vault_status

vault_status returns the current state of the ov mcp serve session. Useful for verifying the bridge is running correctly, confirming which app is active, and checking interceptor activity.

---

Parameters

None.

---

Response

{
  "unlocked": true,
  "intercept_mode": "block",
  "intercepted_count": 0,
  "session_age_seconds": 142
}

Field	Description
unlocked	true if the session is authenticated and the KEK is loaded
intercept_mode	Current interceptor mode: block (default) or warn
intercepted_count	Number of potential secrets intercepted this session
session_age_seconds	Seconds since ov mcp serve was started

---

Confirming the bridge is healthy

Call vault_status at the start of a session to confirm the bridge is running before issuing any vault_run calls:

“Let me verify the vault is unlocked before we start.” → calls vault_status → confirms unlocked: true → proceeds with vault_run

---

Checking interceptor activity

If intercepted_count is greater than 0, something triggered the inbound interceptor during this session — either a command that could exfiltrate injected secrets was blocked, or a raw secret pattern was detected in MCP traffic. Check your audit log for details.

---

Verified context reporting

The response also includes a context block reporting the active (app, env), with a source_verified field:

{
  "context": {
    "app": {"slug": "my-saas", "id": "a2f1c3..."},
    "env": {"slug": "production", "id": "b4d2e8..."},
    "source": ".ov.yaml at /Users/hunter/dev/my-saas/.ov.yaml",
    "source_verified": true
  }
}

vault_status resolves (app, env) per call, the same verified pipeline every other tool uses — it does not just echo the daemon’s startup env. source_verified is true only when that per-call resolution succeeded and was verified against the server for this specific call; it’s false on any fallback (unresolved or unverifiable .ov.yaml), in which case the displayed app/env are unverified defaults, not confirmed context. No failure reason is ever included in this field or elsewhere in the response — treat source_verified: false as “don’t trust this context,” not as a diagnostic.