Providers and agents
OpsCli invokes each tool through its non-interactive native CLI and saves session identifiers to resume interrupted tasks. model, effort, and agent are optional; unset options are omitted.
| Provider | Agent mapping | Model | Effort |
|---|---|---|---|
| OpenCode | Custom agent via --agent |
Optional --model |
Optional --variant |
| Claude Code | Custom agent via --agent |
Optional --model |
Optional --effort |
| Codex | Config profile via --profile |
Optional --model |
model_reasoning_effort via --config |
| Cursor | Mode via --mode (agent, ask, or plan) |
Optional --model |
Unsupported; rejected |
OpenCode
Runs opencode run --format json, passing a configured model, agent, and effort as --model, --agent, and --variant. Resumes sessions with --session <session_id>.
Codex
Runs codex exec --json; resumes with codex exec resume <thread_id>. Effort is passed as --config model_reasoning_effort=<level>. Documented levels include low, medium, high, xhigh, max, and ultra; availability depends on the model and CLI version.
Claude Code
Runs claude -p --output-format stream-json --verbose, passing configured model, agent, and effort through their native flags. Resumes with --resume <session_id>.
Cursor
Runs agent --print --output-format stream-json, passing model with --model and mode with --mode. Effort is not supported.
Permission overrides are also provider-specific (auto_approve for OpenCode, sandbox/auto_review for Codex, permission_mode/permission_prompts for Claude Code, and force/trust for Cursor). If omitted, each CLI keeps its native policy.