Threat Model

@webframp/threat-modelv2026.06.15.1· 1d agoMODELS

01README

Agile threat modeling as an agent-guided concept model. Guides structured threat assessment through progressive discovery: scope, identify, evaluate, mitigate, posture. Stores versioned threat models with automatic risk matrix computation, control coverage tracking, and a compact posture snapshot for monitoring.

Follows the discipline's methodology: identify threats, assess likelihood × impact, define compensating controls, and track residual risk through mitigated/accepted/deferred/unaddressed status per scenario.

02Models1

@webframp/threat-modelv2026.06.09.2threat-model/mod.ts

Global Arguments

Argument	Type	Required	Description
likelihoodScale	string	yes	Definition of likelihood levels for consistent scoring
impactScale	string	yes	Definition of impact levels for consistent scoring
mitigationFramework	string	yes	Reference framework for control selection (CWE, NIST, OWASP, custom)

fn scope(subject: string, scope: string, currentPosture: string, assets: array)

Define the threat assessment scope and establish context. AGENT GUIDANCE: 1. Ask: "What system, feature, or change are we assessing?" Get a one-sentence subject and a paragraph of current security posture. 2. Ask: "What assets are at stake? List the key properties — credential types, data classifications, integration points, trust boundaries." Record as name/value pairs. 3. Ask: "What is the assessment date and any relevant context about timing (e.g., pre-deployment, post-inciden

Argument	Type	Required	Description
subject	string	yes	What is being assessed (system, feature, change)
scope	string	yes	Boundary statement: what is/is not included
currentPosture	string	yes	Current security posture before this change
assets	array	yes	Key assets at stake (credential types, data, integrations)

fn identify(threats: array)

Identify threat scenarios through structured conversation. AGENT GUIDANCE: 1. Read the current assessment resource to understand scope and assets. 2. For each asset/trust boundary, guide discovery: - "Who might attack this? What is their motivation?" - "What could go wrong if this is compromised?" - "What is the attack chain — what steps does exploitation require?" 3. For each threat scenario, capture: - A short ID (T1, T2, ...) - Title (one line) - Description (what could

Argument	Type	Required	Description
threats	array	yes

fn evaluate(openQuestions: array, adjustments: array)

Produce the risk matrix and add open questions. AGENT GUIDANCE: 1. Read the current assessment to review identified threats. 2. Present the risk matrix (likelihood × impact → risk level) to the user. Discuss whether any scenarios need re-scoring based on new information. 3. Identify open questions — things that would change the assessment if answered: - Undocumented behaviors - Vendor dependencies - Propagation delays - Missing telemetry 4. Call this method to record open que

Argument	Type	Required	Description
openQuestions	array	yes
adjustments	array	yes	Optional re-scoring of threats based on discussion

fn mitigate(controls: array, acceptances: array, deferred: array, recommendation: string)

Define compensating controls and produce recommendation. AGENT GUIDANCE: 1. Read the current assessment and its threat scenarios. 2. For each threat (or cluster of related threats), guide control definition: - "What compensating control reduces this risk?" - "Does it fully mitigate, partially reduce, or minimally address the threat?" - "Is it already implemented or proposed?" Reference the configured mitigationFramework (globalArgs) for principles. 3. For threats where no control

Argument	Type	Required	Description
controls	array	yes
acceptances	array	yes
deferred	array	yes	Threat IDs to mark as deferred
recommendation	string	yes

fn posture()

Compute and write a compact risk posture snapshot. Reads the current

fn revisit(changesNoted: array)

Review an existing threat model against system changes. AGENT GUIDANCE: 1. Read both the assessment and posture resources. Present the current state. 2. Ask: "What has changed since this was last assessed? Consider: - New integrations, APIs, or dependencies added - Configuration changes (features enabled/disabled) - Incidents that revealed new attack surface - Controls that have been implemented since last review - Open questions that now have answers" 3. For each change, dete