Skip to content

SecondAgent · Agent Mode

SecondAgent is SecondStack’s agentic mode — it appears as the Cowork Agent inside the chat UI. It turns a chat into a working session with an AI coworker: it runs the Claude Code CLI inside an isolated, per-thread Docker sandbox, equipped with Skills for real tasks — reading and writing files, automating a browser, authoring diagrams and office documents, generating images, and remembering context across a session.

create a hello world program in this workspace
Tools

Done. Created and ran hello_world.py — it prints "Hello, World!" as expected.

hello_world.py
Preview · Download
Attach files
Engineering Wiki
Customer Research
Sales Playbook
+ New Knowledge…
MCP servers
Local Files
Context7
Atlassian Suite
+ Manage MCP…
SSH workspace running
ssh 44a43ef4@localhost -p 2222
Open in VS Code
Auto-stop in 1h · extends on activity
Model
Claude Opus 4.7
Claude Sonnet 4.6
Claude Haiku 4.5
Sandbox files
.plans
hello_world.py 23 B
Type your message here...

The agent works in its own sandboxed workspace — here creating and running a file, then returning it.

  • Each agent session runs the Claude Code CLI inside its own Docker sandbox container, tied to a single chat thread, with an isolated workspace.
  • Sandboxes run as a non-root user with baked, read-only tooling, keeping the runtime constrained.
  • Files move in and out of the sandbox through presigned object storage, and generated artifacts are returned to the chat.

SecondAgent supports two runtimes, which administrators can enable independently:

  • Cloud / server runtime — sandboxes run on the platform’s own infrastructure.
  • Local desktop runtime — a companion desktop app (macOS, Windows, Linux) pairs to the platform and runs sandboxes on the user’s own machine and folders.

Skills give the agent concrete capabilities. SecondStack ships 18+ built-in Skills, and the set is extensible with your own. They include:

  • File operations — read, write, and manipulate files in the workspace.
  • Browser automation — drive a real browser for research and web tasks.
  • Image generation — create and edit images.
  • Diagram authoring — produce diagrams (for example, draw.io).
  • Office documents — author Word, Excel, PowerPoint, and PDF files.
  • Persistent memory — carry context forward within a session.

Additional Skills can be added per user or overlaid for the whole organization.

The agent can run against a Claude subscription (via OAuth) or against API billing through the gateway, depending on the profile an administrator configures.