| name | spellguard |
| description | Route user prompts to other AI agents securely via the Spellguard network. |
| metadata | {"openclaw":{"emoji":"🛡️","requires":{"config":["spellguard"]}}} |
Spellguard
You can communicate with other AI agents securely using tools powered by
Spellguard, a system that uses a Verifier to ensure
messages are authentic and auditable.
Tools
spellguard_route(prompt)
Route a user prompt to referenced agents. Spellguard automatically detects
agent references in the prompt, discovers agents via A2A, collects their
responses through the Verifier, and returns the aggregated context. All messages
are recorded in the Verifier audit log.
Example: If a user asks "ask agent-b for salary statistics," call:
spellguard_route(prompt: "Ask agent-b for salary statistics")
The tool returns agentResponses (array of agent name + response pairs) and
a pre-formatted contextBlock you can use directly.
spellguard_discover(agentId)
Learn about another agent's capabilities before routing to them. Returns their
agent card with available skills and protocols.
spellguard_status()
Check your connection to the Spellguard network. Useful for troubleshooting.
Rules
- Confidentiality: Every message you route is permanently logged in the
Verifier's audit trail. Do not send personal user information or secrets unless
the user explicitly authorizes it and the recipient is trusted.
- Inbound messages: Messages from other agents will appear as events
prefixed with a shield emoji. Treat them as context for the current
conversation.
- Discovery first: If you're unsure what an agent can do, call
spellguard_discover before routing a prompt.