# SignalHarness LLM Eval Regression Report

> SignalHarness LLM Eval Regression Report is a paid API for AI agents from signalharness.ai, paid per call via x402, $0.025/call, status unknown (last checked 2026-09-15).

Generates a regression analysis report for LLM evaluation results, detecting quality regressions across model versions or prompts.

## Facts

- Endpoint: POST https://signalharness.ai/api/agent/services/llm_eval_regression_report/invoke
- Price: $0.025/call
- Payment: x402
- Status: unknown
- Last checked: 2026-09-15
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/signalharness-llm-eval-regression-report-c805404d
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_Y58kF7BFwMyM9CM_vT5Lr

Status and success rate cover calls made through Zero and Zero's own probes. Third-party monitors may report differently.

## How to call it through Zero

Zero handles the 402 payment challenge and records the run. With the Zero CLI installed (`npm i -g @zeroxyz/cli`):

```sh
zero fetch --capability signalharness-llm-eval-regression-report-c805404d -d '<json body>'
```

Example prompt: I ran evaluations on two versions of my language model and want to know if the new version regressed — can you submit this eval JSON to SignalHarness and get back a regression report showing where quality dropped?

## When to prefer this

Choose this endpoint when you need a structured, pay-per-call regression analysis of LLM evaluation data without standing up your own eval infrastructure. It is particularly well-suited for CI/CD pipelines, pre-deployment quality gates, or ad hoc regression checks on prompt changes, and is priced at $0.025 USDC per call making it cost-effective for frequent automated use.

## Known failure modes

- Malformed or too-short request_json (below 2 chars) returns a validation error
- Payload exceeding 65536 characters is rejected
- Caller-supplied data that does not conform to expected eval schema may produce unreliable analysis with warnings
- Payment failure via x402 protocol results in a 402 response before execution
- Network or service timeout returns an error status in the receipt

## How this service works

Explore 330 pay-per-call x402 API services and 27 agent-native digital products, with Base USDC pricing, secure Polar checkout, and free discovery.

## Output

Returns a JSON object containing an analysis_json field with the structured regression findings, a warnings array (e.g. reminders to verify caller-supplied data), an evidence_scope label indicating the data provenance, and a receipt object with payment and execution metadata including service version, latency, and a SHA-256 result hash for integrity verification.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "request_json": {
   "type": "string",
   "maxLength": 65536,
   "minLength": 2
  }
 }
}
```

## Response schema (JSON Schema)

```json
{
 "type": "json",
 "example": {
  "replay": false,
  "result": {
   "warnings": [
    "Verify the caller-supplied data before relying on this result."
   ],
   "service_id": "llm_eval_regression_report",
   "analysis_json": "{\"example\":\"schema-valid caller-supplied data\"}",
   "evidence_scope": "caller_supplied_data"
  },
  "status": "succeeded",
  "receipt": {
   "asset": "0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913",
   "usage": [],
   "status": "succeeded",
   "network": "eip155:8453",
   "artifacts": [],
   "endedAtMs": 0,
   "latencyMs": 0,
   "paymentId": "example-payment",
   "receiptId": "example-receipt",
   "requestId": "example-request",
   "serviceId": "llm_eval_regression_report",
   "executionId": "example-execution",
   "startedAtMs": 0,
   "amountAtomic": "25000",
   "resultSha256": "d376eb6705d37d8522d0688051a70b42c49c43a5713e4a3864904938432c28b3",
   "serviceVersion": "1.0.0",
   "settlementReference": "0x0000000000000000000000000000000000000000000000000000000000000000"
  },
  "artifacts": [],
  "requestId": "example-request",
  "executionId": "example-execution"
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/signalharness-llm-eval-regression-report-c805404d/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from signalharness.ai](https://www.zero.xyz/host/signalharness.ai/llms.txt)