Section 1: Trust + Capability
Why this section matters competitively
Seraph's biggest strategic gap versus OpenClaw, IronClaw, and Hermes is not vision. It is operational credibility. This section exists to close the gap between a compelling guardian product and a runtime people can trust with real work.
Current state in Seraph
- strong product thesis and proactive architecture
- limited execution plane
- early sandboxing and timeout controls
- weak approval and policy model
- limited secret and trust-boundary handling
- single primary model/provider path
Target end state
Seraph can act with broader power while remaining auditable, governable, and meaningfully safer than a typical tool-calling shell.
Key capabilities
- tool policy profiles and approvals
- secret scoping and safer credential handling
- stronger execution isolation
- richer shell and browser execution
- workflow orchestration
- provider routing and fallbacks
- evaluation and observability for agent behavior
Dependencies and risks
- touches backend runtime, tools, settings, and security assumptions
- can sprawl if execution breadth grows faster than guardrails
- may tempt the project into platform work that weakens the guardian product identity