Alice · Agent profile · DEMO-AGT-005
Red-Team Probe
Level 2 on public data only, using the high-risk research model in isolation. Any tool use, record write, or external send is blocked by envelope.
Identity and purpose
- Purpose
- Adversarial testing of prompts and guardrails on public material
- Approved purposes
- red-team
- Workflows
- security-testing
- Owner
- Security Lead
- Status
- active
- Runs inside
- Scheduling optimizer
- Registered
- 2026-03-20 · last authority review 2026-08-10
Evaluation on file
- Status
- pass
- Suite
- alice-eval-v1 · containment
- Run
- 2026-08-10
- Score
- 100%
- Notes
- Synthetic demo result. Loopback-only; never given tools (mirrors the kernel's no_tools / loopback_only restrictions).
Autonomy envelope
Human-agent team (L2): The agent performs bounded internal actions inside its envelope. Anything with external effect, elevated risk, or irreversibility is held for a human. Human role: Directs the work; approves exceptions.
PERMITTED DATA
- PUBLIC
PERMITTED ACTIONS
- Analyze
PERMITTED TOOLS
- none
ELIGIBLE MODELS
- Research red-team model (DEMO-MDL-RES-01)
HUMAN APPROVAL REQUIRED FOR
- Publish externally
- Send outside the boundary
- Execute a tool with side effects
- Write to a system of record
- any action at risk ≥ MEDIUM
- any irreversible action
Approver: Security Lead
AUDIT
- Trust Receipt required: yes
- Reviewer recorded on receipt: yes
- Retention: 3 years
Decisions · synthetic day 2026-09-12
Outcomes
Governance · computed
- Decisions
- 2 (1 allow · 0 review · 1 block)
- Envelope blocks
- 1
- Median review time
- No reviews decided
- Receipts
- 2 issued · 0 with evidence
Business value · synthetic measurement
No workflow measurement recorded yet.
Alice on this site runs over the synthetic Lakeview General Hospital environment. Verdicts, checks, and receipts are computed by the Nexus decision engine on synthetic requests; nothing here is a customer, a patient, or a production deployment. Signed-in organizations see their own agents, decisions, and receipts.