dust-call-agent_skill

This skill helps you access Dust agents to read, create, and manage information and tasks across projects, Slack, Notion, or Drive.
  • TypeScript

1.3k

GitHub Stars

1

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill dust-tt/dust --skill dust-call-agent

  • SKILL.md1.5 KB

Overview

This skill lets you call a Dust agent from the command line to fetch context, perform actions, or run tasks using company-wide knowledge. It integrates with Dust agents that have access to Slack threads, Notion pages, Google Drive docs, and internal Dust context. Use it to query, summarize, automate, or continue existing agent conversations non-interactively.

How this skill works

You invoke the Dust CLI with the chat command and specify an agent (-a) and a message (-m). For new conversations, you may scope the request to a project using --projectName or --projectId. To continue an existing conversation, pass the conversation id with -c returned in the initial JSON. Add -d / --details to return full event streams, tool actions, and raw agent payloads.

When to use it

  • Summarize threads, documents, or project state from company sources non-interactively.
  • Create or update artifacts (issues, calendar events, Slack posts) via an agent action.
  • Query Dust-specific context like recent projects, customers, or discussions.
  • Continue a multi-step agent interaction using a conversation id.
  • Collect detailed tool and event traces for debugging with --details.

Best practices

  • Start new workflows with a clear, specific prompt to the agent and include relevant context or links.
  • Use --projectName or --projectId when the task should be performed inside a specific project space.
  • Use -c with the conversation id to keep multi-step interactions coherent and maintain state.
  • Add -d only when you need debugging info or complete agent traces to avoid noisy output.
  • If a tool errors saying login is required, ask the user to sign in manually before retrying.

Example use cases

  • Create an issue: dust chat -a issueBot -m "create an issue for this: ..."
  • Research a topic: dust chat -a deep-dive -m "Research all info we have on kubernetes probe failures in recent weeks."
  • Project summary in a space: dust chat -a prea --projectName "Engineering" -m "summarize recent incidents"
  • Continue a conversation: dust chat -a issueBot -c 'TdWyn4aDt1' -m "also add a subsequent issue about this: ..."
  • Get full traces: dust chat -a prea -d -m "what's the status of project X?"

FAQ

Use -c CONVERSATION_STRING_ID returned in the initial JSON response to continue the same agent conversation.

How do I scope a request to a project?

For new conversations, include --projectName or --projectId. These flags cannot be used with -c for continuing conversations.

What does --details (-d) return?

--details returns raw event streams, tool actions taken by the agent, and the full agent message payload for debugging and audit.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational
dust-call-agent skill by dust-tt/dust | VeilStrat