code-mode

Code-Mode provides a library/client that lets AI agents execute tool workflows by running TypeScript code in a sandboxed VM and invoking tools registered from UTCP/MCP/HTTP/File/CLI sources. It also offers dynamic tool discovery and auto-generated TypeScript interfaces for IDE/agent guidance.

Evaluated Mar 30, 2026 (22d ago)
Repo ↗ DevTools ai-agents tool-calling mcp utcp code-execution sandboxing typescript python sdk
⚙ Agent Friendliness
62
/ 100
Can an agent use this?
🔒 Security
49
/ 100
Is it safe for agents?
⚡ Reliability
28
/ 100
Does it work consistently?

Score Breakdown

⚙ Agent Friendliness

MCP Quality
55
Documentation
70
Error Messages
--
Auth Simplicity
55
Rate Limits
20

🔒 Security

TLS Enforcement
70
Auth Strength
45
Scope Granularity
20
Dep. Hygiene
50
Secret Handling
60

README claims secure VM sandboxing, no filesystem access, and 'zero network access' from the VM, plus timeout protection and console capture. However, the provided content does not clearly document secret logging practices inside the client/VM, the exact sandbox/network enforcement guarantees, or fine-grained permission/scopes for registered tools.

⚡ Reliability

Uptime/SLA
0
Version Stability
35
Breaking Changes
20
Error Recovery
55
AF Security Reliability

Best When

You want to reduce multi-tool iteration overhead by having the agent generate one coherent TypeScript program that calls registered tools, while relying on UTCP/MCP/HTTP/etc tool registrations.

Avoid When

You cannot tightly control which tools are registered or you need strict separation between model reasoning and executable code execution semantics.

Use Cases

  • Agent tool orchestration via single code execution (multi-step workflows)
  • Dynamic tool discovery and interface introspection for adaptive agent behavior
  • Integrating MCP servers (and other tool sources) into agent workflows with TypeScript interfaces
  • Enterprise-safe code execution with timeouts and sandboxing
  • Generating TypeScript interface definitions for available tools

Not For

  • Running untrusted arbitrary code without strong sandboxing guarantees/verification
  • Use-cases requiring strictly REST-style function-calling (no code execution step)
  • Environments needing fine-grained auditability of every individual tool call rather than aggregated execution logs

Interface

REST API
No
GraphQL
No
gRPC
No
MCP Server
Yes
SDK
Yes
Webhooks
No

Authentication

Methods: Registered tool credentials via UTCP/MCP/HTTP/CLI configuration (e.g., env vars passed to MCP server commands)
OAuth: No Scopes: No

Auth is described indirectly via tool/server configuration (e.g., passing a personal access token via environment variables to a registered MCP server). No first-party auth mechanism (API keys/OAuth) is documented in the provided README content.

Pricing

Free tier: No
Requires CC: No

No pricing information is provided in the supplied content.

Agent Metadata

Pagination
none
Idempotent
False
Retry Guidance
Not documented

Known Gotchas

  • Because the agent runs generated TypeScript code, ensure the sandbox and registered tools are appropriately constrained to prevent unsafe actions.
  • Tool calls occur inside code; failure modes may be harder to localize than per-tool JSON calls unless logs are carefully inspected.
  • Idempotency of tool operations depends on the underlying registered tools/APIs; Code-Mode itself does not document idempotency controls.

Alternatives

Full Evaluation Report

Comprehensive deep-dive: security analysis, reliability audit, agent experience review, cost modeling, competitive positioning, and improvement roadmap for code-mode.

AI-powered analysis · PDF + markdown · Delivered within 30 minutes

$99

Package Brief

Quick verdict, integration guide, cost projections, gotchas with workarounds, and alternatives comparison.

Delivered within 10 minutes

$3

Score Monitoring

Get alerted when this package's AF, security, or reliability scores change significantly. Stay ahead of regressions.

Continuous monitoring

$3/mo

Scores are editorial opinions as of 2026-03-30.

8642
Packages Evaluated
17761
Need Evaluation
586
Need Re-evaluation
Community Powered