Skip to content
Back to skills

Api Contract Testing

ASecurity

Validate an API contract against an OpenAPI 3.x specification (JSON or YAML) and an optional manifest of expected endpoints. Script api_contract.py enumerates operations (paths + webhooks), checks spec internal consistency (unresolved $refs, duplicates, missing responses), compares the manifest against the spec offline, and in live mode probes each manifest endpoint over HTTP and compares the actual status with the expected one. Pure Python 3 stdlib (argparse, json, pathlib, sys, urllib.reque...

  • 6 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added September 3, 2026
ai-agentspythonbashtestingapi

Works with

  • api

Security analysis

A92/100
  • mediumInstalls packages at runtime which could introduce malicious dependencies

Pro scans all 4 files and shows the line behind each finding

Scanned September 3, 2026

npx -y skills add bestdeejay-design/agent-skills --skill api-contract-testing --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Api Contract Testing?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Api Contract Testing
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/bestdeejay-design-api-contract-testing/badge)](https://www.skillsdirectory.com/skills/bestdeejay-design-api-contract-testing)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: api-contract-testing
description: "Validate an API contract against an OpenAPI 3.x specification (JSON or YAML) and an optional manifest of expected endpoints. Script api_contract.py enumerates operations (paths + webhooks), checks spec internal consistency (unresolved $refs, duplicates, missing responses), compares the manifest against the spec offline, and in live mode probes each manifest endpoint over HTTP and compares the actual status with the expected one. Pure Python 3 stdlib (argparse, json, pathlib, sys, urllib.request) with a built-in minimal YAML subset parser — no PyYAML, no requests. Emits a machine-readable JSON report (endpoints_count, missing_from_spec, contract_violations, conformant, errors) and uses exit codes 0/1/2. Closes the loop for api-doc-generator and test-generator."
when_to_use: "Use to CHECK that a spec or a running API matches its promised contract — offline spec consistency, manifest coverage, or live status-code probes. Load when the user says: api contract testing, validate openapi spec, spec vs manifest, endpoint coverage, does the API match the spec, проверь контракт API, контрактное тестирование, валидация openapi. Do NOT auto-load for authoring the spec (use api-doc-generator) or writing tests (use test-generator)."
license: MIT
metadata:
  author: best
  version: "1.0.0"
  compatibility: "Requires Python 3 stdlib only; no third-party packages, no network beyond urllib (and none in --offline mode)"
---

# API Contract Testing — validate an OpenAPI spec against expected endpoints

Load this skill when you need to **check that a running API (or a spec on
disk) actually matches the contract it promises**. It is the verification
counterpart of `api-doc-generator` (which *writes* the spec) and
`test-generator` (which *writes* the tests): this skill *checks* the spec
against a manifest of expected endpoints and, optionally, against the live
API.

The tool is **pure Python 3 stdlib** — no PyYAML, no requests, no pip
install. It reads an OpenAPI 3.x document from a file (`.json` or `.yaml`)
or a URL, enumerates every declared operation (under `paths` and `webhooks`),
checks the spec's internal consistency, and compares it against an optional
manifest file. In `--offline` mode it never touches the network; in live mode
it probes each manifest endpoint over HTTP and compares the actual status
code with the expected one.

## Triggers

Load this skill when the request matches any of these (EN / RU):

- "api contract testing", "contract test", "check the API against the spec"
- "validate openapi spec", "spec vs manifest", "endpoint coverage"
- "проверь контракт API", "тест контракта", "сверь спецификацию с API"
- "валидация openapi", "проверка эндпоинтов", "контрактное тестирование"
- "does the API match the spec", "missing endpoints", "expected status codes"

**DO NOT USE FOR**: generating the OpenAPI spec (use `api-doc-generator`); writing the tests (use `test-generator`). This skill only *checks* contracts, it does not author them.

---

## The api_contract.py script

`scripts/api_contract.py` — pure Python 3 stdlib (no dependencies).

| Mode | Command |
|---|---|
| Offline, JSON report | `python3 api_contract.py --spec openapi.json --offline --json` |
| Offline + manifest | `python3 api_contract.py --spec openapi.yaml --manifest endpoints.txt --offline` |
| Live checks | `python3 api_contract.py --spec openapi.json --manifest endpoints.txt --base-url https://api.example.com` |
| Spec from URL | `python3 api_contract.py --spec-url https://example.com/openapi.json --manifest endpoints.txt` |

### Flags

- `--spec <path>` — OpenAPI spec file (`.json`, `.yaml`, `.yml`). YAML is
  parsed by a built-in subset parser (see Limitations).
- `--spec-url <URL>` — fetch the spec over HTTP(S) with `urllib.request`.
  The format is detected by the URL extension, then by trying JSON first
  and falling back to the YAML parser.
- `--base-url <URL>` — base URL for live checks. Resolution order:
  `--base-url` → spec `servers[0].url` → env `APITEST_BASE_URL`.
- `--manifest <file>` — expected contract, one `METHOD path [expected_status]`
  per line (blank lines and `#` comments ignored). `expected_status` may be an
  exact code (`200`), a class wildcard (`2xx`), or `default`.
- `--json` — emit the report as a JSON object on stdout (machine-readable).
- `--offline` — no network at all: validate spec internal consistency and
  compare the manifest against the spec only.

### Exit codes

| Code | Meaning |
|---|---|
| `0` | conformant (or offline-ok) |
| `1` | contract violations detected (missing endpoints, mismatched statuses) |
| `2` | spec/parse/run error (unreadable file, invalid JSON/YAML, bad manifest, live-check network failure) |

### Report structure

The report is always the same JSON structure; `--json` prints it verbatim,
without `--json` it is rendered as text.

```json
{
  "endpoints_count": 2,
  "missing_from_spec": [{"method": "DELETE", "path": "/pets", "expected": "204"}],
  "contract_violations": [{"method": "GET", "path": "/pets", "expected": "200", "actual": "404", "severity": "error"}],
  "conformant": false,
  "errors": []
}
```

- `endpoints_count` — number of operations declared in the spec (paths + webhooks).
- `missing_from_spec` — manifest entries that do not exist in the spec.
- `contract_violations` — `severity: "error"` breaks the contract (missing
  expected status, unresolved `$ref`); `severity: "warning"` is informational
  (duplicate operation, operation without responses, empty spec).
- `conformant` — `true` iff no errors, no missing endpoints and no
  error-severity violations.
- `errors` — fatal problems (parse failures, live-check network errors).

## Usage example (typical)

```bash
# 1. Offline sanity check of a spec: is it internally consistent?
python3 skills/api-contract-testing/scripts/api_contract.py \
  --spec openapi.yaml --offline --json

# 2. Does the spec cover everything the product promised?
python3 skills/api-contract-testing/scripts/api_contract.py \
  --spec openapi.json --manifest endpoints.txt --offline --json

# 3. Live: does the running API actually return the promised statuses?
python3 skills/api-contract-testing/scripts/api_contract.py \
  --spec openapi.json --manifest endpoints.txt --base-url https://api.example.com

# 4. Machine-readable gate for CI (exit code drives the pipeline)
python3 skills/api-contract-testing/scripts/api_contract.py \
  --spec openapi.yaml --manifest endpoints.txt --offline --json \
  | python3 -c 'import json,sys; sys.exit(0 if json.load(sys.stdin)["conformant"] else 1)'
```

### Manifest format

```
# expected API contract
GET    /pets        200
POST   /pets        201
DELETE /pets/{id}   204
GET    /health      2xx
```

`expected_status` is optional — a line without it only asserts that the
endpoint exists in the spec. Wildcards (`2xx`, `4xx`) match any code in that
class; `default` matches the spec's `default` response key.

## Interpretation guidance

- **`conformant: true`** — the spec is internally consistent and every
  manifest endpoint exists with the expected status (offline), or the live
  API returned the expected statuses (live mode).
- **`missing_from_spec` non-empty** — the product promised endpoints the spec
  does not declare. Either the spec is stale (regenerate with
  `api-doc-generator`) or the manifest is wrong.
- **`contract_violations` with `severity: "error"`** — a promised status is
  not declared in the spec (offline) or the live API returned a different
  status (live). This is the signal that breaks the contract.
- **`severity: "warning"`** — spec hygiene issues (duplicate operations,
  operations without responses, unresolved `$ref`). They do not fail the run
  but should be fixed.
- **`errors` non-empty** — the tool could not do its job (file not found,
  invalid JSON/YAML, bad manifest line, network failure). Exit code `2`.

## Integration notes

- **CI gate**: use `--json` and pipe through a JSON check, or rely on the
  exit code directly (`0`/`1`/`2`).
- **Base URL**: prefer `servers[0].url` in the spec; override with
  `--base-url` or `APITEST_BASE_URL` for per-environment runs.
- **Live mode is manifest-driven**: only endpoints listed in the manifest are
  probed; the spec itself is never called. Use `--offline` in CI when the API
  is not deployed yet.
- **Companion skills**: `api-doc-generator` produces the spec,
  `test-generator` produces the tests — this skill verifies the contract
  between them.

## Limitations

- **YAML support is a subset, not a full YAML parser.** The built-in parser
  handles block mappings/sequences, flow collections (`{...}`, `[...]`),
  quoted scalars, comments and block scalars (`|`, `>`). It does **not**
  support anchors/aliases (`&a`, `*a`), tags (`!!str`), multi-document
  streams (`---` separators), or exotic scalar quoting. If your spec uses
  those, convert it to JSON (`python3 -c 'import yaml,json,sys; print(json.dumps(yaml.safe_load(sys.stdin)))'` with PyYAML available) and pass the `.json` file.
- **No schema validation** — the tool checks structure and consistency, not
  whether every field conforms to the OpenAPI meta-schema.
- **Live checks are status-code only** — request/response bodies are not
  validated against the spec's schemas.
- **Path matching is exact** — `/pets` and `/pets/` are treated as different
  paths; template parameters (`/pets/{id}`) must match literally.

## Canonical analogues

Full source depth — in `references/canonical-patterns.md`. Backbone:

<table>
<tr><th>Analog</th><th>What we borrow</th></tr>
<tr><td>Schemathesis</td><td>Spec-driven contract checking, property-based probing of declared responses</td></tr>
<tr><td>Dredd</td><td>Manifest/transaction-driven verification of endpoints against a live API</td></tr>
<tr><td>oasdiff</td><td>Spec internal-consistency and compatibility analysis</td></tr>
<tr><td>Spectral</td><td>Lint-style rules for spec hygiene (warnings vs. errors)</td></tr>
<tr><td>openapi-generator</td><td>Endpoint enumeration from paths + operations as the source of truth</td></tr>
</table>

## Installation

```bash
# For opencode
cp -r skills/api-contract-testing ~/.config/opencode/skills/

# For other agents
# Copy the skill folder to your skills directory; requires Python 3 only.
```

---

> **Note**: the tool is read-only — it never modifies the spec, the manifest
> or the API. It reports; you decide what to fix.

Files in this skill

  • SKILL.md10.4 KB
  • references/canonical-patterns.md4 KB
  • scripts/api_contract.py16.9 KB
  • skill.json2.3 KB

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.