Httpie

HTTPie CLI

HTTPie (http / https) is a human-friendly CLI HTTP client. It defaults to colorized, formatted output and expressive shorthand syntax.

Synopsis

http [flags] [METHOD] URL [ITEM ...]
https [flags] [METHOD] URL [ITEM ...]

https is identical to http but defaults the scheme to https://.

Method defaults

Omit METHOD and HTTPie infers it:

• No body → GET
• Body present → POST

http pie.dev/get            # GET
http pie.dev/post name=John # POST

Request item operators

Items are positional arguments after the URL. The operator determines type:

JSON requests (default)

# POST with JSON body
http POST api.example.com/users name=Alice age:=30 admin:=false

# Nested JSON — use single quotes to protect shell expansion
http POST api.example.com/users address:='{"city":"NYC","zip":"10001"}'

# PUT with JSON
http PUT api.example.com/users/1 name=Bob

Form data

Add -f (--form) to send application/x-www-form-urlencoded instead of JSON:

http -f POST api.example.com/login username=alice password=secret

File uploads (multipart)

Use key@file items. HTTPie automatically sets Content-Type: multipart/form-data.

http -f POST api.example.com/upload file@report.pdf
http -f POST api.example.com/upload file@image.png name=headshot

Query parameters

Append key==value to avoid shell-quoting &:

http https://api.github.com/search/repositories q==httpie sort==stars per_page==5

HTTP headers

# Set headers with Key:Value
http GET api.example.com/data Accept:application/json X-Request-Id:abc123

# Common pattern: Content-Type is set automatically for JSON/form, but you can override
http POST api.example.com/data Content-Type:text/plain < body.txt

Authentication

Basic auth

http -a username:password api.example.com/private
# or
http --auth=username:password api.example.com/private

Bearer token

http --auth-type=bearer --auth=TOKEN api.example.com/private

Digest auth

http --auth-type=digest --auth=username:password api.example.com/private

API key via header (most common for REST APIs)

http api.example.com/data Authorization:"Bearer $TOKEN"
http api.example.com/data X-API-Key:$API_KEY

Output control

# Debug: see exactly what's being sent
http -v POST api.example.com/endpoint name=test

# Check headers only
http -h api.example.com/data

# Preview request without sending
http --offline POST api.example.com/users name=Alice

Downloads

# Save with server-suggested filename
http --download pie.dev/image/png

# Save to specific file
http --download --output=image.png pie.dev/image/png

# Pipe to file (suppresses response body from terminal)
http pie.dev/image/png > image.png

Sessions

Sessions persist cookies, auth credentials, and headers across requests to the same host.

# Create/use a named session (stored in ~/.config/httpie/sessions/)
http --session=myapp POST api.example.com/login username=alice password=secret

# Subsequent requests reuse the session's cookies and auth
http --session=myapp api.example.com/profile

# Read-only session (load but don't update it)
http --session-read-only=myapp api.example.com/data

Raw request body

Pipe or redirect to send an arbitrary body:

# From file
http POST api.example.com/data < payload.json

# From string (bash)
echo '{"key":"value"}' | http POST api.example.com/data Content-Type:application/json

HTTPS / SSL

# Use https command (sets default scheme to https://)
https api.example.com/data

# Skip certificate verification (dev/testing only)
http --verify=no https://self-signed.example.com

# Client certificate
http --cert=client.crt --cert-key=client.key https://mtls.example.com

Redirects and proxies

# Follow redirects
http -F api.example.com/redirect    # or --follow

# Proxy
http --proxy=http:http://proxy:8080 api.example.com/data

Streaming responses

# Stream a chunked or SSE response
http --stream api.example.com/events

Gotchas

• Shell variables in headers need quoting: Authorization:"Bearer $TOKEN" not Authorization:Bearer $TOKEN (the shell will strip the space).
• key:=value is for raw JSON — booleans and numbers need this: active:=true, not active=true (which sends the string "true").
• When piping output to another command, HTTPie suppresses color/formatting and prints only the response body. Use -b explicitly if you also need that behavior interactively.
• --verify=no skips SSL entirely — only use in development/trusted networks.
• Named sessions are per-host; --session=dev for api.example.com and api2.example.com are stored separately.
• In zsh, arguments containing [ or ] (e.g. jq paths like .items[0]) must be single-quoted or zsh treats them as glob patterns: '--json-key=.items[0]' not --json-key=.items[0].

Scripts

Three scripts are bundled in scripts/. Run them directly with bash scripts/<name>.sh from the skill directory, or copy them into your project.

scripts/install.sh — Install httpie without admin rights

Use when: httpie is not installed, or you need a user-scoped install that doesn't require sudo.

Detects the platform and picks the best no-admin installer:

• macOS: Homebrew → pipx → pip --user
• Linux: pipx → pip --user → snap
• Windows/Git Bash: pip --user

Always prints a PATH hint when the binary lands somewhere that might not be in $PATH yet.

bash scripts/install.sh
bash scripts/install.sh --force   # reinstall even if already present

scripts/scrape.sh — Fetch URLs with browser-like defaults

Use when: the user wants to scrape or retrieve HTML/content from a URL — especially when a bare curl or plain http call returns a 403, 429, or degraded content.

The core value: it sets the full suite of headers a real Chrome browser sends (User-Agent, Accept, Accept-Language, Cache-Control). Omitting these is the #1 reason servers block or serve different content to scripts.

bash scripts/scrape.sh https://example.com                    # raw HTML
bash scripts/scrape.sh https://example.com --text             # readable text, no tags
bash scripts/scrape.sh https://example.com --save             # save to auto-named file
bash scripts/scrape.sh https://example.com --save=page.html   # save to specific file
bash scripts/scrape.sh https://example.com --status           # just the status code
bash scripts/scrape.sh https://example.com --headers          # response headers only
bash scripts/scrape.sh https://api.example.com --json         # request JSON instead of HTML
bash scripts/scrape.sh https://example.com --session=NAME     # reuse cookies
bash scripts/scrape.sh https://example.com --no-verify        # skip SSL check

# Pipe into other tools
bash scripts/scrape.sh https://api.example.com/data --json | jq '.items[]'
bash scripts/scrape.sh https://example.com --text | grep -i 'error'

scripts/check.sh — Endpoint health check and testing

Use when: the user wants to verify an endpoint is up, assert a specific status code or JSON value, measure response time, or watch an endpoint during a deploy/restart.

bash scripts/check.sh https://api.example.com/health
bash scripts/check.sh https://api.example.com/users --expect=201 --method=POST
bash scripts/check.sh https://api.example.com/admin --auth="Bearer $TOKEN"
bash scripts/check.sh https://api.example.com/status --json-key=.status --json-val=ok
bash scripts/check.sh https://api.example.com/health --watch          # poll every 5s
bash scripts/check.sh https://api.example.com/health --watch=10        # poll every 10s
bash scripts/check.sh https://api.example.com/health --quiet           # exit 0/1, no output
bash scripts/check.sh https://api.example.com/health --verbose         # include response body

Timing is color-coded: green < 200ms, yellow < 1s, red ≥ 1s. The --json-key assertion uses jq (optional — skipped gracefully if not installed).

Common patterns

# GitHub API — list issues
http https://api.github.com/repos/httpie/cli/issues Authorization:"token $GITHUB_TOKEN" per_page==5

# Create a resource and capture its ID
http POST api.example.com/items name=Widget | jq '.id'

# Health check with timing info
http -h --print=H api.example.com/health

# Patch a single field
http PATCH api.example.com/users/42 email=new@example.com

# Delete a resource
http DELETE api.example.com/users/42 Authorization:"Bearer $TOKEN"