@veris-ai/veris-sim-opencode
v0.7.0
Published
Test an application against Veris - stateful twins of the external services it calls. Three commands: setup (wire a repo once), build <issue|prompt> (measure the vendor before designing, prove the change through veris-proxy), fix <issue|prompt> (reproduce
Readme
@veris-ai/veris-sim-opencode
Veris for OpenCode: three
engineer-invoked commands plus the veris MCP server. Veris runs stateful
twins of the external services an application calls; veris-proxy reroutes
the application's outbound HTTP(S) into a sandbox from outside the process,
so the code under test keeps its production hostnames, credentials and
client stack, and ends each run with a receipt of what the sandbox received.
Install
opencode plugin @veris-ai/veris-sim-opencode -g(or add "@veris-ai/veris-sim-opencode" to the plugin array in
~/.config/opencode/opencode.json). Then export your credential and restart
opencode:
export VERIS_API_KEY=<key>VERIS_API_BASE overrides the control plane (defaults to
https://svc.api.veris.ai). Until the key is set, opencode mcp list shows
the veris server as failed; the commands still run and setup walks you
through it.
Commands
| command | what it does |
|---|---|
| /veris-sim:setup | wires the repository to a Veris environment, once — credential, environment, transport, one smoke run with proof of arrival |
| /veris-sim:build <issue link \| prompt> | measures every vendor claim the task rests on against the twin before designing, proves the change through veris-proxy with a receipt |
| /veris-sim:fix <issue link \| prompt> | reproduces the failure the issue describes before designing, proves the same failure closed with a receipt |
The commands are plain OpenCode commands: only the engineer can start them, the model cannot invoke them on its own.
Source and the shared skill files: https://github.com/veris-ai/plugins
