Skip to content

Latest commit

 

History

History
74 lines (60 loc) · 5.62 KB

File metadata and controls

74 lines (60 loc) · 5.62 KB

Competitive Analysis — API Verity Lab

Generated: 2026-08-23 · Evidence: live GitHub API metadata (see data/competitor-meta.json) + documented product models. Full machine-readable matrix: data/competitive-capabilities.json.

Method

  • Repo license, stars, last push, archived status and latest release fetched live on 2026-08-23 via the authenticated GitHub API for every competitor.
  • Qualitative capability claims are labeled VERIFIED (fetched metadata / official docs quotes) or KNOWLEDGE-BASED (documented product understanding).
  • No competitor is claimed to lack a feature without verification; unverified areas are unknown.

Landscape snapshot (verified 2026-09-09)

Tool License Stars Last push Latest release Status
oasdiff Apache-2.0 1,356 2026-09-08 v1.31.0 (2026-09-05) active
Schemathesis MIT 3,590 2026-09-09 v4.26.1 (2026-09-08) active
Spectral Apache-2.0 3,205 2026-09-06 v6.16.3 (2026-08-03) active
Pact (pact-js) NOASSERTION 1,813 2026-09-07 v17.1.4 (2026-09-07) active
Optic MIT 1,534 2026-01-08 v1.0.9 (2025-08-10) archived
Dredd MIT 4,222 2024-05-11 dredd@14.1.0 (2021-11-16) archived
Prism Apache-2.0 5,028 2026-09-03 v5.16.0 (2026-07-17) active
WireMock Apache-2.0 7,363 2026-09-08 3.13.2 (2025-11-14) active
Hoverfly Apache-2.0 2,512 2026-09-07 v1.12.13 (2026-08-28) active
Karate MIT 8,950 2026-09-05 v2.1.2 (2026-08-14) active
k6 AGPL-3.0 31,424 2026-09-09 v2.2.0 (2026-08-10) active
Newman Apache-2.0 7,251 2026-08-05 active
GraphQL Inspector MIT 1,767 2026-08-20 release-1787234790800 (2026-08-20) active
Buf Apache-2.0 11,424 2026-09-09 v1.72.0 (2026-07-17) active

What each one owns

  • oasdiff — OpenAPI diff + breaking changes. Focused, fast, healthy. Nothing else.
  • Schemathesis — schema-driven fuzzing, response validation, stateful testing via links, failure minimization. The fuzzing bar.
  • Spectral — the linting/ruleset standard (OpenAPI/Arazzo/AsyncAPI v2 rulesets).
  • Pact (+PactFlow) — consumer-driven contracts, provider verification, broker publication, can-i-deploy. The deploy-decision bar. Collaboration = self-hostable OSS Broker; RBAC/SSO in commercial PactFlow.
  • Prism — spec-accurate mock server with request validation.
  • WireMock / Hoverfly — service virtualization: stubbing, scenarios/state, record-replay, fault injection. JVM-centric (WireMock) / proxy-centric (Hoverfly). Cloud/Team layers are hosted-commercial.
  • Karate — broad test automation DSL incl. mocks and Gatling-based perf.
  • k6 — load engine + thresholds (= budgets). AGPL core; Grafana Cloud commercial.
  • Newman/Postman — collection runner; collaboration/mocks/monitors locked to hosted platform.
  • GraphQL Inspector — reference GraphQL breaking/dangerous taxonomy, schema coverage. Hive console is the hosted layer.
  • Buf — gold-standard protobuf lint/breaking (wire semantics); BSR registry is hosted-first.
  • Dredd — archived (last push 2024-05-11); instructive history only.
  • Optic — archived 2026-01 at 1,534 stars, in precisely this domain. An earlier revision of this file recorded it as "repo gone (404)". It is not gone: opticdev/optic is archived and public. The 404 came from this repository's own fetcher asking for useoptic/optic, which does not exist, and the null being written into the table as a finding. Fixed at the source -- the fetcher now raises on a failed lookup instead of recording it as data.

Capability matrix

See data/competitive-capabilities.json for the full 25-dimension × 14-tool matrix with per-cell evidence notes. Highlights:

  • Drift detection: not a documented focus area for any listed competitor → open niche.
  • Safety-gated replay w/ redaction: WireMock/Hoverfly do record-replay but without contract-aware safety gating or redaction DSL → differentiable.
  • can-i-deploy over plain specs: only Pact offers it, requiring its broker worldview → offering it over OpenAPI/GraphQL/gRPC contracts lowers adoption cost.
  • Unified frontend: most OSS cores are CLI-only; rich UIs sit behind hosted products.

Strategic conclusions for API Verity Lab

  1. The shared model is the moat. One normalized contract graph feeding diff → breaking rules → generators → workflows → runtime → drift/perf → results → deploy decisions is something no single-lane tool provides.
  2. Match specialist depth where cheap, integrate always. e.g., our fuzzing should be deterministic and CI-friendly (a Schemathesis complement, not clone); protobuf rules should adopt Buf's wire-compat categories conceptually.
  3. Own the uncontested niches: drift baselines/trends, redaction-proven traffic corpora, replay dry-runs and destructive-method gates, contract-aware performance budgets.
  4. Licensing: stay Apache-2.0. Community keeps full local capability; team/enterprise value comes from the self-hosted server (collaboration, policy-as-code, audit, workers) — the same layer competitors gate behind hosted products. No non-commercial restrictions.
  5. Do not copy: hosted-only lock-in, AGPL constraints, JVM-centrism, or features of dead tools.

Deliberately not copied

  • Hosted-only collaboration (Postman/Hive/BSR model) — we ship a self-hosted server instead.
  • k6's scripting language/AGPL — we use declarative profiles + budgets.
  • WireMock's hand-authored stub DSL — our virtualization derives from contracts.
  • Browser recorder ecosystems — out of scope by design.