Works with the AI Agents you already use
Threat Modeling
Map what could go wrong before you build it with evidence backed STRIDE threat modeling
Free
Threat Modeling
Example session with this skill installed
Threat model this new payments architecture before implementation. Users authenticate through OIDC. A public API creates payment intents. PostgreSQL stores tenant and transaction records. A webhook worker receives payment provider events. Object storage contains generated receipts. Build the system model, identify trust boundaries, apply STRIDE, prioritize the main threats, and show what should happen next.
- Read your context and instructions
- Compiled the threat modeling
System Model Evidence Gate
The following architecture was supplied directly in the request.
| Component | Evidence status |
|---|---|
| OIDC authentication | USER SUPPLIED |
| Public payments API | USER SUPPLIED |
| PostgreSQL tenant and transaction store | USER SUPPLIED |
| Payment provider webhook worker | USER SUPPLIED |
| Receipt object storage | USER SUPPLIED |
| Specific cloud provider | UNKNOWN / MISSING |
| API gateway | UNKNOWN / MISSING |
| Existing WAF | UNKNOWN / MISSING |
| Existing secrets manager | UNKNOWN / MISSING |
No missing component is treated as an established architectural fact.
System Model
Assets
- User identities and authenticated sessions
- Tenant boundaries
- Payment intent data
- Transaction records
- Payment provider webhook events
- Generated receipts
- Authorization state
Actors
- Authenticated application user
- Payment provider
- External unauthenticated client
- Application services
- Operations or administrative identities, if present
Administrative identities are not described in the supplied architecture, so their exact capabilities remain UNKNOWN.
Primary Trust Boundaries
TB-01
Internet client to public payments API.
TB-02
OIDC identity context to application authorization logic.
TB-03
Payment provider to webhook processing boundary.
TB-04
Application services to PostgreSQL.
TB-05
Application services to receipt object storage.
Threat Register
TM-01 Cross tenant payment access
STRIDE
Elevation of privilege / Information disclosure
Scenario
An authenticated user manipulates a payment or tenant scoped identifier and reaches another tenant's payment data.
Evidence basis
Tenant records and a public API are USER SUPPLIED.
The exact authorization implementation is UNKNOWN.
Priority
HIGH, TENTATIVE
Required response
Define explicit tenant authorization requirements and negative cross tenant verification cases.
Existing control
UNVERIFIED
Owner
UNASSIGNED
TM-02 Forged or replayed payment webhook
STRIDE
Spoofing / Tampering
Scenario
An attacker submits or replays a webhook event that the worker accepts as a legitimate payment provider message.
Evidence basis
The webhook worker and payment provider are USER SUPPLIED.
The signature verification and replay protection mechanisms are UNKNOWN.
Priority
HIGH, TENTATIVE
Required response
Define authenticity, freshness, replay protection, and idempotency requirements for provider events.
Existing control
UNVERIFIED
Owner
UNASSIGNED
TM-03 Receipt object exposure
STRIDE
Information disclosure
Scenario
Receipt objects become accessible across tenant boundaries or through overly broad object access.
Evidence basis
Receipt object storage is USER SUPPLIED.
Bucket policy, object authorization, URL design, and retention controls are UNKNOWN.
Priority
HIGH, TENTATIVE
Required response
Define access control and data exposure requirements for receipt retrieval.
Existing control
UNVERIFIED
Owner
UNASSIGNED
TM-04 Identity context substitution
STRIDE
Spoofing / Elevation of privilege
Scenario
Application authorization relies on identity or tenant attributes that can be substituted, confused, or trusted from an inappropriate source.
Evidence basis
OIDC authentication is USER SUPPLIED.
Token validation rules and authorization claim handling are UNKNOWN.
Priority
HIGH, TENTATIVE
Required response
Define trusted identity sources, token validation requirements, and server side authorization boundaries.
Existing control
UNVERIFIED
Owner
UNASSIGNED
Evidence Gaps
The supplied architecture does not establish:
- the cloud platform
- network segmentation
- administrative identity model
- secret storage mechanism
- existing rate limiting
- API gateway behavior
- webhook signature mechanism
- object storage access policy
- transaction encryption configuration
- logging and monitoring controls
These remain UNKNOWN rather than being assumed.
Recommended Next Workflow
Route the prioritized threats into security-requirements.
Convert TM-01 through TM-04 into explicit security properties, acceptance criteria, and evidence expectations before implementation and focused verification begin.
Connects securely to your tools. The creator never sees your data.
About this skill
Builds an evidence backed security model before implementation or active testing begins. It identifies assets, actors, data flows, trust boundaries, entry points, and abuse cases, then applies STRIDE to produce a prioritized threat register with explicit evidence status and response direction.
What it does
- Builds a structured system model from architecture and project evidence.
- Identifies assets, actors, entry points, data flows, and trust boundaries.
- Applies STRIDE across the modeled attack surface.
- Records abuse cases and security consequences.
- Prioritizes threats according to the available context.
- Distinguishes verified architecture from user supplied, inferred, hypothetical, and missing information.
- Prevents undocumented infrastructure from silently becoming fact.
- Records unknown ownership and controls explicitly rather than inventing them.
- Creates the foundation for security requirements, review, testing, and later release evidence.
System Model Evidence Gate
Every relevant component must be classified according to its evidence status:
- VERIFIED
- USER SUPPLIED
- INFERRED
- HYPOTHETICAL EXAMPLE
- UNKNOWN / MISSING
The workflow does not silently assume multi tenancy, OAuth, API gateways, databases, queues, cloud services, or other architecture that is not supported by the available evidence.
When architecture evidence is insufficient, the result remains an evidence gap model or clearly identified reference model rather than presenting invented infrastructure as fact.
How it works
- System scope identifies the product, feature, service, or architecture being modeled.
- Evidence classification records which architecture facts are established and which remain uncertain.
- Asset identification captures security sensitive data, identities, capabilities, and business operations.
- Actor mapping identifies trusted and untrusted principals.
- Data flow modeling traces how information crosses components and trust boundaries.
- Entry point analysis identifies externally or internally reachable surfaces.
- STRIDE analysis evaluates spoofing, tampering, repudiation, information disclosure, denial of service, and elevation of privilege where applicable.
- Abuse case analysis converts plausible attacker goals into concrete threat scenarios.
- Prioritization ranks threats while preserving uncertainty where evidence is incomplete.
- Handoff identifies required security requirements, reviews, testing, or architecture decisions.
Evidence discipline
A component that is not established by the supplied evidence is not promoted to a verified architectural fact.
Existing controls are not assumed.
Concrete owners are not invented.
Threats that depend on uncertain architecture remain appropriately qualified.
This keeps the threat model useful without creating false certainty.
Boundaries
Threat Modeling identifies what could go wrong and what security properties require attention.
Testable acceptance criteria belong to security-requirements.
Static implementation review belongs to the appropriate review specialist.
Breadth first vulnerability discovery belongs to vulnerability-assessment.
Exploitability proof belongs to exploit-validation.
Authorized active intrusion testing belongs to the appropriate penetration testing specialist.
Final security release assessment belongs to security-release-gate.
This workflow does not exploit targets, modify application code, deploy changes, or claim that a modeled threat is a confirmed vulnerability.
Why this beats prompting it yourself
Generic threat modeling prompts often invent architecture to make the model look complete. That creates convincing but unreliable threats against systems that may not exist. This workflow treats architecture evidence as part of the security problem itself, preserving uncertainty while still producing actionable threat analysis.
Validation
Threat Modeling passed dedicated evidence hardening validation.
The validation verified the System Model Evidence Gate, the five tier evidence taxonomy, explicit evidence gap handling, prohibition of silent assumptions about multi tenancy, OAuth, gateways, and storage, and explicit treatment of unknown owners and controls.
At pack level, JetFerrari Security Engineering passed 58/58 routing integrity checks with zero unresolved cases.
Use cases
- Threat model a new application or service before implementation.
- Review security implications of a proposed architecture.
- Identify trust boundaries and high value assets.
- Generate STRIDE based abuse cases.
- Establish the security risks that should become formal requirements.
- Identify missing architecture evidence before security testing begins.
Known limitations
The quality of the threat model depends on the architecture and project evidence available to the agent. Missing context can leave components, controls, ownership, and priorities uncertain. The workflow distinguishes those gaps rather than silently filling them. Threats identified by the model are risk hypotheses, not confirmed vulnerabilities, and require downstream review or testing where verification is necessary.
Part of JetFerrari Security Engineering.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Security.
The quality of the threat model depends on the architecture, data flows, assets, trust boundaries, and project context available to the agent. Missing information remains UNKNOWN, INFERRED, or otherwise qualified rather than being silently promoted to fact. Threats are risk hypotheses and do not prove that a vulnerability exists. Owners, existing controls, and production architecture cannot be established without supporting evidence.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free to use
No account required to browse
Trust & safety
Security scan
Security scan passed 23 days ago
- Free with an Agensi account
Permissions
Access is limited to architecture documents, source code, configuration, and project context relevant to building the system model. The skill may use local read only inspection commands. It does not modify project files, access production systems, retrieve credentials, or perform active security testing.