class="lead">A practical, tutorial-style reference for installing, configuring, automating, extending, and operating UR-Nexus.</p>
         </div>

package/extensions/jetbrains-ur/build.gradle.kts

@@ -7,7 +7,7 @@ plugins {
 }
 
 group = "dev.urnexus"
-version = "1.58.0"
+version = "1.59.0"
 
 repositories {
     mavenCentral()

package/extensions/vscode-ur-inline-diffs/package.json

@@ -2,7 +2,7 @@
   "name": "ur-inline-diffs",
   "displayName": "UR Inline Diffs",
   "description": "Review, apply, and reject UR inline diff bundles from .ur/ide/diffs inside VS Code.",
-  "version": "1.58.0",
+  "version": "1.59.0",
   "publisher": "ur-nexus",
   "engines": {
     "vscode": "^1.92.0"

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ur-agent",
-  "version": "1.58.0",
+  "version": "1.59.0",
   "description": "UR-Nexus — autonomous engineering workflow engine (plan, execute, test, verify, document, benchmark, reproduce)",
   "type": "module",
   "packageManager": "bun@1.3.14",