openclaw/docs/plugins/onepassword.md

summary	read_when	title
Resolve SecretRefs and give agents curated, audited access to 1Password	You want agents to request curated 1Password secrets You want OpenClaw config credentials to resolve from 1Password You need per-secret approval policy and audit history You are configuring a 1Password service account for OpenClaw	1Password

1Password

The bundled onepassword plugin has two independent, opt-in surfaces:

• a managed exec provider that resolves configured SecretRefs during Gateway startup, reload, audit, and apply preflight
• a policy-controlled agent tool that reads a curated set of 1Password fields

Both use the official op CLI and the same service-account token file. Enabling the plugin alone does not expose the agent tool: that surface also requires a configured item registry.

Security model

• Service-account authentication only. The token stays in a local credentials file and is never accepted in openclaw.json.
• Curated agent registry only. Agents can list configured slugs, but the plugin never enumerates a 1Password vault. SecretRef reads are limited to references explicitly stored on registered OpenClaw credential targets.
• Per-slug auto, approve, or deny policy.
• Approval grants expire. A cached value never bypasses current policy.
• Every access attempt is recorded in OpenClaw's shared SQLite state. Audit rows include the supplied reason; keep reasons non-sensitive. The broker never copies a fetched value or the service token into an audit row.
• After the current tool execution, OpenClaw-owned transcript persistence replaces a successful get value with redacted metadata.
• The value is model-visible for that execution. If the model copies it into a later tool call or reply, that separate record is outside this plugin's persistence hook. Keep policies narrow and do not ask the model to echo a value.
• The plugin invokes op once per cache miss. It does not retry rate limits or other failures.
• Each op call runs with a minimal environment that disables 1Password desktop-app integration (OP_LOAD_DESKTOP_APP_SETTINGS=false, OP_BIOMETRIC_UNLOCK_ENABLED=false), so a 1Password app installed on the Gateway host never triggers biometric or macOS permission dialogs.

Give the service account read access only to the vaults and items used by registered SecretRefs and agent-tool slugs.

Before you begin

You need:

• the 1Password CLI (op) installed on the Gateway host
• a 1Password service account with access to the selected items
• a dedicated service-account token file

Enable the bundled plugin:

openclaw plugins enable onepassword

Create the token directory and file under the OpenClaw state directory:

mkdir -p ~/.openclaw/credentials/onepassword
chmod 700 ~/.openclaw/credentials/onepassword
printf '%s' "$OP_SERVICE_ACCOUNT_TOKEN" > \
  ~/.openclaw/credentials/onepassword/service-account-token
chmod 600 ~/.openclaw/credentials/onepassword/service-account-token
unset OP_SERVICE_ACCOUNT_TOKEN

When OPENCLAW_STATE_DIR is set, replace ~/.openclaw with that directory. The plugin warns once when the token file is readable or writable by group or other users.

Configure SecretRefs

Create a secrets apply plan for common model provider keys:

openclaw onepassword secretref setup \
  --anthropic-id op://Automation/Anthropic/credential \
  --openrouter-id op://Automation/OpenRouter/credential \
  --plan-out ./openclaw-1password-secrets-plan.json

Use --provider-key <provider=id> for another model provider, or --target <path=id> for any registered SecretRef credential target. The command requires at least one target and writes a plan. Inspect it, check the local op and token-file prerequisites, then apply and reload:

openclaw onepassword secretref status
openclaw secrets apply --from ./openclaw-1password-secrets-plan.json --dry-run --allow-exec
openclaw secrets apply --from ./openclaw-1password-secrets-plan.json --allow-exec
openclaw secrets audit --check --allow-exec
openclaw secrets reload

Before apply, status can report that the provider itself is not configured yet; prerequisites ready: yes confirms that the trusted op executable and an accepted non-empty token file are ready. After apply, ready: yes confirms both the provider wiring and prerequisites. Missing or unsafe prerequisites produce actionable next steps without printing the token or raw resolver errors.

Manual provider configuration uses the existing plugin id:

{
  plugins: {
    entries: {
      onepassword: { enabled: true },
    },
  },
  secrets: {
    providers: {
      onepassword: {
        source: "exec",
        pluginIntegration: {
          pluginId: "onepassword",
          integrationId: "onepassword",
        },
      },
    },
  },
  models: {
    providers: {
      openai: {
        apiKey: {
          source: "exec",
          provider: "onepassword",
          id: "op://Automation/OpenAI/credential",
        },
      },
    },
  },
}

References use op://<vault>/<item>/<field> or op://<vault>/<item>/<section>/<field>. Vault, item, section, and field names may contain spaces. The setup command stores references that do not fit OpenClaw's shared exec-id grammar in a plugin-local opaque form and decodes them only inside the resolver. Very long references should use stable 1Password IDs; they are shorter and reduce the number of 1Password API requests.

The SecretRef resolver runs at most four op read processes concurrently, disables the 1Password CLI cache so reloads observe rotated values, never uses desktop-app integration, and does not expose an agent tool for arbitrary reads. Before passing the service-account token, both plugin surfaces resolve the executable and reject paths that another local account can replace; Windows ACL verification must also succeed. Check provider wiring and local readiness with:

openclaw onepassword secretref status --json

Configure registered secrets

Add plugin config to openclaw.json:

{
  "plugins": {
    "entries": {
      "onepassword": {
        "enabled": true,
        "config": {
          "vault": "Automation",
          "defaultPolicy": "approve",
          "cacheTtlSeconds": 300,
          "grantTtlHours": 720,
          "opTimeoutMs": 15000,
          "items": {
            "repository-token": {
              "item": "Repository automation token",
              "field": "credential",
              "policy": "approve",
              "description": "Token for repository automation",
            },
            "model-key": {
              "item": "Model provider key",
              "vault": "Agent credentials",
              "policy": "auto",
            },
          },
        },
      },
    },
  },
}

Slugs use lowercase letters, numbers, and hyphens, start with a letter or number, and contain at most 64 characters. A registry can contain up to 32 slugs; descriptions can contain up to 200 characters. field accepts one field label or ID, must not contain a comma, and defaults to credential. An item-level vault overrides the default vault. opBin can set an absolute path to the op executable; otherwise the plugin resolves op from PATH. Item titles must not start with a hyphen.

Use the agent tool

The tool name is onepassword.

List registered slugs:

{ "action": "list" }

The result contains only the slug, description, policy, and whether a standing grant is active. It never contains a secret value and does not query 1Password.

Request one secret:

{
  "action": "get",
  "slug": "repository-token",
  "reason": "Authenticate the requested repository operation"
}

reason is required, must be non-empty, and is limited to 300 characters. A successful get returns the value plus the configured slug, item title, and field label.

The tool schema also declares an internal authorizationNonce parameter. The policy layer injects it after evaluating the request to hand the authorization to the executing tool call. Never set it manually: the policy hook overwrites any supplied value, and an unknown value fails the request.

Policy tiers and approvals

• auto: fetch immediately and audit the request.
• deny: block and audit the request.
• approve: use an unexpired standing grant, or ask a human to allow once, always, or deny.

Allow once authorizes only the current tool call. Allow always writes a standing grant for that agent and slug to SQLite; other agents must receive their own approval. OpenClaw offers allow always only when the caller has a concrete agent identity. The grant expires after grantTtlHours, which defaults to 720 hours. An unresolved or timed-out approval denies the request; the maximum approval wait is 600 seconds. The plugin retains up to 1,024 standing grants; at that bound, the oldest grant is evicted and its agent must approve the next access.

Each evaluated authorization is single-use and is handed to the executing tool call through shared SQLite state, so the handoff also works when more than one plugin instance is active in the gateway process. Unused authorizations expire after the 600-second approval window.

The in-memory cache defaults to 300 seconds and is bounded by the configured slug registry. Set cacheTtlSeconds to 0 to disable it. Policy is evaluated before every cache lookup, and cache hits are audited. Runtime config reloads take effect at each policy and execution boundary; disabling the plugin or removing, denying, or retargeting a slug invalidates pending authorization and cached values.

Inspect status and audit history

Show readiness and registry counts:

openclaw onepassword status

This reports whether the token file exists, whether op resolved and its path, the registered item count, and per-policy counts. It never reads or prints the token or secret values.

Show the 50 most recent audit rows:

openclaw onepassword audit
openclaw onepassword audit --limit 100

Rows are newest first and show timestamp, agent, slug, outcome, an errorCode when the attempt failed, and a truncated reason. The reason is stored as supplied; the broker never adds the fetched value to the audit log.

1Password CLI behavior

Each cache miss runs op item get with the configured item, vault, and exact field selector, JSON output, a bounded timeout, and --cache=false. The child receives only that field rather than the full item. Only OP_SERVICE_ACCOUNT_TOKEN and HOME are present in the child environment.

The plugin makes one attempt. RATE_LIMITED errors should be handled by waiting before a later agent request; the plugin does not create an automatic retry loop.

Error codes

Failed attempts carry one closed error code in the tool result and the audit row.

1Password access errors:

Code	Meaning
TOKEN_MISSING	Token file is missing or empty
OP_NOT_FOUND	op binary could not be resolved
ITEM_NOT_FOUND	Configured item is not in the vault
FIELD_NOT_FOUND	Configured field is not on the item; available labels are listed
RATE_LIMITED	1Password service-account rate limit reached
AUTH_FAILED	Service-account authentication failed
TIMEOUT	op exceeded opTimeoutMs
OP_ERROR	Any other op failure or invalid output

Policy and validation errors:

Code	Meaning
INVALID_ACTION, INVALID_REASON, INVALID_SLUG	Request failed input validation
UNKNOWN_SLUG	Slug is not in the configured registry
TOOL_CALL_ID_MISSING	Call arrived without a tool call id
POLICY_NOT_EVALUATED	No matching authorization for this call; the request was not policy-approved
POLICY_CHANGED	Config changed between approval and execution
GRANT_EXPIRED	Standing grant lapsed before execution
APPROVAL_CANCELLED	The run was aborted while the approval was pending