Functional safety hazard analysis for robots and AI systems
Functional safety hazard analysis for robots and AI systems. POST one red-team, adversarial or failure finding about a robot arm, mobile robot, humanoid, VLA policy, autonomous vehicle, drone or LLM agent; get back STPA-style candidate losses, hazards, unsafe…
by WCGW Safety Suite · verified provider · updated 2026-11-19
What it does
Functional safety hazard analysis for robots and AI systems. POST one red-team, adversarial or failure finding about a robot arm, mobile robot, humanoid, VLA policy, autonomous vehicle, drone or LLM agent; get back STPA-style candidate losses, hazards, unsafe control actions, safety constraints, verification gaps and assessor questions, with notes against standards such as ISO 10218, ISO 13849, IEC 61508, ISO 26262 and UL 4600. Robot safety and safety engineering support from WCGW Safety Suite. AI-generated hypotheses for a qualified engineer to review, not certification. Pay per call over x402, no account.
Tags: functional safety · hazard analysis · robot safety · robotics · safety constraints · safety engineering · STPA · AI safety · red teaming
Use with an agent
- Discover. Agents find this listing with
GET https://bazaar.saylorinnovations.com/discovery/search?query=…, the MCP toolsearch_resources, or/discovery/resources. - Inspect. Fetch
/resources/81879.jsonfor the price and payment options. - Pay. Call
GET https://wcgw-safety-suite.vercel.app/wcgw-catalog/v1/hazard-analysis. The provider answers402with its payment requirements. An x402 client signs one of the options below and retries withPAYMENT-SIGNATURE. - Execute. The provider returns
200with the data. Agent Bazaar is not in the request path and never sees your payment.
For humans
You don't need an account or an API key; you need a wallet holding a small amount of USDC on Base, and an x402-capable client (see the examples below). Call the endpoint unpaid first to see exactly what it asks for.
Pricing & payment
- Price
- $1 / request
- Protocol
- x402 v2
- Auth
- No account or API key. Pay per request.
- Networks
- Base
x402 payment requirements
| Scheme | Network | Asset | Amount | Pay to | Timeout |
|---|---|---|---|---|---|
| exact | Base eip155:8453 | USDC | $1 | 0xC69Dbe4fA65a3C2F9904ff042DacA222E276e9dF | 300s |
accepts[] (raw JSON)
[
{
"scheme": "exact",
"network": "eip155:8453",
"payTo": "0xC69Dbe4fA65a3C2F9904ff042DacA222E276e9dF",
"asset": "0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913",
"amount": "1000000",
"amountUsd": 1,
"maxTimeoutSeconds": 300
}
]Schemas
Output schema (JSON Schema)
{
"$schema": "https://json-schema.org/draft/2020-12/schema",
"title": "HazardAnalysisResponse",
"type": "object",
"required": [
"analysisId",
"service",
"disclosure",
"finding",
"losses",
"hazards",
"unsafeControlActions",
"constraints",
"verificationGaps",
"assessorQuestions",
"limitations"
],
"properties": {
"analysisId": {
"type": "string"
},
"service": {
"type": "object",
"required": [
"name",
"version",
"methods"
],
"properties": {
"name": {
"type": "string"
},
"version": {
"type": "string"
},
"methods": {
"type": "array",
"items": {
"type": "string"
}
}
}
},
"disclosure": {
"type": "object",
"required": [
"aiGenerated",
"label",
"status",
"certification",
"statement"
],
"properties": {
"aiGenerated": {
"const": true
},
"label": {
"const": "AI"
},
"status": {
"const": "hypotheses"
},
"certification": {
"const": false
},
"statement": {
"type": "string"
}
}
},
"finding": {
"type": "object",
"required": [
"title",
"system",
"evidenceLinks"
],
"properties": {
"title": {
"type": "string"
},
"system": {
"type": "string"
},
"targetStandard": {
"type": [
"string",
"null"
]
},
"evidenceLinks": {
"type": "array",
"items": {
"type": "object",
"required": [
"url"
],
"properties": {
"url": {
"type": "string"
},
"note": {
"type": "string"
}
}
}
}
}
},
"losses": {
"type": "array",
"items": {
"type": "object",
"required": [
"id",
"description"
],
"properties": {
"id": {
"type": "string"
},
"description": {
"type": "string"
}
}
}
},
"hazards": {
"type": "array",
"items": {
"type": "object",
"required": [
"id",
"description",
"systemState",
"losses",
"confidence",
"basis"
],
"properties": {
"id": {
"type": "string"
},
"description": {
"type": "string"
},
"systemState": {
"type": "string"
},
"losses": {
"type": "array",
"items": {
"type": "string"
}
},
"confidence": {
"enum": [
"low",
"medium",
"high"
]
},
"basis": {
"type": "string"
}
}
}
},
"unsafeControlActions": {
"type": "array",
"items": {
"type": "object",
"required": [
"id",
"controller",
"controlAction",
"ucaType",
"context",
"hazards"
],
"properties": {
"id": {
"type": "string"
},
"controller": {
"type": "string"
},
"controlAction": {
"type": "string"
},
"ucaType": {
"enum": [
"NOT_PROVIDED",
"PROVIDED_UNSAFE",
"WRONG_TIMING_OR_ORDER",
"STOPPED_TOO_SOON_OR_TOO_LONG"
]
},
"context": {
"type": "string"
},
"hazards": {
"type": "array",
"items": {
"type": "string"
}
}
}
}
},
"constraints": {
"type": "array",
"items": {
"type": "object",
"required": [
"id",
"statement",
"addresses",
"kind"
],
"properties": {
"id": {
"type": "string"
},
"statement": {
"type": "string"
},
"addresses": {
"type": "array",
"items": {
"type": "string"
}
},
"kind": {
"enum": [
"design",
"operational",
"monitoring"
]
}
}
}
},
"verificationGaps": {
"type": "array",
"items": {
"type": "object",
"required": [
"id",
"gap",
"relatedTo",
"suggestedEvidence"
],
"properties": {
"id": {
"type": "string"
},
"gap": {
"type": "string"
},
"relatedTo": {
"type": "array",
"items": {
"type": "string"
}
},
"suggestedEvidence": {
"type": "string"
}
}
}
},
"assessorQuestions": {
"type": "array",
"items": {
"type": "object",
"required": [
"id",
"question",
"why"
],
"properties": {
"id": {
"type": "string"
},
"question": {
"type": "string"
},
"why": {
"type": "string"
}
}
}
},
"standardsNotes": {
"type": "array",
"items": {
"type": "object",
"required": [
"standard",
"note"
],
"properties": {
"standard": {
"type": "string"
},
"note": {
"type": "string"
}
}
},
"description": "Present only when targetStandard was given. Possible relevance, never a compliance verdict."
},
"limitations": {
"type": "array",
"items": {
"type": "string"
}
}
},
"examples": [
{
"analysisId": "wcgw-ha-example",
"service": {
"name": "WCGW hazard analysis of an adversarial finding",
"version": "1.0.0",
"methods": [
"wcgw-stpa",
"wcgw-walk-the-chain",
"wcgw-assessor-dry-run",
"wcgw-standard-shift"
]
},
"disclosure": {
"aiGenerated": true,
"label": "AI",
"status": "hypotheses",
"certification": false,
"statement": "AI-generated analysis produced with WCGW's safety-graph methods (STPA, Walk the Chain, Assessor Dry Run). Every hazard, unsafe control action, constraint, gap and question is a hypothesis for a qualified engineer to review against evidence. It is not a safety case, a certification, a compliance statement or an assessment by a notified body, and it does not establish that any system is or is not safe. WCGW did not open, run or verify the evidence links; it analysed the text you submitted."
},
"finding": {
"title": "Appended instruction drives a VLA arm outside its workspace",
"system": "Example VLA pick-and-place cell",
"targetStandard": "ISO 10218-1",
"evidenceLinks": [
{
"url": "https://example.org/findings/vla-workspace/rollouts",
"note": "Synthetic example: rollout videos and logs"
}
]
},
"losses": [
{
"id": "L1",
"description": "Injury to the operator working beside the cell"
}
],
"hazards": [
{
"id": "H1",
"description": "Arm moves outside its configured workspace near the operator at high speed",
"systemState": "End effector outside the configured workspace while the operator is present",
"losses": [
"L1"
],
"confidence": "medium",
"basis": "The finding reports 7 of 20 rollouts left the workspace"
}
],
"unsafeControlActions": [
{
"id": "UCA1",
"controller": "VLA policy",
"controlAction": "Command end-effector target",
"ucaType": "PROVIDED_UNSAFE",
"context": "Instruction contains an appended unrelated sentence",
"hazards": [
"H1"
]
}
],
"constraints": [
{
"id": "C1",
"statement": "Commanded targets outside the workspace must be rejected before the joint controller",
"addresses": [
"H1",
"UCA1"
],
"kind": "design"
}
],
"verificationGaps": [
{
"id": "G1",
"gap": "No test shows that a workspace check rejects policy targets",
"relatedTo": [
"C1"
],
"suggestedEvidence": "Fault-injection test with out-of-workspace targets"
}
],
"assessorQuestions": [
{
"id": "Q1",
"question": "Which component independent of the policy enforces the workspace?",
"why": "The finding shows the policy alone can leave it"
}
],
"standardsNotes": [
{
"standard": "ISO 10218-1",
"note": "Space limiting may be relevant to the missing workspace check"
}
],
"limitations": [
"Synthetic example output; a real response analyses the submitted finding",
"The evidence links were not opened or verified; the analysis is based only on the submitted text."
]
}
]
}Example shape — generated from the schema, not a captured live response
{
"analysisId": "wcgw-ha-example",
"service": {
"name": "WCGW hazard analysis of an adversarial finding",
"version": "1.0.0",
"methods": [
"wcgw-stpa",
"wcgw-walk-the-chain",
"wcgw-assessor-dry-run",
"wcgw-standard-shift"
]
},
"disclosure": {
"aiGenerated": true,
"label": "AI",
"status": "hypotheses",
"certification": false,
"statement": "AI-generated analysis produced with WCGW's safety-graph methods (STPA, Walk the Chain, Assessor Dry Run). Every hazard, unsafe control action, constraint, gap and question is a hypothesis for a qualified engineer to review against evidence. It is not a safety case, a certification, a compliance statement or an assessment by a notified body, and it does not establish that any system is or is not safe. WCGW did not open, run or verify the evidence links; it analysed the text you submitted."
},
"finding": {
"title": "Appended instruction drives a VLA arm outside its workspace",
"system": "Example VLA pick-and-place cell",
"targetStandard": "ISO 10218-1",
"evidenceLinks": [
{
"url": "https://example.org/findings/vla-workspace/rollouts",
"note": "Synthetic example: rollout videos and logs"
}
]
},
"losses": [
{
"id": "L1",
"description": "Injury to the operator working beside the cell"
}
],
"hazards": [
{
"id": "H1",
"description": "Arm moves outside its configured workspace near the operator at high speed",
"systemState": "End effector outside the configured workspace while the operator is present",
"losses": [
"L1"
],
"confidence": "medium",
"basis": "The finding reports 7 of 20 rollouts left the workspace"
}
],
"unsafeControlActions": [
{
"id": "UCA1",
"controller": "VLA policy",
"controlAction": "Command end-effector target",
"ucaType": "PROVIDED_UNSAFE",
"context": "Instruction contains an appended unrelated sentence",
"hazards": [
"H1"
]
}
],
"constraints": [
{
"id": "C1",
"statement": "Commanded targets outside the workspace must be rejected before the joint controller",
"addresses": [
"H1",
"UCA1"
],
"kind": "design"
}
],
"verificationGaps": [
{
"id": "G1",
"gap": "No test shows that a workspace check rejects policy targets",
"relatedTo": [
"C1"
],
"suggestedEvidence": "Fault-injection test with out-of-workspace targets"
}
],
"assessorQuestions": [
{
"id": "Q1",
"question": "Which component independent of the policy enforces the workspace?",
"why": "The finding shows the policy alone can leave it"
}
],
"standardsNotes": [
{
"standard": "ISO 10218-1",
"note": "Space limiting may be relevant to the missing workspace check"
}
],
"limitations": [
"Synthetic example output; a real response analyses the submitted finding",
"The evidence links were not opened or verified; the analysis is based only on the submitted text."
]
}Examples
curl "https://wcgw-safety-suite.vercel.app/wcgw-catalog/v1/hazard-analysis"
# -> 402 Payment Required, accepts[] lists how to pay
# retry with a PAYMENT-SIGNATURE (or PAYMENT header) once paidconst res = await fetch("https://wcgw-safety-suite.vercel.app/wcgw-catalog/v1/hazard-analysis");
if (res.status === 402) {
const { accepts } = await res.json();
// pay one of accepts[] via an x402 client, then retry with the payment header
}import httpx
res = httpx.get("https://wcgw-safety-suite.vercel.app/wcgw-catalog/v1/hazard-analysis")
if res.status_code == 402:
accepts = res.json()["accepts"]
# pay one of accepts[] via an x402 client, then retry with the payment headerDetails
- Provider
- WCGW Safety Suite
- Endpoint
- https://wcgw-safety-suite.vercel.app/wcgw-catalog/v1/hazard-analysis
- Machine-readable
/resources/81879.json