ur-agent 1.47.1 → 1.49.0

package/documentation/app.js

@@ -14,13 +14,19 @@ const featureGroups = [
   {
     title: 'Agent platform',
     tags: ['spec', 'workflow', 'pattern', 'crew', 'goal', 'worktree'],
-    text: 'Spec-driven development, durable workflows, collaboration patterns, parallel crews, long-horizon goals, live execution boards, and resumable checkpoint state.',
-    commands: ['ur spec', 'ur workflow', 'ur pattern', 'ur crew', 'ur goal', 'ur worktree'],
+    text: 'Spec-driven development, durable workflows, collaboration patterns, parallel crews, long-horizon goals, live execution boards, multi-repository coordination, and resumable checkpoint state.',
+    commands: ['ur spec', 'ur workflow', 'ur pattern', 'ur crew', 'ur goal', 'ur worktree', 'ur workspace'],
+  },
+  {
+    title: 'Frontier agent operations',
+    tags: ['managed', 'steering', 'playbooks', 'Agentic CI', 'trajectory', 'desktop QA'],
+    text: 'Managed PASS-and-safe-branch fan-out, idempotent live steering, approval-gated learned playbooks, citation-validated memory, patch-only Agentic CI, privacy-minimized trajectory gates, redaction-safe Electron QA, durable tool-free side chats, dependency-aware workspace coordination, and verified anonymous model judging.',
+    commands: ['ur cloud', 'ur bg steer', 'ur learn playbooks', 'ur context-pack memory', 'ur agent-ci', 'ur eval gate', 'ur desktop-qa', '/btw', 'ur workspace', 'ur arena --judge hybrid'],
   },
   {
     title: 'Judging, escalation, and repair',
     tags: ['oracle', 'arena', 'CI', 'artifacts'],
-    text: 'Capability-aware fast/oracle model routing, best-of-N agent judging, test-first execution loops, self-healing CI loops, and reviewable artifacts for diffs, test runs, plans, and feedback.',
+    text: 'Capability-aware fast/oracle routing, proof-gated best-of-N judging, test-first and self-healing CI loops, patch-only Agentic CI, and reviewable artifacts for diffs, test runs, plans, attachments, and feedback.',
     commands: ['ur escalate', 'ur arena', 'ur test-first', 'ur ci-loop', 'ur artifacts', 'ur bg'],
   },
   {
@@ -38,14 +44,14 @@ const featureGroups = [
   {
     title: 'Knowledge and memory',
     tags: ['retrieval', 'provenance', 'citations'],
-    text: 'Durable memory, semantic memory, curated knowledge sources, lexical or embedding retrieval, citations, claim ledgers, and research graph primitives.',
-    commands: ['ur knowledge', 'ur semantic-memory', 'ur context-pack remember', 'ur claim-ledger'],
+    text: 'Durable memory, semantic memory, curated knowledge sources, lexical or embedding retrieval, digest-validated file/run citations, explicitly unverifiable user/web citations, claim ledgers, and research graph primitives.',
+    commands: ['ur knowledge', 'ur semantic-memory', 'ur context-pack remember --cite-file', 'ur context-pack memory revalidate', 'ur claim-ledger'],
   },
   {
     title: 'Evaluation and verification',
-    tags: ['evals', 'review', 'QA'],
-    text: 'Replayable eval suites, self-review PR gate, browser QA fixtures, verifier reminders, test-first traces, trace inspection, reviewable artifacts, and subagent timelines.',
-    commands: ['ur eval', 'ur test-first', 'ur agent-task', 'ur browser-qa', 'ur artifacts', '/verify', '/trace'],
+    tags: ['evals', 'trajectory', 'review', 'desktop QA'],
+    text: 'Isolated replayable evals with fail-closed trajectory gates, self-review and patch-state binding, browser/Electron QA fixtures, verifier reminders, trace inspection, safe attachment delivery, and subagent timelines.',
+    commands: ['ur eval gate', 'ur agent-ci', 'ur desktop-qa', 'ur test-first', 'ur agent-task', 'ur browser-qa', 'ur artifacts', '/verify', '/trace'],
   },
   {
     title: 'Interoperability',
@@ -107,8 +113,8 @@ const commands = [
     name: 'bg',
     category: 'Agent Platform',
     aliases: ['background-agent'],
-    summary: 'Run and manage detached local background agents with optional worktrees and PR creation.',
-    examples: ['ur bg run "fix the flaky parser test" --worktree', 'ur bg list --json', 'ur bg status <id> --json'],
+    summary: 'Run and manage detached local background agents with optional worktrees, cancellation, and bounded idempotent live steering.',
+    examples: ['ur bg run "fix the flaky parser test" --worktree', 'ur bg steer <id> --message "cover timeouts" --request-id timeout-1', 'ur bg status <id> --json', 'ur bg kill <id>'],
   },
   {
     name: 'worktree',
@@ -163,8 +169,8 @@ const commands = [
     name: 'arena',
     category: 'Agent Platform',
     aliases: ['best-of'],
-    summary: 'Run multiple agents on the same task in isolated worktrees, score their diffs, and optionally apply the winning patch.',
-    examples: ['ur arena "implement a debounce helper" --agents 2 --dry-run', 'ur arena "implement the rate limiter" --agents 3', 'ur arena "fix the parser" --agents 3 --apply'],
+    summary: 'Run isolated candidates, admit only PASS + safe + verified bounded patches, and judge them deterministically or through a strict anonymous model/hybrid decision.',
+    examples: ['ur arena "implement a debounce helper" --agents 2 --dry-run', 'ur arena "implement the rate limiter" --agents 3 --judge hybrid --verify "bun test"', 'ur arena "fix the parser" --agents 3 --apply'],
   },
   {
     name: 'artifacts',
@@ -205,8 +211,8 @@ const commands = [
     name: 'context-pack',
     category: 'Knowledge',
     aliases: ['project-manifest', 'ctx-pack'],
-    summary: 'Summarize repository architecture from manifests and instructions, record task memory, and compress old context under `.ur/context`.',
-    examples: ['ur context-pack scan', 'ur context-pack remember --decision "Use package scripts first"', 'ur context-pack remember --command "bun run typecheck"', 'ur context-pack compress'],
+    summary: 'Summarize repository architecture and maintain source-labelled task memory with digest-validated file/run citations and explicit user/web provenance.',
+    examples: ['ur context-pack scan', 'ur context-pack remember --decision "Keep streaming" --cite-file src/parser.ts --lines 20:48', 'ur context-pack memory revalidate', 'ur context-pack memory search --query "streaming"', 'ur context-pack compress'],
   },
   {
     name: 'config',
@@ -261,8 +267,8 @@ const commands = [
     name: 'eval',
     category: 'Verification',
     aliases: ['evals'],
-    summary: 'Create, validate, run, and report public agent eval suites with deterministic grading plus execution metrics and a local HTML dashboard.',
-    examples: ['ur eval init', 'ur eval list', 'ur eval validate starter', 'ur eval run starter --dry-run', 'ur eval run starter --metrics --json', 'ur eval report starter --dashboard', 'ur eval dashboard'],
+    summary: 'Run isolated agent evals, grade privacy-minimized control-flow trajectories, and enforce fail-closed pass, trajectory, test, cost, duration, or regression gates.',
+    examples: ['ur eval init', 'ur eval validate starter', 'ur eval run starter --metrics --json', 'ur eval gate starter --min-pass-rate 1 --min-trajectory-score 0.9', 'ur eval report starter --dashboard'],
   },
   {
     name: 'goal',
@@ -443,8 +449,8 @@ const commands = [
     name: 'cloud',
     category: 'Agent Platform',
     aliases: [],
-    summary: 'Run detached best-of-N tasks, inspect their results, and explicitly apply a selected winner.',
-    examples: ['ur cloud run "speed up parser" --attempts 3', 'ur cloud list', 'ur cloud show <id>', 'ur cloud apply <id>'],
+    summary: 'Run local best-of-N or managed fan-out. Managed selection requires explicit PASS plus a safe review branch, supports cancellation/steering, and never fetches or merges.',
+    examples: ['ur cloud run "speed up parser" --runner managed --environment <id> --attempts 3', 'ur cloud sync', 'ur cloud steer <id> --message "preserve the API" --request-id review-1', 'ur cloud cancel <id>'],
   },
   {
     name: 'recipe',
@@ -497,12 +503,12 @@ const slashGroups = [
   {
     title: 'Project intelligence',
     items: ['/dna', '/project', '/context-pack', '/workspace', '/read', '/search', '/index', '/analyze', '/summarize'],
-    text: 'Understand the workspace, build indexes, and read or summarize project files.',
+    text: 'Understand one repository or coordinate a dependency-aware, one-writer-per-repository workspace; build indexes and read or summarize project files.',
   },
   {
     title: 'Agents and orchestration',
-    items: ['/agents', '/agent-templates', '/spec', '/workflow', '/pattern', '/crew', '/goal', '/arena', '/route', '/role-mode', '/bg'],
-    text: 'Manage agents, install role modes, run specs and workflows, and coordinate multi-agent work.',
+    items: ['/agents', '/agent-templates', '/spec', '/workflow', '/pattern', '/crew', '/goal', '/arena', '/route', '/role-mode', '/bg', '/btw'],
+    text: 'Manage agents, run specs/workflows and verified arenas, steer background work, and keep durable tool-free side chats alongside the main task.',
   },
   {
     title: 'Agent skills',
@@ -511,13 +517,13 @@ const slashGroups = [
   },
   {
     title: 'Memory and evidence',
-    items: ['/memory', '/remember', '/forget', '/knowledge', '/semantic-memory', '/claim-ledger', '/evidence', '/graph'],
-    text: 'Persist useful facts, search memory and knowledge, and keep claim provenance.',
+    items: ['/memory', '/remember', '/forget', '/context-pack', '/knowledge', '/semantic-memory', '/claim-ledger', '/evidence', '/graph'],
+    text: 'Persist useful facts, validate cited project memory, search knowledge, and keep claim provenance.',
   },
   {
     title: 'Automation and evals',
-    items: ['/automation', '/trigger', '/ci-loop', '/eval', '/browser-qa', '/actions', '/stability'],
-    text: 'Run recurring prompts, webhook-triggered runs, self-healing CI loops, browser smoke checks, evals, and stability diagnostics.',
+    items: ['/automation', '/trigger', '/ci-loop', '/agent-ci', '/eval', '/browser-qa', '/desktop-qa', '/actions', '/stability'],
+    text: 'Run recurring prompts, patch-only Agentic CI, self-healing loops, browser/Electron QA, trajectory-gated evals, and stability diagnostics.',
   },
   {
     title: 'Models, tools, and interop',
@@ -624,8 +630,33 @@ const projectFiles = [
   },
   {
     title: '.ur/context/',
-    text: 'Architecture summaries, task memory JSONL, and compressed context for decisions, constraints, commands, diffs, and notes.',
-    example: 'ur context-pack compress',
+    text: 'Architecture summaries, citation-validated task memory JSONL, and compressed context for decisions, constraints, commands, diffs, and notes.',
+    example: 'ur context-pack memory revalidate',
+  },
+  {
+    title: '.ur/cloud/',
+    text: 'Private bounded manifests, steering receipts, cursors, and redacted logs for local or managed fan-out.',
+    example: 'ur cloud sync',
+  },
+  {
+    title: '.ur/learning/',
+    text: 'Proof-backed playbook candidates plus the private archive created when an unchanged approved workflow is disabled.',
+    example: 'ur learn playbooks list',
+  },
+  {
+    title: '.ur/agentic-ci/',
+    text: 'Agentic CI policies, isolated temporary worktree state, and hash-addressed patch-only run manifests.',
+    example: 'ur agent-ci validate default',
+  },
+  {
+    title: '.ur/desktop-qa/',
+    text: 'Electron QA fixtures and bounded run evidence; selector-redacted fixtures cannot enable raw trace or video.',
+    example: 'ur desktop-qa validate .ur/desktop-qa/fixtures/smoke.json',
+  },
+  {
+    title: '.ur/workspaces/',
+    text: 'Pinned multi-repository specifications and durable dependency-aware run state.',
+    example: 'ur workspace validate checkout',
   },
   {
     title: '.ur/repo-edit/',
@@ -682,8 +713,48 @@ const examples = [
   },
   {
     title: 'Best-of-N agent run',
-    text: 'Let isolated agents attempt the same task and surface the strongest diff.',
-    code: 'ur arena "implement a debounce helper" --agents 2 --dry-run\nur arena "fix the parser" --agents 3 --apply',
+    text: 'Admit only PASS + safe + verified bounded patches, then use deterministic or strict anonymous model judging.',
+    code: 'ur arena "implement a debounce helper" --agents 2 --dry-run\nur arena "fix the parser" --agents 3 --judge hybrid --verify "bun test"\nur arena "fix the parser" --agents 3 --apply',
+  },
+  {
+    title: 'Managed fan-out and steering',
+    text: 'Launch isolated managed candidates, steer active sessions idempotently, and review only explicit PASS candidates with safe branches.',
+    code: 'ur cloud environments\nur cloud run "repair the parser" --runner managed --environment <id> --attempts 3\nur cloud steer <task-id> --message "preserve the API" --request-id review-1\nur cloud sync\nur cloud show <task-id>',
+  },
+  {
+    title: 'Approval-gated learned playbook',
+    text: 'Mine verified repetitions, inspect the candidate, explicitly promote it, and archive it safely when disabled.',
+    code: 'ur learn playbooks mine --min-runs 3\nur learn playbooks show <id>\nur learn playbooks approve <id> --name parser-repair\nur learn playbooks run <id>\nur learn playbooks disable <id>',
+  },
+  {
+    title: 'Cited project memory',
+    text: 'Bind durable decisions to a file excerpt or run artifact and revalidate before reuse.',
+    code: 'ur context-pack remember --decision "Keep streaming" --cite-file src/parser.ts --lines 20:48\nur context-pack memory revalidate\nur context-pack memory search --query "streaming"',
+  },
+  {
+    title: 'Patch-only Agentic CI',
+    text: 'Generate the pinned read-only workflow and emit a post-check, state-bound patch artifact for separate review.',
+    code: 'ur agent-ci init default\nur agent-ci validate default\nur agent-ci workflow default --force\nur agent-ci run default --event "$GITHUB_EVENT_PATH" --event-name "$GITHUB_EVENT_NAME" --output-dir "$RUNNER_TEMP/ur-agentic-ci"',
+  },
+  {
+    title: 'Trajectory-gated eval',
+    text: 'Run cases in isolated worktrees and fail CI when outcome or control-flow evidence misses a declared threshold.',
+    code: 'ur eval run starter\nur eval gate starter --min-pass-rate 1 --min-trajectory-score 0.9 --min-test-pass-rate 1',
+  },
+  {
+    title: 'Redaction-safe Electron QA',
+    text: 'Validate bounded fixtures before launch. Selector masks require raw video and trace to remain disabled.',
+    code: 'ur desktop-qa init\nur desktop-qa validate .ur/desktop-qa/fixtures/smoke.json\nur desktop-qa doctor\nur desktop-qa run .ur/desktop-qa/fixtures/smoke.json',
+  },
+  {
+    title: 'Durable side chat',
+    text: 'Ask a tool-free one-turn question without blocking the main task, then continue or close its private hash-chained history.',
+    code: '/btw Why does this parser use a sentinel?\n/btw continue <chat-id> What invariant does it protect?\n/btw show <chat-id>\n/btw close <chat-id>',
+  },
+  {
+    title: 'Dependency-aware repository workspace',
+    text: 'Pin repository identities, serialize each repository writer, verify every branch, and print non-executing handoff plans.',
+    code: 'ur workspace init checkout\nur workspace add checkout api ../api --base main --verify "bun test"\nur workspace task checkout api-contract --repo api --prompt "add the response field"\nur workspace validate checkout\nur workspace run checkout --max-concurrency 4\nur workspace verify checkout\nur workspace pr-plan checkout',
   },
   {
     title: 'Reliable repo rename',

package/documentation/index.html

@@ -31,6 +31,7 @@
         <a href="#architecture">Architecture</a>
         <a href="#mental-model">Mental Model</a>
         <a href="#features">Feature Map</a>
+        <a href="#frontier">Frontier Agent Workflows</a>
         <a href="#tutorials">Tutorials</a>
         <a href="#commands">Command Reference</a>
         <a href="#slash">Slash Commands</a>
@@ -44,7 +45,7 @@
     <main id="content" class="content">
       <header class="topbar">
         <div>
-          <p class="eyebrow">Version 1.47.1</p>
+          <p class="eyebrow">Version 1.49.0</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>
@@ -165,7 +166,7 @@ ur config set responses.store false</code></pre>
           </article>
           <article>
             <h3>Status bar and updates</h3>
-            <pre><code>Ollama | llama3 | ask | main | update 1.47.1 available</code></pre>
+            <pre><code>Ollama | llama3 | ask | main | update 1.48.0 available</code></pre>
             <p>The interactive status bar shows only important runtime state: provider, model, mode, branch, active tasks, checks status when known, and update availability. It is hidden in CI, dumb terminals, and print mode.</p>
           </article>
           <article>
@@ -229,6 +230,61 @@ ur config set responses.store false</code></pre>
         <div id="featureGrid" class="feature-grid"></div>
       </section>
 
+      <section id="frontier" class="section">
+        <div class="section-heading">
+          <p class="eyebrow">Production Agent Operations</p>
+          <h2>Ten frontier workflows with explicit trust boundaries</h2>
+          <p>These surfaces keep untrusted task text separate from policy, make publication explicit, and fail closed when evidence is incomplete or changes during verification.</p>
+        </div>
+        <div class="concept-grid">
+          <article>
+            <h3>1. Managed cloud fan-out</h3>
+            <p><code>ur cloud run --runner managed</code> starts isolated candidates. Selection requires a completed session, explicit <code>PASS</code>, and a safe non-empty review branch; ordering is deterministic, not comparative judging. UR never fetches or merges that branch. Cancellation also closes sessions that finish starting concurrently.</p>
+          </article>
+          <article>
+            <h3>2. Idempotent live steering</h3>
+            <p><code>ur cloud steer</code> and <code>ur bg steer</code> accept bounded messages only for active work. Request IDs are reserved before delivery, digest-bound, and deduplicated across concurrent retries. A2A steering additionally enforces authenticated task ownership.</p>
+          </article>
+          <article>
+            <h3>3. Learned playbooks</h3>
+            <p><code>ur learn playbooks</code> mines repeated proof-backed safe runs into candidates. Execution requires explicit approval. Rejection is terminal; disable verifies the promoted workflow and moves it into a private disabled archive so it can no longer run.</p>
+          </article>
+          <article>
+            <h3>4. Cited task memory</h3>
+            <p><code>ur context-pack remember</code> can cite bounded project-file excerpts and run artifacts by SHA-256. Stale, missing, rejected, and superseded entries are excluded by default. User and web sources stay explicitly unverifiable until reopened; memory search performs no network request.</p>
+          </article>
+          <article>
+            <h3>5. Patch-only Agentic CI</h3>
+            <p><code>ur agent-ci</code> runs from a trusted base in a detached worktree with read-only GitHub permissions and isolated check credentials. Exact NUL-delimited path metadata enforces rename-source, deletion, allow, and deny policy. Pre/post check state digests bind verification to the post-check patch; any staged, unstaged, untracked, or index-visibility mutation suppresses output.</p>
+          </article>
+          <article>
+            <h3>6. Trajectory-aware eval gates</h3>
+            <p><code>ur eval gate</code> can require pass, trajectory, test, cost, duration, and baseline-regression thresholds. Captures retain control-flow metadata only—never prompts, paths, tool inputs, tool outputs, or assistant text—and a requested missing metric fails closed.</p>
+          </article>
+          <article>
+            <h3>7. Electron desktop QA</h3>
+            <p><code>ur desktop-qa</code> runs bounded fixtures with guaranteed teardown and masked screenshots. Raw video or trace is refused whenever selector redaction is configured. Evidence is copied into bounded non-symlink attachments; downloads use private/no-store headers, sandbox CSP, a safe inline allow-list, and octet-stream fallback.</p>
+          </article>
+          <article>
+            <h3>8. Durable side chats</h3>
+            <p><code>/btw</code> creates a private hash-chained side conversation that survives restarts while the main task continues. Every request is a cancellable one-turn, tool-free fork; chat count, turns, content, and storage are bounded, and closed chats reject continuation.</p>
+          </article>
+          <article>
+            <h3>9. Multi-repository coordination</h3>
+            <p><code>ur workspace</code> pins canonical repository/remote identities and executes a dependency DAG in isolated worktrees. Independent repositories may run concurrently, but each repository has one serialized writer. Changed specifications cannot resume, and PR/rollback commands print plans only.</p>
+          </article>
+          <article>
+            <h3>10. Verified model arena</h3>
+            <p><code>ur arena --judge model|hybrid</code> exposes only bounded, redacted, anonymous eligible candidates to a one-turn tool-free judge. Candidates need <code>PASS</code>, a non-empty safe patch, and successful checks. Oversized full diffs are excluded rather than truncated into eligibility; malformed decisions yield no winner.</p>
+          </article>
+        </div>
+        <pre><code>ur cloud run "repair the parser" --runner managed --environment &lt;id&gt; --attempts 3
+ur agent-ci validate default
+ur eval gate starter --min-pass-rate 1 --min-trajectory-score 0.9
+ur desktop-qa validate .ur/desktop-qa/fixtures/smoke.json
+ur arena "repair the cache race" --agents 3 --judge hybrid --verify "bun test"</code></pre>
+      </section>
+
       <section id="tutorials" class="section">
         <div class="section-heading">
           <p class="eyebrow">Tutorials</p>

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

@@ -7,7 +7,7 @@ plugins {
 }
 
 group = "dev.urnexus"
-version = "1.47.1"
+version = "1.49.0"
 
 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": "1.47.1",
+  "version": "1.49.0",
   "publisher": "ur-nexus",
   "engines": {
     "vscode": "^1.92.0"

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ur-agent",
-  "version": "1.47.1",
+  "version": "1.49.0",
   "description": "UR-Nexus — autonomous engineering workflow engine (plan, execute, test, verify, document, benchmark, reproduce)",
   "type": "module",
   "packageManager": "bun@1.3.14",