Keploy API Testing & Mock Platform
Auto-generate API tests and mocks from real user traffic with zero code changes — faster than unit tests.
- Source repo
- keploy/keploy
- Stars
- ★ 19k
- Last updated
- today
- License
- Apache-2.0
- Primary language
- Go
- FA score
- 49/100 · Major gaps
At a glance
- How it runs
- Works with
- Universal · cross-platform
- Cost
- Free tier plus a paid hosted plan
- Setup effort
- Low · running in minutes
- You'll need
- Typical use
- A developer uses
keploy recordduring development to automatically capture real API traffic and turn it into regression tests without writing test code. - Not a fit if
- Teams on Windows/macOS or without Linux kernel eBPF support
- CI setups that cannot grant root or privileged containers
- Teams needing dependencies whose protocols are not open-sourced
- Source review
- 49/100 · Major gaps 2 safety controls not found
What does this agent do, and when should you use it?
Keploy is an open-source API and integration testing platform that captures traffic at the network layer using eBPF, automatically turning API calls, database queries, and streaming events into test cases and mocks. It requires no SDKs or code modifications and works with any language or framework. Keploy virtualizes databases, message queues, and external APIs, enabling deterministic test replay. It provides unified coverage reports including statement, branch, and API schema coverage. AI-assisted test generation expands coverage by analyzing existing recordings and OpenAPI schemas. It integrates with CI/CD pipelines and offers a centralized mock registry.
Keploy runs the application under test with the keploy record command, intercepting network traffic via eBPF. It automatically records API requests/responses, database calls, and streaming events to generate test cases and mocks. The keploy test command replays these recordings offline, with optional time freezing for determinism. It calculates statement, branch, and API schema coverage. AI analyzes OpenAPI schemas and existing recordings to generate additional test cases covering boundary values, missing fields, and error scenarios. Generated mocks can also be used as server tests. Unified reports and a mock registry support CI/CD integration.
- A developer uses
keploy recordduring development to automatically capture real API traffic and turn it into regression tests without writing test code. - A QA engineer relies on Keploy's API schema coverage reports to verify business use-case coverage, replacing subjective assessments.
- A DevOps team integrates Keploy into CI pipelines (GitHub Actions, Jenkins) to run tests and generate coverage reports without external dependencies.
- A backend team tests complex distributed flows involving Kafka, RabbitMQ, or Postgres by virtualizing these dependencies and replaying deterministically.
- An organization improves test depth by having AI generate edge cases (boundary values, wrong types) beyond existing recordings and schemas.
How do you install or deploy this agent?
Install the Keploy Agent on a Linux environment with curl and bash:
curl --silent -O -L https://keploy.io/install.sh && source install.shHow do you use this agent?
- Start your application under Keploy:
keploy record -c "CMD_TO_RUN_APP"(e.g.,keploy record -c "python main.py"). 2. Exercise the application to generate traffic. 3. Run tests offline:keploy test -c "CMD_TO_RUN_APP" --delay 10. Refer to the official documentation for advanced CLI options and CI/CD integration.
What are this agent's strengths and limitations?
- Zero code changes: captures traffic via eBPF at the network layer, supporting any language without SDKs.
- Full infrastructure virtualization: mocks databases, message queues, and external APIs beyond HTTP, enabling deterministic tests.
- Comprehensive coverage: measures statement, branch, and API schema coverage, catering to both developers and QA.
- AI-driven test expansion: automatically generates additional test cases for boundary values and missing fields from OpenAPI schemas.
- eBPF requires Linux kernel support; Windows/macOS may need VMs or containers.
- Non-open-source protocols for certain dependencies are not supported (likely enterprise-only).
- Installation depends on curl and bash, which may fail in restricted environments.
- Running eBPF may require elevated privileges (e.g., root), complicating some CI setups.
How does this agent compare with similar options?
Key facts side by side with the most closely related agents.
| Agent | Source review | Form / cost | Stars | Updated | Language | Full support on |
|---|---|---|---|---|---|---|
| Keploy API Testing & Mock Platform This agent | 49 · Major gaps | CLIFreemium | ★ 19k | today | Go | — |
| Kane CLI | 71 · Some gaps | CLIFreemium | ★ 251 | 3d ago | JavaScript | Codex · Claude Code |
| Supercov | 75 · Some gaps | CLIFree + model costs | ★ 156 | today | Rust | Codex · Claude Code |
| Carbon Code | 64 · Some gaps | CLIFree + model costs | ★ 198 | 2mo ago | TypeScript | OpenAI API |
How does FollowAgents rate this agent?
Why each dimension lost points
Evidence shows: README describes using eBPF to capture network traffic, requiring root privileges, but no least-privilege configuration or privilege reduction is provided; installation script is piped via curl without user confirmation; data flow transparency is limited, only mentioning recording API calls and database queries, not storage or transmission; sensitive data handling lacks encryption or redaction; many dependencies but no security audit or vulnerability scanning evidence; external effects include network capture and possible system modifications, but no rollback mechanism; source attribution has Apache-2.0 license and copyright, but publisher unverified. Deductions: lack of user confirmation, rollback, detailed data flow and sensitive data handling.
Evidence shows: README and test files are consistent, e2e test scripts are detailed with error handling, but many dependencies without availability guarantees; failure messages are clear in test scripts, but product's own error handling not detailed. Deductions: dependency availability unverified, product failure messages not fully shown.
Evidence shows: README clearly targets developers and QA, scenarios include API, integration, and E2E testing; capability boundaries via supported languages and protocols; trigger is CLI record and replay, precise; environment fit includes local, CI, and Kubernetes, but lacks detailed configuration. Deductions: environment fit details insufficient.
Evidence shows: README structure clear, includes installation, quick start, docs links; install notes provide curl command; naming stable, but version info not explicit; examples and FAQ links exist; known limitations mention some non-open-source dependencies; license Apache-2.0; versioning and changelog not in files; maintenance responsibility via contribution guide and community links. Deductions: versioning and changelog missing, known limitations incomplete.
Evidence shows: output is test cases and mocks, directly usable; marginal value high, auto-generating tests saves time; cost-benefit not quantified, but open-source free. Deductions: cost-benefit lacks specific data.
Evidence shows: README claims have doc links, but no independent verification; test files exist, but no test results; facts and inferences not clearly separated. Deductions: claim traceability insufficient, lack of cross-validation.
- Not found in source: confirmation before actingTurn on (or add) a confirmation step before it acts, and try it in a sandbox or test environment before real data.
- Not found in source: rollback or recovery pathBack up first, or work on a git branch or snapshot, so its changes can be undone.
- Installation script is piped via curl, posing supply chain risk; review script content.
- eBPF network capture requires high privileges, may impact system security; evaluate carefully.
- Many dependencies without security audit; recommend vulnerability scanning.
Recent releases
Latest entries from the project's GitHub Releases. Notes are quoted from upstream as written, not rewritten or tested by FollowAgents.
-
Release notes excerpt
What's Changed * fix(replay): a 2xx answer to CONNECT ends at its headers, and what follows is the tunnel by @slayerjain in https://github.com/keploy/keploy/pull/4844 * Batch K-B22: zero-copy end stamp, and a parser says which mock variables it follows by @slayerjain in https://github.com/keploy/keploy/pull/4856 Full Changelog: https://github.com/keploy/keploy/compare/v3.6.143...v3.6.144View on GitHub ↗ -
Release notes excerpt
What's Changed * fix(ci): Windows lanes take keploy.exe from the machine that built it, not a download by @slayerjain in https://github.com/keploy/keploy/pull/4841 * Batch K-B20: PlaceOf, S1 settled reads, boot judge on runnable time, S12 worker scope by @slayerjain in https://github.com/keploy/keploy/pull/4851 Full Changelog: https://github.com/keploy/keploy/compare/v3.6.142...v3.6.143View on GitHub ↗ -
Release notes excerpt
What's Changed * Batch K-B19: docker agent args quoted, unchecked-hook reasons, in-flight gap answers (D2) by @slayerjain in https://github.com/keploy/keploy/pull/4846 Full Changelog: https://github.com/keploy/keploy/compare/v3.6.141...v3.6.142View on GitHub ↗