Code Execution & Workspace Tools

Tools for Code Execution & the Developer Workspace

Give agents a real, governed workspace: run code in a sandbox, read and patch files, execute tests and builds, and drive Git end to end — from status to commit to pull request — all inside your perimeter.

Browse all tools
18Prebuilt tools in this category
100%On-premise & sovereign-cloud ready
Any agentAssignable under role-based policy
AuditedEvery action logged & traceable
Code Execution & Workspace Tools

18 tools in this category

These tools turn an agent from a suggester into a doer. Assign sandboxed execution, file operations, test and build runners, and the full Git action set so an engineering agent can implement, verify, and ship changes — under role-based policy and full audit logging.

code-execution

Sandboxed Code Execution

Run agent-written code safely in an isolated sandbox.

Code Execution & Workspace Tools Explore tool code-execution

Terminal Execute

Run governed shell commands in the agent workspace.

Code Execution & Workspace Tools Explore tool code-execution

File Read

Read a file from the agent workspace.

Code Execution & Workspace Tools Explore tool code-execution

File Write

Create or overwrite a file in the workspace.

Code Execution & Workspace Tools Explore tool code-execution

File List

List files and folders in the workspace.

Code Execution & Workspace Tools Explore tool code-execution

File Search

Find files and text across the workspace fast.

Code Execution & Workspace Tools Explore tool code-execution

File Patch

Apply a precise, reviewable edit to a file.

Code Execution & Workspace Tools Explore tool code-execution

Test Runner

Run the test suite and report pass/fail.

Code Execution & Workspace Tools Explore tool code-execution

Build Runner

Compile or build the project and surface errors.

Code Execution & Workspace Tools Explore tool code-execution

Lint Runner

Check code style and catch issues before review.

Code Execution & Workspace Tools Explore tool code-execution

Git Status

See what changed in the working tree.

Code Execution & Workspace Tools Explore tool code-execution

Git Diff

Show exactly what an agent changed.

Code Execution & Workspace Tools Explore tool code-execution

Git Branch

Create and switch branches for isolated work.

Code Execution & Workspace Tools Explore tool code-execution

Git Commit

Stage and commit changes with a clear message.

Code Execution & Workspace Tools Explore tool code-execution

Git Push

Publish a branch to the remote, under policy.

Code Execution & Workspace Tools Explore tool code-execution

Git Create Pull Request

Open a pull request from the agent’s branch.

Code Execution & Workspace Tools Explore tool code-execution

Git Create Issue

File a tracked issue straight from the workspace.

Code Execution & Workspace Tools Explore tool code-execution

Git Add Comment

Comment on a PR or issue with findings.

Code Execution & Workspace Tools Explore tool
In depth

Code Execution & Workspace Tools on VDF AI

A real, governed developer workspace for AI agents

To ship code, an agent needs to do more than suggest a diff — it needs to run things. This category gives an agent a full workspace: sandboxed code execution and a terminal to run commands safely; file operations to read, write, list, search, and patch source; a test runner, build runner, and linter to verify changes; and the complete Git action set — status, diff, branch, commit, push, and pull request creation. With these, an engineering agent can implement a change, run the tests, and open a PR the way a developer would.

Execution that is sandboxed, policy-bound, and on-premise

Letting an agent execute code is only safe if execution is contained and controlled. On VDF AI every command runs in a governed sandbox under role-based policy with full audit logging, and the whole workspace runs on-premise or in your sovereign cloud, so your source and build environment never leave your perimeter. Pair these tools with the code intelligence category — where an agent reviews and documents — to build a governed VDF AI Network that reads, reviews, implements, verifies, and ships a change end to end.

FAQ

Code Execution & Workspace Tools — frequently asked questions

What code execution and workspace tools can agents use?

Agents can be assigned sandboxed code execution, a terminal, file read/write/list/search/patch operations, test, build, and lint runners, and the full Git action set from status and diff through commit, push, and pull request creation — a complete developer workspace.

Is it safe to let an AI agent execute code?

Yes, when execution is contained. Code runs in a governed sandbox under role-based policy with full audit logging, so an agent can run commands, tests, and builds without unrestricted access to your systems, and every action is traceable.

Can an agent run tests and open a pull request?

Yes. Assign the test runner and Git tools and an agent can implement a change, run the test suite to verify it, commit the work, and open a pull request — the same loop a human engineer follows, under policy and logging.

How is this different from the development tools category?

The development tools understand and document code — review, security scan, dependency and architecture analysis. The code execution tools act on it — running, editing, testing, and shipping. Agents are commonly assigned tools from both so they can review a change and then implement it.

Does agent code execution happen in our environment?

Yes. The sandbox and workspace run on-premise or in your sovereign cloud, so source code, build artifacts, and command output stay inside your perimeter and never reach a third-party runtime.

How do I assign execution tools to an agent?

Attach the tools to an engineering agent from its toolkit under role-based policy. The agent can then execute, verify, and ship within a workflow, and several agents can be orchestrated as a VDF AI Network that delivers changes end to end.

Keep exploring

Other tool categories

Assign code execution & workspace tools to your agents

See these tools assigned to agents and orchestrated together as a governed, on-premise network.

Explore VDF AI Networks