Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
8 changes: 7 additions & 1 deletion build/tools/customize-tool/inputs.mdx
Original file line number Diff line number Diff line change
Expand Up @@ -96,9 +96,9 @@

The schema describes the overall structure of the data and the types that each item can be. Use the documentation at [json-schema.org](https://json-schema.org/) for more options.

<Warning> Note that the only supported keywords are `properties` - all other schema keywords (e.g. `additionalProperties`, `examples`, etc.) will be ignored.</Warning>

Check warning on line 99 in build/tools/customize-tool/inputs.mdx

View check run for this annotation

Mintlify / Mintlify Validation (relevanceai) - vale-spellcheck

build/tools/customize-tool/inputs.mdx#L99

Did you really mean 'additionalProperties'?

### List of JSONs

Check warning on line 101 in build/tools/customize-tool/inputs.mdx

View check run for this annotation

Mintlify / Mintlify Validation (relevanceai) - vale-spellcheck

build/tools/customize-tool/inputs.mdx#L101

Did you really mean 'JSONs'?

Handles multiple JSON objects in a list structure

Expand Down Expand Up @@ -202,4 +202,10 @@
border: '3px solid #5E43CE',
borderRadius: '10px',
}}/>
</div>
</div>

---

## Auto-fill from run context

Text inputs can be configured to automatically fill with information about the user who triggered the run. The server resolves the triggering user's identity and injects their verified email address or display name directly into the tool's parameters — the AI never sees or provides these values. See [Text input](/build/tools/user-inputs/text-input) for details on configuring auto-fill options.
4 changes: 4 additions & 0 deletions build/tools/tool-steps/getting-started.mdx
Original file line number Diff line number Diff line change
Expand Up @@ -37,6 +37,10 @@ Tips:
4. The order of a step can be changed using the drag indicator or the up and down buttons. Steps can be ran individually, which is helpful for testing, by pressing the play icon.
</Tip>

<Tip>
Auto-filled inputs — such as triggering user's email or name — are referenced in steps using the same variable syntax as any other input. They're available throughout the tool execution wherever you can use input variables. See [Text input](/build/tools/user-inputs/text-input) for details on configuring auto-fill options.
</Tip>

## General Tool step features
<Snippet file="components/tools/run-step.mdx" />

Expand Down
31 changes: 31 additions & 0 deletions build/tools/user-inputs/text-input.mdx
Original file line number Diff line number Diff line change
Expand Up @@ -20,6 +20,37 @@ or the side-bar.
<Snippet file="components/inputs/general-input-setting.mdx" />
<Snippet file="components/inputs/default-value-tip.mdx" />

## Fill automatically from the run

Text inputs support automatic filling with information about the user who triggered the agent run. Select one of the auto-fill options and the server resolves the user's identity at run time, injecting the value directly into the tool's parameters without the AI providing or seeing it.

Open the **Fill automatically from the run** dropdown on any text input. The two available options are **Triggering user's email** (the verified email address of the human who started the run) and **Triggering user's name** (the display name of the human who started the run).

<Note>
These values are resolved and injected server-side. They are hidden from the AI and cannot be spoofed or overridden by the model.
</Note>

<Warning>
Scheduled runs, webhook triggers, and embed triggers have no associated user. Auto-fill inputs remain empty for non-human triggers. If an auto-fill input is marked as required and the run is triggered without a human user, the run will be rejected.
</Warning>

### Use cases

<CardGroup cols={2}>
<Card title="Logging" icon="list">
Record which user performed which action in your tool outputs or connected systems.
</Card>
<Card title="Personalization" icon="user">
Customize tool responses or outputs based on the identity of the person running it.
</Card>
<Card title="Audit trails" icon="shield-check">
Maintain records of who triggered specific operations for compliance or review.
</Card>
<Card title="Security" icon="lock">
Tie actions to verified user identities without relying on user-supplied values.
</Card>
</CardGroup>

## Access the entered text
To access the entered text use the variable mode `{{}}` and the input step name. Under the default setting it is accessible
via `{{text}}` (or `params.text` in a JavaScript step).
Loading