AEGIS Advisor
Interactive governance decision helper — answer plain-language questions and get an AEGIS decision without writing JSON or choosing parameter values.
Try it now: AEGIS Advisor — 8-step wizard, no sign-up required, runs against the live production API.
What it is
The AEGIS Advisor is a guided, 8-step decision-support wizard for users who want an AEGIS governance evaluation but don't want to learn the parameter model, write JSON, or configure thresholds themselves. It asks plain-language factual questions about your decision (what domain, what's changing, what the baseline looks like, how much is at stake) and translates your answers into the quantitative parameters that the six AEGIS gates consume.
Behind the scenes it calls the same production /evaluate endpoint as the SDK and MCP server — so the decision it produces is the same decision you'd get from a fully-configured programmatic integration, just with the parameter-derivation step automated.
When to use it
| Situation | Use Advisor? |
|---|---|
| First-time exploration of AEGIS before integrating the SDK | Yes — fastest path to seeing a real decision |
| Non-engineering stakeholder needs a decision and doesn't have an API key | Yes — no sign-up required |
| One-off decision that doesn't warrant writing code | Yes |
| Programmatic integration in a CI pipeline | No — use the SDK, REST API, or GitHub Action |
| Batch processing many proposals | No — use the REST API directly |
How it maps to the six gates
The wizard collects answers across 8 steps:
| Step | Question theme | Derived parameter(s) |
|---|---|---|
| 1 | Domain selection (Investment, Hiring, Engineering, Content, Operations, Other) | Domain-specific parameter templates and default thresholds |
| 2 | Proposal summary (free-text one-liner) | proposal_summary |
| 3 | Blast radius / scope of affected systems | complexity_score baseline, estimated_impact |
| 4 | Current baseline risk (plain-language: low / medium / high) | risk_baseline |
| 5 | Proposed risk after the change | risk_proposed, drives Risk gate |
| 6 | Expected benefit / profit proxy | profit_baseline, profit_proposed, drives Profit gate |
| 7 | Novelty / how different this is from prior decisions | novelty_score, drives Novelty gate |
| 8 | Quality confidence / evidence level | quality_score, drives Quality gate |
The Utility gate is computed automatically from the combined risk/profit deltas; you don't answer a direct question for it. The Complexity gate uses the scope answer (step 3) plus a safety floor that can never be overridden.
What you get back
A full AEGIS decision object with:
- Status:
proceed,pause,halt, orescalate - Confidence: Bayesian posterior probability
- Per-gate breakdown: which of the six gates passed or failed, with the specific threshold each was measured against
- Rationale: human-readable explanation of why the decision came out the way it did
- Next steps: concrete, domain-specific recommendations
The result is NOT persisted to your AEGIS customer account (the Advisor uses the sandbox evaluation path, which is rate-limited to 10 requests per IP per day). If you want an audited, persistent decision, sign up at portal.undercurrentholdings.com and call the authenticated /evaluate endpoint instead.
Privacy
- Your answers are sent to the AEGIS production API for evaluation and then discarded.
- No account required — the Advisor uses the sandbox path (no customer ID, no audit log entry).
- The Advisor is a client-side SPA — everything runs in your browser until you click "Get decision."
Links
- Open the Advisor: aegis.undercurrentholdings.com/advisor/
- Full SDK: Python SDK Quickstart
- Full REST API: REST API Reference
- MCP server: MCP Tools
- GitHub Action: Action Reference
Guardrail Framework Governance
Governance procedures for the Hardened Quantitative Guardrail Framework including approval workflows, recalibration processes, and dual-control requirements.
AEGIS Roadmap
Single source of truth for AEGIS future work including active PRs, open issues, and release milestones.