How long does the integration take?

Change one base URL in an OpenAI-compatible client, or install the Python or TypeScript SDK; the quickstart routes a first request in a few minutes.

Will this lock us into another vendor?

No. The router is Apache-2.0 open source, your provider keys stay yours, and the API is OpenAI-compatible, so leaving is a change of base URL.

What happens when a provider fails?

The router falls back to the next model your policy allows and records the fallback in the trace, so an outage costs a retry, not an incident.

asas-security-ai-safety - Skill

Name: asas-security-ai-safety
Author: asas-erp-saas-1

Imported from asas-erp-saas-1/As (docs/engineering-history/branches/053-fix__archive-snapshot-ci-exclusion-2026-10-16/docs/engineering-history/branches/051-platform-architecture-2026/engineering-history/branch-snapshots/051--platform-architecture-2026/.agents/skills/asas-security-ai-safety/SKILL.md). Install upstream with npx skills add asas-erp-saas-1/As --skill asas-security-ai-safety. Copyright stays with the author.

Security and AI Safety

Threat model

Always consider tenant escape, IDOR, privilege escalation, mass assignment, injection, XSS, CSRF, webhook forgery/replay, secret exposure, PII leakage, financial manipulation, reservation races, API abuse, cross-tenant analytics and prompt/tool injection.

For each material threat record: attack path → control → enforcement layer → test → evidence → residual risk → owner.

Authorization

Deny by default. Authorize server-side before sensitive access. Evaluate tenant/scope before loading protected data. Never trust client-provided role, tenant, organization, branch or approval claims.

AI authority

Intent → Proposal → Tool Request → Policy → Caller Authorization → Domain Service → Transaction → Audit → Event

AI must inherit caller authority and may not widen scope, bypass domain services, mutate authoritative finance/reservation/contract data directly or perform destructive operations without explicit authorization.

Webhooks and integrations

Require authentication/signature verification, replay protection, idempotency, bounded retries, safe error handling and tenant attribution.

Security closure

Do not call a security control complete because code exists. Require adversarial tests and evidence for the enforcement layer that actually protects the asset.

Install asas-security-ai-safety and it becomes one more candidate the router can pick - when it fits.

A listing is a routing target with a manifest: what it does, which domains it covers, what it costs and who publishes it. Once installed it sits beside your own models and tools, is scored like any other candidate for each request, and shows up in the trace when it wins. Ratings come from workspaces that installed it, one per account.

{ "ocm": "1", "id": "asas-erp-saas-1-as-asas-security-ai-safety", "kind": "skill", "name": "asas-security-ai-safety", "description": "Threat-model and verify ASAS tenancy, authorization, RLS, API security, AI tool boundaries and sensitive-data protections.", "publisher": "asas-erp-saas-1", "version": "1.0.0", "capabilities": { "domains": [ "coding", "legal" ], "tags": [ "skill-md", "github" ], "languages": [ "en" ] }, "quality_prior": 0.6, "examples": [ "Threat-model and verify ASAS tenancy, authorization, RLS, API security, AI tool boundaries and sensitive-data protections." ], "primary": false, "metadata": { "source": { "provider": "github", "repository": "https://github.com/asas-erp-saas-1/As", "path": "docs/engineering-history/branches/053-fix__archive-snapshot-ci-exclusion-2026-10-16/docs/engineering-history/branches/051-platform-architecture-2026/engineering-history/branch-snapshots/051--platform-architecture-2026/.agents/skills/asas-security-ai-safety/SKILL.md", "ref": "2672247aab8ce830091db1df83d1b233aa88d17f", "url": "https://github.com/asas-erp-saas-1/As/blob/2672247aab8ce830091db1df83d1b233aa88d17f/docs/engineering-history/branches/053-fix__archive-snapshot-ci-exclusion-2026-10-16/docs/engineering-history/branches/051-platform-architecture-2026/engineering-history/branch-snapshots/051--platform-architecture-2026/.agents/skills/asas-security-ai-safety/SKILL.md", "key": "asas-erp-saas-1/As/docs/engineering-history/branches/053-fix__archive-snapshot-ci-exclusion-2026-10-16/docs/engineering-history/branches/051-platform-architecture-2026/engineering-history/branch-snapshots/051--platform-architecture-2026/.agents/skills/asas-security-ai-safety/SKILL.md" } }, "instructions": "# Security and AI Safety\n\n## Threat model\n\nAlways consider tenant escape, IDOR, privilege escalation, mass assignment, injection, XSS, CSRF, webhook forgery/replay, secret exposure, PII leakage, financial manipulation, reservation races, API abuse, cross-tenant analytics and prompt/tool injection.\n\nFor each material threat record:\n`attack path → control → enforcement layer → test → evidence → residual risk → owner`.\n\n## Authorization\n\nDeny by default. Authorize server-side before sensitive access. Evaluate tenant/scope before loading protected data. Never trust client-provided role, tenant, or", "cost": { "context_tokens": 336 } }

asas-security-ai-safety

About

Security and AI Safety

Threat model

Authorization

AI authority

Webhooks and integrations

Security closure

Use it

Install asas-security-ai-safety and it becomes one more candidate the router can pick - when it fits.

What it costs to adopt - in engineering time.

One base URL

No vendor lock-in

Failover built in

Manifest

Reviews