Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →lloydchang/openswe-code-review
Advanced code review using OpenSWE Deep Agents with sandbox execution for security analysis, performance testing, and automated fixes.
Passed 19 of 22 test cases with the skill, and 7 of 22 without it.
and 10 more in the full report
12 fixed7 passed either way3 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 12 test cases the skill fixed. Same model, same prompt, with and without the skill.
What exact tool name in the OpenSWE tool schema enables the sandbox agent to commit code changes and open a pull request directly?
| Checked | Without skill | With skill |
|---|---|---|
| The tool name is 'commit_and_open_pr'. | Missed | Passed |
The instructions your model reads when the skill loads.
This skill provides advanced code review capabilities using OpenSWE Deep Agents with sandbox execution for comprehensive analysis, security scanning, and automated issue resolution.
| Field | Type | Description |
|---|---|---|
layer | string | Must be "openswe" for OpenSWE skills |
execution_environment | string | "sandbox" for isolated execution |
integration_points | []string | Supported integrations (github, slack, linear) |
sandbox_config | object | Sandbox provider and resource configuration |
openswe_config | object | OpenSWE-specific configuration |
sandbox_config:
provider: modal|daytona|runloop|langsmith
timeout: 1800 # seconds
resources:
cpu: 2
memory: 4Gi
gpu: optional
disk: 20Gi
network_access: true
persistent_storage: false
openswe_config:
agent_type: "deep-agent" # Currently only deep-agent supported
model: "anthropic:claude-opus-4-6" # Model specification
middleware: # List of middleware hooks
- GitOpsSafetyMiddleware
- ToolErrorMiddleware
test cases passed, vs 7 / 22 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.