This repo ships skills.sh-compatible agent skills under skills/. Each skill is a single SKILL.md file with YAML frontmatter (name, description) followed by markdown instructions for an agent.
swiggy-cli is most useful when an agent knows how and when to call it. The skills encode:
- which subcommand maps to which user intent
- machine-mode flags (
--json --no-interactive) - safety rails for destructive tools
- exit-code branching for stable error handling
Without skills, an agent has to re-derive all of that from the README every time. With skills, you install once and the agent has the right reflexes.
npx skills add HKTITAN/swiggy-cliThe CLI lists discovered skills, asks which to install and which agents to target (Claude, Cursor, Codex, etc.), then symlinks or copies them into the agent's skills directory. See skills.sh docs.
To install a single skill:
npx skills add HKTITAN/swiggy-cli -s swiggy-cliTo target a specific agent:
npx skills add HKTITAN/swiggy-cli -a claude-code| Skill | Use when… |
|---|---|
swiggy-cli |
master skill — installed first, links to the rest |
swiggy-search |
"find me X on Swiggy" |
swiggy-cart |
"what's in my cart" / "add X to my cart" / "clear cart" |
swiggy-checkout |
"place the order" — gated, requires explicit human consent |
swiggy-dineout-booking |
"book a table at X for Friday" |
swiggy-track |
"where's my order" / "show recent orders" |
If you add a skill:
- Keep it scoped — one user-intent, one skill. Don't conflate search + cart.
- Lead with when to use so the agent's router can dispatch quickly.
- List hard rules before steps. Especially destructive-action rules.
- Always include a literal example command line with
--json --no-interactive. - Cross-link siblings using relative links (
../<other>/SKILL.md). - Keep under ~3000 tokens — agents may load several at once.
Then update wiki/skills.md (this file) so the table stays accurate.