omh-application-threat-model
[omh] Application threat model workflow: turn a system's components and data flows into assets, trust boundaries, attack scenarios, controls, and the security test that proves each control holds. Use when the user says: application-threat-model, application threat model, threat model, threat modeling, threat modelling, threat modeling session, threat modeling workshop, security threat model.
rlaope · AI review on file
Installation
Read the source and dependencies before running a command. Copying does not install or execute anything.
npx skills add rlaope/oh-my-hermes --skill omh-application-threat-modelAbout this skill
name: "omh-application-threat-model" description: "[omh] Application threat model workflow: turn a system's components and data flows into assets, trust boundaries, attack scenarios, controls, and the security test that proves each control holds. Use when the user says: application-threat-model, application threat model, threat model, threat modeling, threat modelling, threat modeling session, threat modeling workshop, security threat model." metadata: hermes: tags: [workflow, oh-my-hermes, review] category: review phase: application-threat-model role: reviewer quality_tier: security-safety-gated
Application Threat Model
This is an OMH application-threat-model workflow skill, projected for Agent Skills hosts (Claude Code, Codex, Cursor, opencode, OpenClaw, pi).
Why This Exists
application-threat-model exists because the nearest neighbour does not merely miss this request. security-safety-review maps the agent's own prompt, tool, credential, and dependency surface, so an application threat-model request came back as an agent tool inventory under a near-identical name — a confident wrong artifact rather than a miss, in the one domain where that costs most.
Do Not Use When
- The subject is the agent's own prompts, tools, files, credentials, dependencies, or destructive actions; use
security-safety-review, which maps that runtime surface. - The user wants defects found in a diff or a file; use
code-review. - The user asks whether a release is ready across rollout, rollback, and observability; use
production-audit. - The user asks which commands prove a merge is safe; use
verification-gate. - The user asks for a contractual or regulatory obligation rather than an attacker; use
legal-compliance-review.
Examples
Good example:
- Prompt: build a threat model for our payment service architecture
- Expected behavior: Prepare application_threat_model/v1: ask for the component map and data flows, register card data and settlement records as assets, mark the merchant API edge and the PSP callback as trust boundaries, derive scenarios per boundary, decide a control for each, and name the test that fails when the control is removed.
- Why: The subject is an application the user operates, and the goal needs assets, boundaries, scenarios, controls, and tests.
Bad example:
- Prompt: application-threat-model check whether this agent can be prompt-injected through its file tool
- Expected behavior: Route to
security-safety-review: prompts, tools, and credentials are the agent's runtime surface, not an application this workflow models. - Why: The two surfaces share vocabulary and nothing else; modeling the agent's runtime here is how the artifacts get confused.
Completion Checklist
- Every asset carries a data class and one named loss; every boundary names what crosses it and what authenticates the crossing.
- Every scenario resolves to mitigate, transfer, accept, or eliminate, with an owner.
- Every mitigating control carries a security test and the observable that fails without it.
- Controls read deployed, planned, or
unverified; none is inferred from the architecture description. - Residual risk is listed, and the model is not offered as a scan, a penetration test, or an attestation.
Recovery Notes
- If the architecture is not described, ask for the component map and the data flows before modeling; never substitute a generic checklist for the real system.
- If a scenario has no boundary and no asset, drop it with the reason rather than carrying an unreachable threat.
- If the user asks for exploit code, give the precondition and the detection signal instead, then hand remediation to an executor.
- If the request turns out to be about the agent's own prompts, tools, or credentials, stop and hand it to
security-safety-review.
Use When
Use when Hermes must model the security of an application, service, or deployed system the user operates: which assets are worth taking, where trust changes hands, how an attacker reaches each asset, which control stops them, and which security test fails when that control is removed. The subject is the modeled system, never the agent's own runtime.
Strong routing signals: `application-threat-model`, `application threat model`, `threat model`, `threat modeling`, `threat modelling`, `threat modeling session`, `threat modeling workshop`, `security threat model`, `build a threat model`, `model the threats`, `threat scenarios`, `stride analysis`, `stride model`, `trust boundary`, `trust boundaries`, `attack scenario`, `attack scenarios`, `attack tree`, `attack trees`, `abuse case`, `abuse cases`, `security design review`, `security architecture review`, `architecture security review`, `how would an attacker`, `how could an attacker`, `what could an attacker do`, `attacker perspective`
Catalog Metadata
Category: review
Phase: application-threat-model
Quality tier: security-safety-gated
Reasoning demand: standard
Quality bar:
- Name every component, data store, and external dependency of the real system before naming one threat; a model of a system nobody described is a checklist.
- Give each asset a data class and exactly one loss: disclosure, corruption, unavailability, or fraud.
- For each trust boundary, state what crosses it, what authenticates the crossing, and what the receiver assumes without checking.
- Run all six STRIDE prompts from
omh-application-threat-model/references/threat-model-method.mdper boundary; drop an unreachable scenario with its reason instead of carrying it. - Resolve every scenario to one decision (mitigate, transfer, accept, eliminate) with an owner, and give every mitigating control a test whose observable fails when the control is removed.
Required inputs:
- the system under review: components, which component calls which, and where each is deployed
- data flows and data classes: what every store, queue, and message carries
- known trust