aevatar-codex-exec-workflow-sample

Verify Aevatar codex_exec

Mount this skill from Ornn and run exactly one bundled workflow for the configured target:

• codex-exec-check: canonical managed OpenSandbox proof; no caller input.
• codex-exec-private-ssh-check: private node-backed SSH proof; accepts only service and principal routing.

Configuration, health, node-online, and direct SDK/SSH checks are prerequisites, not completion evidence. Report codex_exec as usable only after the selected Aevatar workflow returns exact CODEX_EXEC_READY.

Guardrails

• Fetch the public skill from Ornn. Do not validate only a local workflow copy.
• Keep the fixed probe prompt. Do not replace it with a real task.
• Never place tokens, keys, auth.json, CODEX_HOME, local paths, model flags, images, providers, or sandbox flags in workflow input.
• Do not mix managed and private fields. Managed requires target.kind=managed_sandbox plus workspace.kind=empty_git; private requires nested target.private_ssh and no workspace.
• Run through Aevatar as the NyxID account being verified.

Mount

Call use_skill with workflow mounting enabled:

{
  "skill": "aevatar-codex-exec-workflow-sample",
  "mount_workflows": true
}

Wait for the mount command to be accepted. Read-model visibility can propagate asynchronously. If mounting is unavailable, fetch this public Ornn version and submit the YAML under assets/ as explicit inline draft-run input; state clearly that it was an inline run.

Managed proof

Start the canonical workflow without caller-controlled routing:

{
  "workflow_id": "codex-exec-check",
  "inputs": {
    "prompt": ""
  },
  "wait": "stream"
}

The workflow owns this exact tool payload:

{
  "target": { "kind": "managed_sandbox" },
  "workspace": { "kind": "empty_git" },
  "prompt": "Reply with exactly CODEX_EXEC_READY",
  "timeout_secs": 180
}

Success requires the managed result to contain all of:

• status equal to succeeded
• target equal to managed_sandbox
• output equal to CODEX_EXEC_READY after trimming
• exit_code equal to 0
• a non-empty sanitized diagnostic_id

Treat missing fields, extra model text, or any typed failure as a failed verification.

Private SSH proof

Start the private workflow with exactly the environment-owned service and Unix principal:

{
  "workflow_id": "codex-exec-private-ssh-check",
  "inputs": {
    "prompt": "{\"service\":\"your-service-slug\",\"principal\":\"your-unix-user\"}"
  },
  "wait": "stream"
}

Pass the SSH UserService slug or UUID, never a node ID. Success requires exit_code: 0, timed_out: false, and stdout equal to CODEX_EXEC_READY after trimming.

Diagnose

Preserve the structured error and classify it before changing configuration:

• managed disabled/allowlist: ask the Aevatar operator to enable the target for the authenticated NyxID subject.
• nyxid_binding_required, revoked binding, or llm_proxy_scope_missing: repeat the normal Aevatar/NyxID login consent; never forward the reusable caller bearer into the sandbox.
• managed capacity/provisioning/Credential Vault failures: inspect the deployment and redacted diagnostic ID.
• Landlock/isolation failure: stop; do not weaken the sandbox.
• sandbox_cleanup_failed: treat as an operations incident.
• node_offline, service, target, principal, key, or host-key failures: repair the private NyxID node route.
• Codex PATH, login, Git root, or timeout failures: repair the fixed private runner wrapper.

Use aevatar-codex-exec-node-setup for setup and detailed repair. Never declare readiness from configuration inspection alone.