Merge remote-tracking branch 'origin/main' into feat/uipath-rpa-smoke-tests

# Conflicts:
#	.github/workflows/smoke-skills.yml

Author: AlvinStanescu

.claude/commands/generate-tasks.md

@@ -112,7 +112,7 @@ Domain mapping:
 | uipath-feedback | feedback |
 | uipath-planner | planner |
 | uipath-human-in-the-loop | hitl |
-| uipath-case-management | case |
+| uipath-maestro-case | case |
 
 If a skill is not listed, derive `<domain>` by stripping the `uipath-` prefix and using the shortest unambiguous segment (e.g., `uipath-new-thing` → `newthing`). Check existing `task_id` values for the skill's domain prefix and follow suit.
 

.claude/commands/test-coverage.md

@@ -196,7 +196,7 @@ Weights reflect what tests actually catch:
 | Skill shape | Typical N/A dimensions | Effective weights |
 |---|---|---|
 | Workflow-heavy, multi-path (e.g. `uipath-rpa`, `uipath-agents`, `uipath-maestro-flow`) | — | Comp 45% / Steps 25% / Rules 15% / Path 15% |
-| Workflow-heavy, single-path (e.g. `uipath-case-management`, `uipath-human-in-the-loop`) | Path | Comp 55% / Steps 30% / Rules 15% |
+| Workflow-heavy, single-path (e.g. `uipath-maestro-case`, `uipath-human-in-the-loop`) | Path | Comp 55% / Steps 30% / Rules 15% |
 | Command-catalog skills (e.g. `uipath-platform`, `uipath-servo`, `uipath-test`, `uipath-feedback`, `uipath-data-fabric`) | Rules, Path, Steps (often) | Comp 100% (or Comp 65% / Steps 35% if the skill has explicit workflow steps) |
 | Planning skills (e.g. `uipath-planner`, `uipath-solution-design`) | Components (often), Rules (sometimes), Path | Steps 70% / Rules 30% (or Steps 100% if no rules section) |
 | Agent-orchestration skills (e.g. `uipath-diagnostics`) | Path, sometimes Components | Components 55% (sub-agents + phases) / Steps 30% / Rules 15% |

.gemini/commands/generate-tasks.toml

@@ -66,7 +66,7 @@ Domain mapping:
 | uipath-feedback | feedback |
 | uipath-planner | planner |
 | uipath-human-in-the-loop | hitl |
-| uipath-case-management | case |
+| uipath-maestro-case | case |
 
 If a skill is not listed, derive `<domain>` by stripping the `uipath-` prefix and using the shortest unambiguous segment (e.g., `uipath-new-thing` → `newthing`). Check existing `task_id` values for the skill's domain prefix and follow suit.
 

CODEOWNERS

@@ -5,7 +5,10 @@
 /.github/ @uipreliga
 
 # Hooks
-/hooks/ @DragosUnguru
+/hooks/ @DragosUnguru @RaduAna-Maria
+
+# Plugin agents
+/agents/ @RaduAna-Maria
 
 # Plugin-namespaced slash commands (e.g., /uipath:install-permissions)
 /commands/ @DragosUnguru @tmatup
@@ -39,7 +42,7 @@
 /tests/tasks/uipath-human-in-the-loop/ @dushyant-uipath
 
 # Case Management skill
-/skills/uipath-case-management/ @charlesliu9 @song-zhao-25 @jundayin
+/skills/uipath-maestro-case/ @charlesliu9 @song-zhao-25 @jundayin
 
 # Coded Apps skill
 /skills/uipath-coded-apps/ @Raina451 @deepeshrai-tech @Sandeepan-Ghosh-0312 @swati354

agents/uipath-project-discovery-agent.md

@@ -1,6 +1,6 @@
 ---
 name: uipath-project-discovery-agent
-description: "Auto-discover UiPath project structure, dependencies, conventions, and generate context files for Claude Code and UiPath Autopilot. This agent is a prerequisite for the uipath-rpa skill — it MUST run before it. TRIGGER when: User asks to create, edit, or work with UiPath workflows (coded or RPA) and project context has not been established yet in this session; User explicitly asks to generate project context or project documentation, analyze project structure, create AGENTS.md, or regenerate/refresh context or documentation. DO NOT TRIGGER when: Project context is already loaded in the conversation (visible as project-context.md rules content); User explicitly asked to regenerate but context already exists and they did not confirm."
+description: "Auto-discover UiPath project structure, dependencies, and conventions; returns context document for Claude Code/Autopilot. Spawn before workflow authoring or when user asks to refresh project context / regenerate AGENTS.md."
 model: sonnet
 tools: Bash, Read, Glob, Grep
 ---

skills/uipath-agents/SKILL.md

@@ -47,6 +47,7 @@ Determine the agent mode before proceeding:
 | Discover solution resources (processes/apps/indexes/buckets/connections) | Low-code | [lowcode/cli-commands.md](references/lowcode/cli-commands.md) § Solution Resource Discovery | |
 | Add Integration Service tool | Low-code | [lowcode/quickstart.md](references/lowcode/quickstart.md) § Scenario 5 | `lowcode/agent-json-format.md` § Integration Service tool |
 | Add index-backed context (Context Grounding) | Low-code | [lowcode/quickstart.md](references/lowcode/quickstart.md) § Scenario 7 | `lowcode/agent-json-format.md` § Context resource |
+| Add guardrails to an agent | Low-code | [lowcode/guardrails-guide.md](references/lowcode/guardrails-guide.md) (MUST read — do not guess the schema) | `lowcode/quickstart.md` § Scenario 8 |
 | Embed agent in a flow | Low-code | [lowcode/embedding-in-flows.md](references/lowcode/embedding-in-flows.md) | |
 | Wire multi-agent solution | Low-code | [lowcode/agent-solution-guide.md](references/lowcode/agent-solution-guide.md) | |
 | Inline agent node structure in a flow | Low-code | [lowcode/agent-flow-integration.md](references/lowcode/agent-flow-integration.md) | |

skills/uipath-agents/references/lowcode/agent-json-format.md

@@ -74,10 +74,13 @@ Primary configuration file. Edit directly.
       ]
     }
   ],
+  "guardrails": [],
   "projectId": "<AUTO_GENERATED_UUID>"
 }
 ```
 
+> **`guardrails`** — array of guardrail objects that inspect agent inputs/outputs for policy violations. See [guardrails-guide.md](guardrails-guide.md) for the full schema, validator reference, and examples.
+
 ### Settings
 
 | Field | Description |
@@ -267,7 +270,7 @@ The `validate` command reads these files, resolves `referenceKey` for solution t
     "exampleCalls": []                // Required for external tools
   },
   "guardrail": {
-    "policies": []
+    "policies": []              // Auto-populated by `uip agent validate` from root-level guardrails. Do not edit manually. See guardrails-guide.md.
   },
   "id": "<uuid>",              // Stable; generate once, never change
   "referenceKey": "<release-key-guid>", // For external: the release Key (lowercase GUID from /odata/Releases API). For solution-internal: leave empty, validate resolves it.
@@ -351,7 +354,7 @@ Integration Service tools call connector activities (e.g., Slack Send Message, W
   },
   "iconUrl": "<connector image URL — see rules below>",
   "settings": {},
-  "guardrail": { "policies": [] },
+  "guardrail": { "policies": [] },     // Auto-populated by validate from root-level guardrails. Do not edit manually. See guardrails-guide.md.
   "isPreview": false,
   "properties": {
     "toolPath": "<path from metadata, e.g. /v2/webSearch>",
@@ -803,7 +806,7 @@ The root `agent.json` does not contain a `resources` field. Resources are define
     "folderPath": "solution_folder"
   },
   "guardrail": {
-    "policies": []
+    "policies": []              // Legacy placeholder — do not populate. See guardrails-guide.md.
   },
   "id": "<uuid>",
   "referenceKey": "",           // Leave empty; validate resolves it and writes it back to disk

skills/uipath-agents/references/lowcode/cli-commands.md

@@ -35,6 +35,22 @@ uip agent init "<FLOW_PROJECT_DIR>" --inline-in-flow --output json
 
 After scaffolding, add a `uipath.agent.autonomous` node to the flow with `model.source = <ProjectId>`. See [embedding-in-flows.md](embedding-in-flows.md) for the full structure.
 
+### `uip agent guardrails list`
+
+List available guardrail validator definitions with their allowed scopes, stages, and parameters.
+
+```bash
+uip agent guardrails list --output json
+```
+
+Returns an array of validator definitions. Each entry contains:
+- `Validator` — the `validatorType` string to use in `builtInValidator` guardrails
+- `AllowedScopes` — valid values for `selector.scopes`
+- `GuardrailStages` — object mapping each scope to its valid execution stages
+- `Parameters` — array of parameter definitions (`Type`, `Id`, `Required`)
+
+Run this before adding built-in validator guardrails to discover which validators are available and what parameters they accept.
+
 ### `uip agent validate`
 
 Validate agent project structure and schema, then migrate all project files to the latest schema version.
@@ -373,6 +389,7 @@ uip solution deploy run \
 | Scaffold agent | `uip agent init "<NAME>" --output json` | Solution directory |
 | Scaffold inline agent | `uip agent init "<FLOW_PROJECT_DIR>" --inline-in-flow --output json` | Any directory |
 | Register project | `uip solution project add "<PATH>" --output json` | Solution directory |
+| List guardrail validators | `uip agent guardrails list --output json` | Any directory |
 | Validate + migrate | `uip agent validate [path] --output json` | Agent dir or any directory with path |
 | Upload to Studio Web | `uip solution upload . --output json` | Solution directory |
 | Pack | `uip solution pack . ./dist -v "1.0.0" --output json` | Solution directory |

skills/uipath-agents/references/lowcode/embedding-in-flows.md

@@ -63,7 +63,7 @@ Same schema as a standalone agent (see [agent-json-format.md](agent-json-format.
 - `projectId` matches the folder name UUID
 - `inputSchema.properties` is empty (flow wires data via node connections)
 - `messages` have empty `content` and `contentTokens` initially (edit agent.json to set prompts with `type: "simpleText"` and `rawString`)
-- `guardrails: []` at root level
+- `guardrails: []` at root level — can be populated with guardrail objects. See [guardrails-guide.md](guardrails-guide.md)
 - No `metadata.targetRuntime` field
 
 Example: