The future control plane for your agents
Planned registry and governance built on top of Guard and CLI.
Capiscio Platform is the planned control plane for your agent ecosystem. It will sit above the Guard and CLI, giving you a central place to register agents, define policies, and understand how agent traffic behaves across your organization.
We are currently in the design phase and building v1 with a small number of partners.
Why teams are asking for a platform
"Where are all our agents?"
Problem: Agents are scattered across repos, teams, and environments. There is no single source of truth. New hires spend weeks just discovering what exists.
→ Platform is being designed as a central registry for agents and their metadata, anchored to A2A Agent Cards and Guard events.
"Which agents are actually safe to use?"
Problem: No easy way to see which agents are passing agent identity/integrity checks, which are failing, and how that changes over time.
→ Platform will aggregate Guard and CLI results so you can see pass/fail trends, reasons for failures, and which services are at risk.
"Manual governance doesn't scale."
Problem: Spreadsheets, email threads, and tribal knowledge are being used to approve or deprecate agents. There is no audit trail.
→ Platform is intended to provide basic policy and approval workflows so you can say "only agents with these properties may talk to this system" and prove it later.
How Platform will connect to Guard and CLI
Platform is not a new SDK. It is designed to consume events from the tools you already use: the Guard, the Go sidecar, and the CLI.
An opt in mode where capiscio validate can register or update agents in a central registry as part of dev and CI workflows.
An optional channel for Guards and sidecars to emit anonymized "pass/fail + reason" events, so you can see which agents are failing checks and why.
Workspaces where developers, leads, and compliance teams can see the same view of agents, approve changes, and export audit reports.
Trust badges and public registry views for teams that want to show that a given agent is being checked and monitored.
Guard and CLI will always work standalone. Platform integration will be opt in.
What we're planning to build
The exact feature set will be shaped with design partners, but these are the core areas we're aiming at.
Central index of agents, their A2A Agent Cards, owners, and environments.
Rules about which agents may talk to which services, and what properties they must have (signing keys, environments, labels).
Aggregated view of Guard / CLI events over time: when checks pass, when they fail, and why.
Optional public registry views and badges to signal that specific agents are being checked and governed.
Start with the CLI today
The CLI is available now. Use it to validate A2A Agent Cards and test live endpoints in dev and CI. Platform will build on this data later — nothing you do today is wasted.
Explore CLIEnforce with the Guard
The Guard and Go sidecar enforce agent identity, integrity, and freshness in production today. Platform is being designed to consume these events and surface them in one place.
Explore GuardHelp shape the Platform
We're looking for a small group of teams that:
You are:
- ✓ Already running agents in production or staging
- ✓ Care about agent identity, integrity, and governance
- ✓ Willing to share real requirements and workflows
You'll get:
- ★ Direct influence on the feature set and priorities
- ★ Early access to prototypes and internal builds
- ★ Favorable commercial terms when Platform becomes generally available