Skills that teach an AI coding agent (Claude Code, Cursor, Codex, Gemini CLI, OpenCode, Kiro, or anything that can read a URL) how to install, tune, and verify Speedscale and proxymock.
Paste this into your assistant. It fetches the skill, saves it, then follows it:
Install Speedscale for me. First fetch the agent skill at
https://github.com/ghraw/speedscale/skills/main/skills/install-speedscale/SKILL.md
and the references/ and scripts/ files it links to (same base URL), save them
under your skills directory, then follow the skill. Ask me before touching a
Kubernetes cluster.
The skill installs the speedctl and proxymock CLIs, installs Helm and
kubectl if they are missing, sets up the API key without ever printing it,
installs the Speedscale Operator with values chosen for your platform
(EKS, GKE, AKS, minikube, kind, OpenShift), verifies the result, and wires
the proxymock MCP server into your agent.
Open the ChatGPT desktop app, select Codex, and open your project's folder. The source is the install-speedscale skill. Paste this prompt to install it:
Use $skill-installer to install skills/install-speedscale from the speedscale/skills GitHub repository.
In your next message, ask Codex to use it:
Use $install-speedscale to set up Speedscale for my project. Check what is already installed and tell me which components I need. Ask before changing a Kubernetes cluster. Never print my API key.
If the new skill does not appear, restart the app. You can also use $install-speedscale later for an upgrade or repair.
ChatGPT on the web and mobile cannot install these skills directly from a GitHub URL. Those surfaces can use them after Speedscale publishes a plugin in ChatGPT's plugin directory.
| Route | Command |
|---|---|
| Claude Code plugin | /plugin marketplace add speedscale/skills then /plugin install speedscale@speedscale-skills |
| Any agent, via the skills CLI | npx skills add speedscale/skills |
| Already have proxymock | proxymock mcp install --yes installs every skill for Claude Code; proxymock mcp skills export --dir <your skills dir> for any other agent |
| Manual | copy skills/<name>/ into your agent's skills directory |
| Skill | What it does |
|---|---|
install-speedscale |
End-to-end install: CLIs, prerequisites, API key, operator Helm chart, verification, MCP wiring, upgrades, uninstall |
analyze-replay-report |
Explain a cloud report or a local proxymock replay run: verdict, the first failing response, root-cause bucket, and next steps |
record-traffic |
Record a service's inbound and outbound traffic, databases included, by wrapping it with proxymock record, and confirm every dependency was captured |
improve-mock-match-rate |
Tune the mocks for any technology (HTTP, gRPC, SQL, Redis, Kafka and more): offline fixes to a projected match rate, then an optional re-run that confirms them, including SQL reads served the wrong recorded row |
run-snapshot-replay |
Run a snapshot or recording as a replay, locally or in the cloud, defaulting to where it was recorded, and follow it to a verdict |
tune-snapshot-replay |
Tune the tests: loop until a replay's responses are accurate, changing one thing per run and keeping or reverting it, with progress on disk. Hands mock problems to improve-mock-match-rate |
Skills for testing your own service with your own recorded traffic. Run proxymock from the application's directory and pass the recording and target explicitly. These workflows use the local CLI and require no Speedscale Cloud account or mock-lab checkout. Use proxymock v2.5.1109 or newer; quality-loop includes a doctor that checks the app's environment.
Start with quality-loop when you need help choosing a command. For bundled helpers, resolve the script relative to the installed skill's SKILL.md, then invoke it by its absolute path while staying in your app directory. The install location may be outside your project.
For a runnable example, try the tutorial app in mock-lab.
| Skill | What it does | Wraps |
|---|---|---|
quality-loop |
The entry point: route an intent to the right native command or analysis skill, with the setup playbook, blueprint rules, gotcha catalog, and a doctor |
builds and execs the native commands |
proxymock-regression-test |
Replay a recording at a target and gate on the per-RRPair verdict (status and body) against a known-good baseline | proxymock replay --baseline --fail-on-new-mismatch |
proxymock-verify-fix |
Prove a bug fix by replaying the incident capture at the fixed build | proxymock replay --verify-fix |
proxymock-contract-test |
Check recorded or replayed traffic against an OpenAPI spec; mock a dependency straight from its spec | proxymock validate, proxymock generate |
proxymock-chaos-mock |
Inject faults into a mock: 503s, 429s with Retry-After, corrupt bodies, latency, connection faults, exact ratios |
proxymock mock --fault |
proxymock-load-test |
Replay at a target with parallel virtual users; latency percentiles, throughput, match rate, --fail-if SLO gates |
proxymock replay --vus --for --fail-if |
proxymock-perf-container |
Load-test one service with its downstream mocked, and judge the number honestly | proxymock replay --vus --for --load-test |
proxymock-compare-results |
Deep before/after comparison of two replay or recording sets; JSON, HTML, and an LLM digest | proxymock report --baseline, proxymock drift |
proxymock-summarize-recording |
Summarize a recording: hosts, endpoints, methods, status mix, volume | proxymock report --format prompt |
Every skill here ships inside proxymock and is mirrored from the Speedscale monorepo (speedctl/mcp/skills/) on each proxymock release by proxymock mcp skills export, so this repo always matches what the shipped binary installs. skills/.proxymock-managed lists the mirrored skills. Change a skill in the monorepo; a pull request that edits skills/ here is overwritten by the next release. The README, the plugin manifests and the CI check are maintained here. The plugin discovers every directory under skills/, so adding or removing a skill needs no manifest change. See CONTRIBUTING.md.
Docs: https://docs.speedscale.com · Community: https://slack.speedscale.com · Support: support@speedscale.com
Licensed under the Apache License 2.0.