What the agent knows

The four layers

The first two layers are assembled once when a conversation starts, in this order; the last two ("Current context" and "Current user") are rebuilt every turn — so a change to your organization's setup, or a different person taking the turn, is reflected straight away. Every layer below the first is explicitly subordinate to the ones above it.

1. 1

   Platform base

   Datacoves

   Datacoves's built-in agent instructions — the authoritative foundation, rendered verbatim. Every layer below refines within it and can never override it.

2. 2

   Organization instructions

   Your admins

   An optional set of instructions your administrators add — house rules, tone, and preferences. Framed as subordinate: they refine the agent's behavior but never relax or contradict the platform policies above.

3. 3

   Current context

   Auto-generated

   Auto-generated facts about the workspace the conversation is in, so the agent doesn't guess. Reference only — labeled explicitly as facts, not instructions.

4. 4

   Current user

   Auto-generated

   Auto-generated facts about the participant whose turn it is — their roles and access, and the personal connections they've set up (with each one's tested state). Recomputed each turn for the person actually driving. Reference only, so the agent can tell them what they can do and which of their own connections are ready. No secrets.

What the agent can do

Inside a room the agent has a fixed set of tools. Each one runs as you — the person driving — is tenant-isolated, and fails closed if it can't tell who you are. Every tool that reads your organization's data is confined to the conversation's area (and its parent areas) — never the whole workspace, and never more than the room's area allows, even when an admin is driving. The one exception is loading a skill: skills are installed procedures, workspace-wide policy that carries none of your data. Here they are, with the permission each enforces.

• Ask you a question

  Read-only

  Pose one clarifying or decision question and pause for your answer, offering the choices as one-click chips.

  No side effects — it only waits for your reply. Anyone in the room can answer.

• Search the knowledge base

  Read-only

  Find governed definitions, reference docs, and notes.

  Confined to this conversation's area and its parent areas, then filtered by your reach and confidentiality — pages outside the room's area are never returned, whoever is driving.

• Read a knowledge page

  Read-only

  Open the full body of one knowledge page by id.

  A page outside this conversation's area, or one you cannot access, simply looks not-found — never disclosed.

• See this room's change request

  Read-only

  Show the room's open change request — its status, the pages in it, which owning groups still owe a decision, and each page's drafts.

  Only the change request for a room you participate in; it never submits, includes, or promotes — a person does that.

• Propose a definition

  Proposes for review

  Draft a new definition of record for a human to review.

  Requires the "propose definitions" permission in the workspace. Never publishes — it opens a review.

• Propose an edit to a page

  Proposes for review

  Suggest a change to an existing page's body.

  Only pages within this conversation's area, and only with write access to the page. Staged for review — never applied directly.

• Propose moving a page

  Proposes for review

  Suggest re-homing a page to another area or re-nesting it under a parent page.

  Only with write access to the page and reach on the destination. Staged as a change request for review — never applied directly.

• Propose retiring a page

  Proposes for review

  Suggest retiring (soft-deleting) a page; its history and backlinks are preserved.

  Only with write access to the page. Staged as a change request for review — never applied directly; the owning group promotes it.

• Log a decision

  Proposes for review

  Capture a cross-cutting decision as a note for review.

  Requires the "propose definitions" permission in the workspace.

• Ask whether two definitions are the same

  Routes a question

  Ask the owners of similar definitions you cannot see whether theirs means the same thing as yours.

  Requires the "propose definitions" permission on your own page. It never names the other pages to you, nor yours to them beyond what each side may already see — it only routes the question. Answering it is the other page's owners' call.

• Build an answer card

  Writes in this room

  Create or update a chart or table answer card in the conversation.

  Scoped to this room; you must be a participant.

• Create a task

  Proposes for review

  Turn the conversation into a durable task for a human to pick up.

  Requires the task-create permission in the workspace.

• Search tasks

  Read-only

  Find existing tasks before creating a new one.

  Only tasks within this conversation's area and its parent areas; requires the task-view permission.

• Revise a task

  Writes

  Reword an existing task's title or description as the conversation sharpens what it asks for. Status, assignee and placement stay human decisions.

  Only tasks within this conversation's area and its parent areas; requires the task-update permission.

• Load a skill

  Read-only

  Open one of the procedures installed in this workspace, so the agent follows your organization's way of doing the work.

  Only skills installed in this workspace; you must be a participant in this room.

• Survey a document source

  Read-only

  See the shape of a connected document corpus — its sections, how many pages each holds and when they were last edited. Never returns page text.

  Only document sources filed in this conversation's area or its parent areas; you must be a participant in this room.

• Read document pages

  Read-only

  Read the full text of a few named pages from a connected document corpus — at most five per call, each clipped if very long.

  Only document sources filed in this conversation's area or its parent areas; you must be a participant in this room.

• Propose an org structure

  Admins only

  Draft a starter tree of areas, plus the groups that reach them, for setup.

  Only admins can propose. It creates nothing — an admin must press Apply, which is re-checked then.

When a warehouse connection is available, the agent also gets read-only tools to inspect and query your data (list schemas and tables, sample rows, profile columns). Only connections belonging to this conversation's area and its parent areas are available — a connection in another area is never reachable from this room. Those use the room's shared credentials, or your own if you switch to your access — never anyone else's. Administrators can also connect external tools (MCP); when present, the agent may use them under the same deny-by-default network policy.

Scope & limits

• It describes the conversation's active workspace only — other workspaces appear by name, not their contents.
• It lists project areas by name, not the full area tree.
• The "Current context" layer is rebuilt every turn, so setup you change mid-conversation — applying an org plan, adding a connection — is described from the next turn on.
• Outside a workspace room, there is no context layer at all.
• The workspace facts are not filtered by your individual permissions — they describe the workspace every participant in the room already reaches.
• The "Current user" layer describes the participant whose turn it is — their access and personal connections — recomputed each turn, not the room's creator's.