@uipath/skills 1.196.0 → 1.197.0

package/skills/uipath-agents/references/lowcode/agent-definition.md

@@ -19,14 +19,15 @@ After `uip agent init <name>`:
 
 ## agent.json
 
-Primary configuration file. Edit directly.
+### Autonomous agent.json
+Primary configuration file for autonomous agent. Edit directly.
 
 ```json
 {
   "version": "1.1.0",
   "settings": {
     "model": "<MODEL_IDENTIFIER>",
-    "maxTokens": 16384,
+    "maxTokens": 128000,
     "temperature": 0,
     "engine": "basic-v2",
     "maxIterations": 25,
@@ -79,20 +80,72 @@ Primary configuration file. Edit directly.
 }
 ```
 
+### Conversational agent.json
+Primary configuration file for conversational agent. Edit directly.
+
+```json
+{
+  "version": "1.1.0",
+  "settings": {
+    "model": "<MODEL_IDENTIFIER>",
+    "maxTokens": 64000,
+    "temperature": 0,
+    "engine": "conversational-v1",
+    "mode": "standard"
+  },
+  "inputSchema": {
+    "type": "object",
+    "properties": {
+      "<FIELD_NAME>": {
+        "type": "string",
+        "description": "<FIELD_DESCRIPTION>"
+      }
+    },
+    "required": ["<FIELD_NAME>"]
+  },
+  "outputSchema": {
+    "type": "object",
+    "properties": {}
+  },
+  "metadata": {
+    "storageVersion": "50.0.0",
+    "isConversational": true,
+    "showProjectCreationExperience": false,
+    "targetRuntime": "pythonAgent"
+  },
+  "type": "lowCode",
+  "messages": [
+    {
+      "role": "system",
+      "content": "<SYSTEM_PROMPT>",
+      "contentTokens": [
+        { "type": "simpleText", "rawString": "<SYSTEM_PROMPT>" }
+      ]
+    },
+    {
+      "role": "user",
+      "content": "",
+      "contentTokens": []
+    }
+  ],
+  "projectId": "<AUTO_GENERATED_UUID>"
+}
+```
+
 > **`guardrails`** — array of guardrail objects that inspect agent inputs/outputs for policy violations. See [capabilities/guardrails/guardrails.md](capabilities/guardrails/guardrails.md) for the full schema, validator reference, and examples.
 
 ### Settings
 
 | Field | Description |
 |-------|-------------|
-| `model` | LLM identifier. Discover valid values with `uip agent model list` and select per [model-selection-guide.md](model-selection-guide.md) — **override the scaffold default `gpt-4o-2024-11-20`** (illustrative GA: `"anthropic.claude-sonnet-4-6"`, `"gpt-5.4"` — verify against the tenant). |
+| `model` | LLM identifier. Discover valid values with `uip agent model list` and select per [model-selection-guide.md](model-selection-guide.md) — **override the scaffold default** (illustrative GA: `"anthropic.claude-sonnet-4-6"`, `"gpt-5.4"` — verify against the tenant). |
 | `maxTokens` | Max output tokens. Must not exceed the chosen model's `MaxTokens` cap (from `uip agent model list`). |
 | `temperature` | 0 = deterministic, higher = creative |
-| `engine` | Use `"basic-v2"` |
-| `maxIterations` | Max agent loop iterations. Default 25. |
+| `engine` | Keep `"basic-v2"` for autonomous, `"conversational-v1"` for conversational |
+| `maxIterations` | Max autonomous agent loop iterations. Default 25. Keep as omitted for conversational. |
 | `mode` | Use `"standard"` |
 
-> Prompt **quality** (system/user prompt structure, tool-call criteria, output contract) lives in [agent-prompting-guide.md](agent-prompting-guide.md). This file owns the **mechanics** (schema, `contentTokens` sync).
+> Prompt **quality** (system/user prompt structure, tool-call criteria, output contract) lives in [prompting/agent-prompting-guide.md](prompting/agent-prompting-guide.md). This file owns the **mechanics** (schema, `contentTokens` sync).
 
 ### Schema Types
 
@@ -164,17 +217,35 @@ Runtime note: attachments cannot be supplied via `uip` CLI. Test from Studio Web
 | Field | Value |
 |-------|-------|
 | `storageVersion` | Managed by `uip agent refresh` — do not edit |
-| `isConversational` | `false` (autonomous agents) |
+| `isConversational` | `false` for autonomous agents, `true` for conversational agents. Do not edit. |
 | `showProjectCreationExperience` | `false` |
-| `targetRuntime` | `"pythonAgent"` |
+| `targetRuntime` | `"pythonAgent"` for autonomous. **`null` for conversational** — conversational agents are not yet in PROD, so the field is intentionally left null until the runtime value is finalized. |
+
+### Input Schema
+
+For **autonomous agents**, the `inputSchema` defines the input properties of the agent which can be templated into the system-prompt and user-prompt.
+
+For **conversational agents**, each agent run handles one conversational exchange - a single run corresponds to a single turn of messages/tool-calls taken in response to a user-initiated message. The runtime supplies the following inputs per run:
+- **`messages`** (reserved, implicit) — current conversation history, including user's latest message. Always present; never declare it (or other fields representing conversation-history or user's input message) in `inputSchema`, per [critical-rules/conversational-critical-rules.md](critical-rules/conversational-critical-rules.md) anti-pattern 4.
+- **Custom `inputSchema` fields** — additional per-exchange context variables which can be templated into the system-prompt. Leave `inputSchema` blank unless the use case genuinely needs per-exchange context variables from the chat-surface beyond the conversation history and user's input messages, which are already captured in `messages`.
+
+
+### Output Schema
+
+For **autonomous agents**, the `outputSchema` defines the output properties of the agent.
+
+For **conversational agents**, the `outputSchema` should not be modified, and thus always left empty. See [critical-rules/conversational-critical-rules.md](critical-rules/conversational-critical-rules.md) anti-pattern 1.
+
 
 ## Messages
 
-This section covers message **structure** and `contentTokens` mechanics. For prompt **quality** — system-prompt skeleton, tool-call criteria, output contract, worked examples — see [agent-prompting-guide.md](agent-prompting-guide.md).
+This section covers message **structure** and `contentTokens` mechanics. For prompt **quality** — system-prompt skeleton, tool-call criteria, output contract, worked examples — see [prompting/agent-prompting-guide.md](prompting/agent-prompting-guide.md).
 
 ### System Message
 
-Sets the agent's role and behavior. Typically plain text with no variables:
+Sets the agent's role and behavior. 
+
+For autonomous, typically plain text with no variables:
 
 ```json
 {
@@ -186,9 +257,21 @@ Sets the agent's role and behavior. Typically plain text with no variables:
 }
 ```
 
+For conversational, typically general behavior and steps to respond to users. Should also generally have no variables unless the conversational inputs feature is used, see § Input Schema.
+
+```json
+{
+  "role": "system",
+  "content": "You are a helpful assistant that runs tools and responds to users based on their queries.",
+  "contentTokens": [
+    { "type": "simpleText", "rawString": "You are a helpful assistant that runs tools and responds to users based on their queries." }
+  ]
+}
+```
+
 ### User Message
 
-Templates input fields into the prompt using `{{input.fieldName}}`. For `job-attachment` fields the token renders metadata only (see § File Attachments).
+For autonomous agents, templates input fields into the prompt using `{{input.fieldName}}`. For `job-attachment` fields the token renders metadata only (see § File Attachments).
 
 ```json
 {
@@ -203,6 +286,8 @@ Templates input fields into the prompt using `{{input.fieldName}}`. For `job-att
 }
 ```
 
+For conversational agents, the user-message should be ignored after initialization and left blank. The user message from the `agent.json` is ignored since each user message is received during the actual conversation.
+
 ## contentTokens Construction
 
 Every message needs both `content` (string) and `contentTokens` (array). Keep them in sync.
@@ -318,7 +403,7 @@ The `validate` command reads these files and resolves `referenceKey` for solutio
 | `"solution"` | Typically `"solution_folder"` (the in-solution declared folder) | `Folder` field from `uip solution resources list` |
 | `"external"` | Literal Orchestrator folder, slash-separated (e.g., `"Shared/Sales"`) | `Folder` field from `uip solution resources list` |
 
-The author writes the value verbatim into `resource.json` (or into the guardrail action under `agent.json`); `uip agent refresh` propagates it into `bindings_v2.json` as `folderPath` (App resources translate `folderName` → binding `folderPath`). Connection (Integration Service) resources are exempt — bound by `connection.id`, no `folderPath`. See [critical-rules.md](critical-rules.md) Rule 11 and [solution-resources.md](solution-resources.md) § Bindings.
+The author writes the value verbatim into `resource.json` (or into the guardrail action under `agent.json`); `uip agent refresh` propagates it into `bindings_v2.json` as `folderPath` (App resources translate `folderName` → binding `folderPath`). Connection (Integration Service) resources are exempt — bound by `connection.id`, no `folderPath`. See [critical-rules/critical-rules.md](critical-rules/critical-rules.md) Rule 11 and [solution-resources.md](solution-resources.md) § Bindings.
 
 For each resource type's full schema, see the relevant capability file:
 

package/skills/uipath-agents/references/lowcode/capabilities/built-in-tools/analyze-attachments.md

@@ -105,11 +105,11 @@ Schema is canonical — copy verbatim. Only `id` (fresh UUID) and optionally `de
 
 - `properties.toolType` MUST be exactly `"analyze-attachments"` (kebab-lowercase). Anything else is silently ignored.
 - The `definitions.job-attachment` block belongs inside the tool's `inputSchema`, not at the agent root. Each schema (agent input, agent output, tool input) carries its own copy.
-- `{{input.<field>}}` only surfaces metadata; without this tool the agent has no way to read file contents. See [../../critical-rules.md](../../critical-rules.md) Critical Rule 18.
+- `{{input.<field>}}` only surfaces metadata; without this tool the agent has no way to read file contents. See [../../critical-rules/critical-rules.md](../../critical-rules/critical-rules.md) Critical Rule 17.
 - Agents can return a `job-attachment` from `outputSchema` (e.g. a generated file), but `analyze-attachments` itself only reads — it does not produce attachments.
 
 ## References
 
 - [built-in-tools.md](built-in-tools.md) — capability overview
 - [../../agent-definition.md](../../agent-definition.md) § File Attachments — `job-attachment` schema
-- [../../critical-rules.md](../../critical-rules.md) Critical Rules 18–21
+- [../../critical-rules/critical-rules.md](../../critical-rules/critical-rules.md) Critical Rules 17–20

package/skills/uipath-agents/references/lowcode/capabilities/built-in-tools/batch-transform/impl-json.md

@@ -4,17 +4,34 @@
 
 ## Resource Shape
 
-`resources/<resource-name>/resource.json`:
+`resources/<resource-name>/resource.json` — the full shape `uip agent refresh`/`validate` accept:
 
 ```json
 {
   "$resourceType": "tool",
-  "id": "<UUID>",
+  "id": "<FRESH_UUID>",
+  "name": "BatchTransform",
+  "description": "<what this batch transform does, e.g. add MCC Code and Confidence columns to each CSV row>",
   "type": "internal",
-  "referenceKey": null,
+  "inputSchema":  { "type": "object", "properties": {} },
+  "outputSchema": { "type": "object", "properties": {} },
+  "settings": {},
   "isEnabled": true,
+  "referenceKey": null,
+  "guardrail": { "policies": [] },
+  "argumentProperties": {},
   "properties": {
-    "toolType": "batch-transform"
+    "toolType": "batch-transform",
+    "settings": {
+      "contextType": "attachments",
+      "query": { "variant": "dynamic" },
+      "folderPathPrefix": { "variant": "static", "value": "<output-bucket-path>" },
+      "outputColumns": [
+        { "name": "MCC Code",   "description": "Return the 4-digit MCC code for this row. Output only the 4-digit string, e.g. 5411; output 0000 if undeterminable." },
+        { "name": "Confidence", "description": "Confidence the MCC Code is correct. Output exactly one of: HIGH, MEDIUM, or LOW." }
+      ],
+      "webSearchGrounding": { "value": "Disabled" }
+    }
   }
 }
 ```
@@ -22,26 +39,35 @@
 | Field | Constraint |
 |---|---|
 | `$resourceType` | `"tool"` |
-| `id` | UUID-shaped string |
+| `id` | Fresh UUID-shaped string |
+| `name` | Required — and the resource folder MUST be named exactly this (`resources[0].name: missing name` if absent; `folder must be named after the resource name` if the folder differs) |
+| `description` | Free text |
 | `type` | `"internal"` (built-in tools) |
+| `inputSchema` / `outputSchema` | Empty object schema `{ "type": "object", "properties": {} }` — batch-transform declares no I/O schema; the CSV and columns flow through `properties.settings` |
+| `settings`, `argumentProperties` | `{}` (backward-compat scaffolding) |
+| `guardrail.policies` | `[]` |
 | `referenceKey` | `null` (non-null identifies an external tool) |
 | `isEnabled` | truthy |
 | `properties.toolType` | `"batch-transform"` (others: `analyze-attachments`, `load-attachments`, `deep-rag`) |
+| `properties.settings` | Tool config — see [Tool Configuration](#tool-configuration) |
 
-Resource directory name is free-form. Validator scans `<agent>/resources/**/resource.json`.
+Validator scans `<agent>/resources/**/resource.json`; each `resource.json`'s parent folder must equal its `name`.
 
-## Tool Configuration
+> The repo's static checker asserts only `$resourceType`, `type`, `referenceKey`, `id`, `isEnabled`, and `properties.toolType`. The full shape above is what `uip agent refresh`/`validate` require — author it directly. Do not reduce it to that minimal field set, and do not treat a `refresh` schema rejection as a reason to reverse-engineer the CLI: fix the resource against this template.
 
-`resource.json` only registers the tool's existence — it does NOT carry per-call inputs (prompt, output columns, destination, web grounding). Those are derived at runtime from the agent's instructions and tool-call output.
+## Tool Configuration
 
-What the agent must produce at invocation time (the runtime forwards these to the underlying API):
+batch-transform's `inputSchema`/`outputSchema` are empty — the per-run config lives in `properties.settings`:
 
-- A non-empty top-level prompt
-- A list of `BatchTransformOutputColumn` entries (`name` 1–500 chars, regex `^[\w\s\.,!?-]+$`; `description` 1–20000 chars, the per-column LLM instruction)
-- An output destination
-- Whether to enable web grounding
+| Setting | Meaning |
+|---|---|
+| `contextType` | `"attachments"` — CSV arrives as a runtime attachment |
+| `query.variant` | `"dynamic"` — per-row prompt derived at runtime |
+| `folderPathPrefix` | Output destination: `{ "variant": "static", "value": "<bucket-path>" }`. Use a unique suffix per run to avoid overwrites |
+| `outputColumns[]` | Per-column LLM instructions — `name` (1–500 chars, regex `^[\w\s\.,!?-]+$`) + `description` (1–20000 chars, the per-column instruction). See [Output Column Descriptions](#output-column-descriptions) |
+| `webSearchGrounding.value` | `"Disabled"` by default; enable only when rows need fresh external data |
 
-The exact JSON field names the runtime sends to the platform (e.g., `useWebSearchGrounding`, `targetFileGlobPattern`) are SDK-version-specific — see the SDK source `uipath/platform/context_grounding/_context_grounding_service.py` (`_batch_transform_*_creation_spec` methods) for the body shape your version emits. Authoring the agent's system prompt (next section) is what shapes the runtime call, not editing `resource.json`.
+The `properties.settings` keys above are the Studio Web authoring shape. At runtime the platform receives a separate API body (e.g. `useWebSearchGrounding`, `targetFileGlobPattern`) whose names are SDK-version-specific — see `uipath/platform/context_grounding/_context_grounding_service.py` (`_batch_transform_*_creation_spec`). If `uip agent refresh` rejects a setting, verify the key against your CLI version.
 
 ## Standalone vs Inline-in-Flow
 

package/skills/uipath-agents/references/lowcode/capabilities/built-in-tools/built-in-tools.md

@@ -15,6 +15,7 @@ For process tools (RPA / agent / API / agentic), see [../process/process.md](../
 2. **`properties.toolType` is the discriminator** — fixed per built-in, kebab-lowercase. Copy from the per-tool walkthrough; do not invent.
 3. **No solution-level files.** Built-in tools need no `uip solution resources refresh`. Validate the agent and bundle.
 4. **Input/output schemas are fixed.** Do not edit them. Each tool's walkthrough lists the canonical schema.
+5. **Inline agents need the flow node too.** When the built-in tool is on an *inline* agent (embedded in a flow), authoring the `resource.json` is **not enough** — also wire a `uipath.agent.resource.tool.builtin.<toolType>` flow node to the autonomous node's `tool` handle. Fetch the node manifest with `uip maestro flow registry get` and hand the node + edge authoring to the `uipath-maestro-flow` skill (Critical Rule 16). Without the flow node the tool is never reachable at runtime. See [../inline-in-flow/inline-in-flow.md](../inline-in-flow/inline-in-flow.md).
 
 ## Resource Shape
 
@@ -45,7 +46,7 @@ For process tools (RPA / agent / API / agentic), see [../process/process.md](../
 | `inputSchema` / `outputSchema` | Fixed per tool — copy from walkthrough |
 | `referenceKey` | Always `null` (no Orchestrator binding) |
 | `guardrail.policies` | Always `[]` — required for backward compatibility |
-| `id` | Fresh UUID per resource — see [../../critical-rules.md](../../critical-rules.md) Anti-pattern 9 |
+| `id` | Fresh UUID per resource — see [../../critical-rules/critical-rules.md](../../critical-rules/critical-rules.md) Anti-pattern 9 |
 
 ## Lifecycle
 
@@ -64,7 +65,7 @@ For process tools (RPA / agent / API / agentic), see [../process/process.md](../
 
 ## Gotchas
 
-- See [../../critical-rules.md](../../critical-rules.md) Critical Rules 18–21 and Anti-patterns 21–22 for the canonical rule list.
+- See [../../critical-rules/critical-rules.md](../../critical-rules/critical-rules.md) Critical Rules 17–20 and Anti-patterns 20–21 for the canonical rule list.
 - Pairing with file inputs: a `job-attachment` field renders metadata only in `{{input.<field>}}`. The agent reads contents only by calling a file-handling built-in tool. See [../../agent-definition.md](../../agent-definition.md) § File Attachments.
 - Runtime test path: built-in tools cannot be exercised end-to-end through the `uip` CLI. Test from Studio Web or via Orchestrator job invocation.
 
@@ -72,4 +73,4 @@ For process tools (RPA / agent / API / agentic), see [../process/process.md](../
 
 - [analyze-attachments.md](analyze-attachments.md) — Analyze Files walkthrough
 - [../../agent-definition.md](../../agent-definition.md) § File Attachments — `job-attachment` schema
-- [../../critical-rules.md](../../critical-rules.md) — canonical rules
+- [../../critical-rules/critical-rules.md](../../critical-rules/critical-rules.md) — canonical rules

package/skills/uipath-agents/references/lowcode/capabilities/built-in-tools/deeprag/impl-json.md

@@ -4,17 +4,29 @@
 
 ## Resource Shape
 
-`resources/<resource-name>/resource.json`:
+`resources/<ResourceName>/resource.json` — the full shape `uip agent refresh`/`validate` accept. **The folder name must exactly match the `name` field** (case- and whitespace-sensitive) or the validator fails with `folder must be named after the resource name`:
 
 ```json
 {
   "$resourceType": "tool",
-  "id": "<UUID>",
+  "id": "<FRESH_UUID>",
+  "name": "DeepRag",
+  "description": "<what this deep research pass does, e.g. synthesize an answer across uploaded PDFs with citations>",
   "type": "internal",
-  "referenceKey": null,
+  "inputSchema":  { "type": "object", "properties": {} },
+  "outputSchema": { "type": "object", "properties": {} },
+  "settings": {},
   "isEnabled": true,
+  "referenceKey": null,
+  "guardrail": { "policies": [] },
+  "argumentProperties": {},
   "properties": {
-    "toolType": "deep-rag"
+    "toolType": "deep-rag",
+    "settings": {
+      "contextType": "attachments",
+      "query": { "variant": "dynamic" },
+      "folderPathPrefix": { "variant": "static" }
+    }
   }
 }
 ```
@@ -22,13 +34,33 @@
 | Field | Constraint |
 |---|---|
 | `$resourceType` | `"tool"` |
-| `id` | UUID-shaped string |
+| `id` | Fresh UUID-shaped string |
+| `name` | Required — and the resource folder MUST be named exactly this (`missing name` if absent; `folder must be named after the resource name` if the folder differs) |
+| `description` | Free text |
 | `type` | `"internal"` (built-in tools) |
+| `inputSchema` / `outputSchema` | Empty object schema `{ "type": "object", "properties": {} }` — deep-rag declares no I/O schema; attachments flow through at runtime |
+| `settings`, `argumentProperties` | `{}` (backward-compat scaffolding) |
+| `guardrail.policies` | `[]` |
 | `referenceKey` | `null` (non-null identifies an external tool) |
 | `isEnabled` | truthy |
 | `properties.toolType` | `"deep-rag"` (others: `analyze-attachments`, `load-attachments`, `batch-transform`) |
+| `properties.settings` | Tool config — see [Tool Configuration](#tool-configuration) |
+
+Validator scans `<agent>/resources/**/resource.json`; each `resource.json`'s parent folder must equal its `name`.
+
+> The repo's static checker asserts only `$resourceType`, `type`, `referenceKey`, `id`, `isEnabled`, and `properties.toolType`. The full shape above is what `uip agent refresh`/`validate` require — author it directly. Do not reduce it to that minimal field set, and do not treat a `refresh` schema rejection as a reason to reverse-engineer the CLI: fix the resource against this template.
+
+## Tool Configuration
+
+deep-rag's `inputSchema`/`outputSchema` are empty — the config lives in `properties.settings`:
+
+| Setting | Meaning |
+|---|---|
+| `contextType` | `"attachments"` — documents arrive as runtime attachments |
+| `query.variant` | `"dynamic"` — the research prompt is derived at runtime |
+| `folderPathPrefix` | `{ "variant": "static" }` — folder scoping for the ephemeral index |
 
-Resource directory name is free-form. Validator scans `<agent>/resources/**/resource.json`.
+Unlike `batch-transform`, deep-rag carries no `outputColumns` and no `webSearchGrounding`. The context-index form of DeepRAG (not this tool form) instead takes `citationMode` (`"Inline"`/`"Skip"`) — see [planning.md](planning.md) § Tool resource vs context-index resource. The `properties.settings` keys are the Studio Web authoring shape; if `uip agent refresh` rejects one, verify it against your CLI version.
 
 ## Standalone vs Inline-in-Flow
 

package/skills/uipath-agents/references/lowcode/capabilities/context/attachments.md

@@ -45,7 +45,7 @@ None. No solution-level file is produced — attachments are runtime-only.
 
 ## Gotchas
 
-`contextType` value MUST be `"attachments"` (all lowercase) — see [../../critical-rules.md](../../critical-rules.md) Anti-pattern 12.
+`contextType` value MUST be `"attachments"` (all lowercase) — see [../../critical-rules/critical-rules.md](../../critical-rules/critical-rules.md) Anti-pattern 12.
 
 ## References
 

package/skills/uipath-agents/references/lowcode/capabilities/context/context.md

@@ -26,7 +26,7 @@ Contexts feed retrievable information into an agent at runtime. Three variants e
 
 ## Casing Rule
 
-`contextType` and `retrievalMode` values are lowercase. See [../../critical-rules.md](../../critical-rules.md) § What NOT to Do — Anti-pattern 12.
+`contextType` and `retrievalMode` values are lowercase. See [../../critical-rules/critical-rules.md](../../critical-rules/critical-rules.md) § What NOT to Do — Anti-pattern 12.
 
 ## Sibling Files
 

package/skills/uipath-agents/references/lowcode/capabilities/context/datafabric.md

@@ -43,7 +43,7 @@ No `indexName` and no `settings` for DataFabric contexts. The shape is entirely
 
 ## Gotchas
 
-`contextType` value MUST be `"datafabricentityset"` (all lowercase) — see [../../critical-rules.md](../../critical-rules.md) Anti-pattern 12.
+`contextType` value MUST be `"datafabricentityset"` (all lowercase) — see [../../critical-rules/critical-rules.md](../../critical-rules/critical-rules.md) Anti-pattern 12.
 
 ## References
 

package/skills/uipath-agents/references/lowcode/capabilities/context/index.md

@@ -7,7 +7,7 @@ For other context variants, see [context.md](context.md).
 ## When to Use
 
 - Agent needs to retrieve from a knowledge base of indexed documents
-- The index already exists in Context Grounding and is backed by an Orchestrator storage bucket
+- The index already exists in Context Grounding and is backed by an Orchestrator storage bucket. To create or manage that index from the CLI, see [uipath-platform/references/context-grounding/index-management.md](../../../../../uipath-platform/references/context-grounding/index-management.md)
 
 `uip solution resources refresh` emits an `index` binding into `bindings_v2.json`, resolves the backing storage bucket via ECS + Orchestrator, and writes all three artifacts automatically: `resources/solution_folder/index/<IndexName>.json`, `resources/solution_folder/bucket/orchestratorBucket/<BucketName>.json`, and two `debug_overwrites.json` entries (`kind: "index"`, `kind: "bucket"`). No manual solution-level authoring is required.
 
@@ -95,7 +95,7 @@ Wrapper-level `apiVersion` is `"ecs.uipath.com/v2"` — matches what refresh wri
 
 **`folderPathPrefix.variant`:** `"static"` (no prefix) or `"argument"` (scope by a folder path provided at runtime).
 
-**Casing matters.** All `contextType` and `retrievalMode` values are lowercase. See [../../critical-rules.md](../../critical-rules.md) Anti-pattern 12.
+**Casing matters.** All `contextType` and `retrievalMode` values are lowercase. See [../../critical-rules/critical-rules.md](../../critical-rules/critical-rules.md) Anti-pattern 12.
 
 ## Solution-Level Files
 
@@ -212,6 +212,10 @@ All failures (index not found, ambiguous name match, non-StorageBucket data sour
 
 See § Agent-Level Resource Shape above for the full field reference, including the three variants (`index`/`attachments`/`datafabricentityset`) and per-`retrievalMode` settings (`citationMode` for `deeprag`, `webSearchGrounding` + `outputColumns` for `batchtransform`).
 
+### Step 4b — Inline agents only: wire the context flow node
+
+**Skip if the agent is standalone.** If the context is on an **inline** agent (embedded in a flow), the `resource.json` alone is never reached at runtime — you MUST also add a `uipath.agent.resource.context.index.<index-name>.<index-id>` flow node connected to the autonomous node's `context` handle (bottom port). Fetch its manifest with `uip maestro flow registry get "<NodeType>" --output json`, then hand the node + edge authoring to the `uipath-maestro-flow` skill (Critical Rule 16 — this skill does not author `.flow` graphs directly). Run Step 5's refresh/validate with `--inline-in-flow` plus `--bindings-target <FlowProjectDir>/bindings_v2.json`. See [../inline-in-flow/inline-in-flow.md](../inline-in-flow/inline-in-flow.md).
+
 ### Step 5 — Refresh and validate
 
 ```bash
@@ -254,7 +258,7 @@ The upload response includes a `Data.DesignerUrl` — open it to verify the cont
 
 ## Gotchas
 
-`contextType` and `retrievalMode` values MUST be lowercase — see [../../critical-rules.md](../../critical-rules.md) Anti-pattern 12.
+`contextType` and `retrievalMode` values MUST be lowercase — see [../../critical-rules/critical-rules.md](../../critical-rules/critical-rules.md) Anti-pattern 12.
 
 ## References