Hurl logo

Hurl

Community
damusix
hurl

Write and run HTTP API tests in Hurl's plain-text format. Use when testing REST/GraphQL APIs, writing integration or smoke tests for HTTP endpoints, chaining requests with captured values, polling async APIs, converting curl commands into maintainable tests, or producing executable API documentation. Trigger on .hurl files, the hurl or hurlfmt CLI, jsonpath/xpath response assertions, or any ask to 'test an API' with a simple text-based tool.

Overview

Publisherdamusix
Repositoryskills
Skill namehurl
Stars
63
Forks
3
Bundled files
6
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 6 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by damusix on GitHub. Read the source before you install it.

Installation

Install the Hurl AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/damusix/skills.git /tmp/skills
mkdir -p .claude/skills
cp -r /tmp/skills/hurl .claude/skills/hurl
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Hurl in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Hurl on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Hurl is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Hurl

Hurl runs HTTP requests defined in a plain-text format and asserts on the responses. One binary (hurl, built on libcurl), one readable file format: a .hurl file is a sequence of entries, each a request plus an optional response spec with asserts and captures. The same files work as smoke tests, integration tests, CI gates, and living API documentation — they read like the API contract they verify.

Current version: 8.0.1. Reference material in this skill is verified against hurl.dev for that release.

A minimal entry:

GET https://api.example.org/health
HTTP 200
[Asserts]
jsonpath "$.status" == "RUNNING"

Entries in one file run sequentially and share cookies and captured variables (session behavior). Files are independent and run in parallel under --test. That asymmetry drives all suite design: one user flow per file.

  1. Header vs section placement changes meaning. key: value lines directly after the URL are request headers — no section name. The same line inside [Query] is a query parameter. Body always comes last; [Form]/[Multipart] are the body and exclude a literal one.
  2. Use current section names [Query], [Form], [Multipart]. The legacy [QueryStringParams]/[FormParams]/[MultipartFormData] still parse — recognize them, don't write them.
  3. Predicates are typed. == true is a boolean check, == "true" a string check; == 458 number, == "458" string. Same rule for templates: == {{count}} compares typed, == "{{count}}" as string. matches on a non-string is a runtime error — convert with toInt/toString first.
  4. HTTP 200 is itself an assert (any version, status 200). HTTP/2 200 also asserts the protocol. Use HTTP * plus explicit status >= 200 asserts for ranges. Asserts run with or without --test — test mode only changes output and adds the recap.
  5. Redirects are not followed by default. Either assert each hop as its own entry, or set [Options] location: true (or --location) and assert on the final response — then redirects count/redirects nth 0 location audit the chain and url gives the final URL.
  6. retry turns asserts into wait conditions. [Options] retry: 10 + retry-interval: 500ms re-runs the entry until its asserts pass — the polling primitive for async jobs and server-readiness gates. Retries trigger on assert, capture, and runtime errors.
  7. Option precedence is env < CLI flag < [Options] section. [Options] applies to its entry only, except variable: which persists to later entries. skip: true exists only in [Options]; --test, --report-*, --variables-file, --secret are CLI-only.
  8. Secrets, not variables, for credentials. --secret name=value, HURL_SECRET_name, or a trailing redact on a capture redacts the value from logs and reports. Redaction is exact-match (register transformed variants separately) and does NOT apply to stdout or JSON-report response dumps.
  9. Reports accumulate. Every --report-html/json/junit/tap appends to an existing report. Clean the report path at the start of each CI run.
  10. Regex literals beat quoted patterns. /^\d{4}$/ needs no double escaping; "^\\d{4}$" does. The regex query and filter assert/extract the first capture group — a pattern without a group fails.
  11. Templates resolve variables only{{host}}, {{newUuid}}, {{newDate}}; no arithmetic or expressions. Compute in shell and inject with --variable or HURL_VARIABLE_name. Raw XML bodies are not template-aware; fence them as ```xml multiline strings to template them.
  12. 8.0 changes to honor: JSONPath engine is now RFC 9535 (edge cases differ from pre-8.0), --interactive was removed (use --from-entry/--to-entry, per-entry [Options] verbose: true, --curl replay), env vars are HURL_VARIABLE_name (old bare HURL_name form is gone), decode/format filters are deprecated for charsetDecode/dateFormat.

Build files incrementally — each step is runnable, so verify as you go.

  1. Smoke first. One entry, status only: GET {{host}}/health + HTTP 200. Run it: hurl --variable host=http://localhost:3000 health.hurl.
  2. Add asserts. Tighten the contract with [Asserts]: jsonpath/xpath/header queries, typed predicates. Prefer explicit asserts over an exact body literal unless you want golden-file equality.
  3. Chain the flow. Capture what later entries need ([Captures] token: jsonpath "$.access_token" redact), use it via {{token}}. Cookies flow automatically.
  4. Handle async. Entries that poll get [Options] retry: N + retry-interval, with the completion condition as an assert.
  5. Parameterize. Replace hosts and credentials with {{variables}}; create per-environment variables files (vars/local.env, vars/staging.env); pass tokens as secrets.
  6. Run as a suite. hurl --test tests/ (parallel files, recap, exit code 4 on assert failure). Add --error-format long so failures log the actual response.
  7. Wire CI. --report-junit for ingestion, --report-html for humans; clean report paths first; gate readiness with a stdin one-liner: printf 'GET %s\nHTTP 200' "$URL" | hurl --retry 60 --retry-interval 2s.

When a file fails: re-run with --very-verbose --to-entry N to isolate the entry with full bodies on stderr, or hurl --curl repro.txt file.hurl to export the exact requests as curl commands.

POST {{host}}/api/login
{"username": "{{user}}", "password": "{{password}}"}
HTTP 200
[Captures]
token: jsonpath "$.access_token" redact

GET {{host}}/api/me
Authorization: Bearer {{token}}
HTTP 200
[Asserts]
jsonpath "$.username" == "{{user}}"

Run: hurl --variable host=http://localhost:3000 --variable user=bob --secret password=$PASS login.hurl

POST {{host}}/jobs
HTTP 201
[Captures]
job_id: jsonpath "$.id"

GET {{host}}/jobs/{{job_id}}
[Options]
retry: 10
retry-interval: 500ms
HTTP 200
[Asserts]
jsonpath "$.state" == "COMPLETED"
rm -rf build/hurl-report build/hurl-junit.xml
hurl --test \
     --variables-file vars/staging.env \
     --error-format long \
     --report-junit build/hurl-junit.xml \
     --report-html build/hurl-report \
     tests/
echo "curl -X POST https://api.example.org/users -H 'Content-Type: application/json' -d '{\"name\":\"bob\"}'" | hurlfmt --in curl

References

  • File Format -- Entries, request/response anatomy, all sections and body types, templating, grammar gotchas
  • Asserting -- Every query type and predicate, implicit vs explicit asserts, typing rules
  • Captures and Filters -- Capture syntax and scope, the full filter table, chaining, secrets and redaction
  • CLI -- hurl and hurlfmt: invocation, test mode, all option groups, reports, exit codes, [Options] precedence
  • Testing Workflows -- Suite layout, retry-until polling, environments, CI integration, debugging, executable-docs export
  • Recipes -- 30 complete runnable examples: auth flows, chained requests, uploads, GraphQL, CI snippets

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Hurl AI skill do?

Write and run HTTP API tests in Hurl's plain-text format. Use when testing REST/GraphQL APIs, writing integration or smoke tests for HTTP endpoints, chaining requests with captured values, polling async APIs, converting curl commands into maintainable tests, or producing executable API documentation. Trigger on .hurl files, the hurl or hurlfmt CLI, jsonpath/xpath response assertions, or any ask to 'test an API' with a simple text-based tool.

Why use Hurl on TypingMind?

Because you install it once and use it with any model. Hurl is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Hurl in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/damusix/skills/tree/main/hurl. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Hurl?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Hurl?

As many as you like. As long as a model supports skills, you can use Hurl with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Hurl AI skill free?

It is published on GitHub by damusix. Check the repository for licensing terms. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇