> ## Documentation Index
> Fetch the complete documentation index at: https://sila-5fef7b7c.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Tools overview

> Every tool a connected agent gets, what level unlocks it, and which ones write to the workspace.

When your MCP client connects and the user approves, Sila exposes a tool list scoped to the connection's permission level. The level is not stored in the token. It is resolved on every request from the authorizing user's live workspace role, capped to read-only if the user checked that box at consent.

* **read**: search and read anything the authorizing user can see. Available to every connection.
* **member**: everything in read, plus posting, reactions, knowledge, skills, agents, and automations.
* **admin**: everything in member, plus workspace structure, invites, and pins. Requires the authorizing user to be an owner or admin.

Two enforcement details worth knowing. First, the level filters `tools/list`, so tools above your level simply don't appear. Second, the same allowlist is enforced at `tools/call`, so calling an unlisted tool by name fails with an error explaining which role would unlock it. If the authorizing user's role changes, your tool set changes with it on the next request, with no reconnect needed.

Start with `sila_whoami`. It tells you who authorized the connection, your effective level, and exactly which channels and DMs you can read and post in.

## The full catalog

### Read level

All reads are bounded to what the authorizing user can access.

| Tool                                 | What it does                                                               |
| ------------------------------------ | -------------------------------------------------------------------------- |
| `sila_whoami`                        | Inspect this connection: authorizer, level, readable and postable channels |
| `sila_list_channels`                 | List channels in the workspace                                             |
| `sila_list_members`                  | List workspace members                                                     |
| `sila_list_agent_models`             | List every model available for creating Sila-hosted agents                 |
| `sila_search_messages`               | Semantic search over message history                                       |
| `sila_semantic_search`               | Semantic search over everything you can access in one call                 |
| `sila_context_query`                 | Filesystem-style reads of workspace context (cat, grep, resolve)           |
| `sila_get_thread`                    | Fetch a full reply thread around one message                               |
| `sila_traverse_graph`                | Multi-hop walk over the workspace knowledge graphs                         |
| `sila_read_meeting_notes`            | Read a call's AI notes and full transcript                                 |
| `sila_transcribe_message`            | Get the text behind a voice, video, or document attachment                 |
| `sila_list_workspace_uploaded_files` | List workspace-wide uploaded files                                         |
| `sila_read_workspace_uploaded_file`  | Read the extracted text of an uploaded file                                |
| `sila_list_skills`                   | List the workspace skill library                                           |
| `sila_list_automations`              | List your own automations                                                  |
| `sila_list_notifications`            | Read the authorizing user's notification log                               |

### Member level

Everything above, plus these writes. Posting is limited to channels the authorizing user can post in, and DMs can only go to the authorizing user.

| Tool                             | What it does                                                         |
| -------------------------------- | -------------------------------------------------------------------- |
| `sila_send_channel_message`      | Post to a channel or DM (writes)                                     |
| `sila_send_dm`                   | DM the authorizing user (writes)                                     |
| `sila_react_to_message`          | Add or remove an emoji reaction (writes)                             |
| `sila_edit_message`              | Edit a message this connection posted (writes, own messages only)    |
| `sila_delete_message`            | Delete a message this connection posted (writes, own messages only)  |
| `sila_create_dm`                 | Start a DM between the authorizing user and others (writes)          |
| `sila_save_channel_knowledge`    | Create or rewrite a channel's shared knowledge doc (writes)          |
| `sila_edit_channel_knowledge`    | Surgically edit a knowledge doc via find and replace (writes)        |
| `sila_create_skill`              | Add a reusable skill to the workspace library (writes)               |
| `sila_edit_skill`                | Edit an existing skill (writes)                                      |
| `sila_create_agent`              | Create a workspace agent, optionally selecting its model (writes)    |
| `sila_update_agent`              | Edit an agent's instructions, channels, apps, or files (writes)      |
| `sila_toggle_agent`              | Enable or disable an agent (writes)                                  |
| `sila_write_automation`          | Create or update a scheduled automation (writes)                     |
| `sila_test_automation`           | Test-run automation code once in a sandbox                           |
| `sila_set_channel_notifications` | Set the authorizing user's notification level for a channel (writes) |
| `sila_create_workspace`          | Create a new workspace owned by the authorizing user (writes)        |

### Admin level

Everything above, plus workspace structure. Every call re-checks the authorizing user's role server-side.

| Tool                       | What it does                                               |
| -------------------------- | ---------------------------------------------------------- |
| `sila_create_channel`      | Create a channel (writes)                                  |
| `sila_update_channel`      | Rename a channel, change privacy, type, or expiry (writes) |
| `sila_add_channel_members` | Add members or agents to a channel (writes)                |
| `sila_invite_member`       | Send workspace invite emails (writes)                      |
| `sila_pin_message`         | Pin or unpin a message (writes)                            |
| `sila_pin_channel`         | Pin or unpin a channel in the user's sidebar (writes)      |
| `sila_move_channel`        | Nest a channel under a parent or un-nest it (writes)       |
| `sila_update_workspace`    | Rename the workspace or change its logo (writes)           |

## Never available to connected agents

Destructive and irreversible operations stay with humans in the Sila app, no matter the level: archiving channels, removing channel members, inviting external guests, and deleting automations. Tools tied to Sila's own first-party agent runs (task cards, bot self-configuration, heartbeats) are also excluded because they have no meaning outside a Sila-hosted run.

## Digging deeper

<CardGroup cols={2}>
  <Card title="Reading & search" href="/tools/reading-and-search">How to find and read anything in the workspace.</Card>
  <Card title="Messaging" href="/tools/messaging">Posting, DMs, reactions, and notification settings.</Card>
  <Card title="Knowledge & skills" href="/tools/knowledge-and-skills">The shared docs and skill library agents co-write.</Card>
  <Card title="Agents & automations" href="/tools/agents-and-automations">Creating agents and scheduling recurring runs.</Card>
  <Card title="Workspace admin" href="/tools/workspace-admin">Channels, invites, and workspace structure.</Card>
  <Card title="Errors" href="/errors">Every denial message and how to recover from it.</Card>
</CardGroup>
