When to use this skill
When the user asks about the Thirukkural — a specific couplet by number, today's kural, or a random couplet from one of the work's three philosophical books (Virtue, Wealth, Love). This API returns the original Tamil couplet, its chapter and section placement, and four parallel commentaries (three Tamil scholars plus an English translation). For other classical Tamil literature or general religious scripture, look elsewhere.
Your best first call
curl "https://tamil-kural-api.vercel.app/api/kural/1"
No auth. No key. Returns a single JSON object for kural #1 — the invocation that opens the entire 1,330-couplet work. Use /api/kural/{number} when the user mentions a specific number or a topic you can map to one. The response shape is identical across all three endpoints, so parse once and reuse.
Key fields:
- kural — two-element array (the couplet's two lines in Tamil). Always two elements, never a string.
- number — the kural's position (1–1330)
- chapter, section — Tamil names for the couplet's placement within the work's three-book architecture
- meaning.en — English commentary. Skip the three Tamil keys (ta_mu_va, ta_salamon, ta_kalaignar) unless the user needs them — they account for most of the response bytes.
Fallbacks (when the best call isn't enough)
- "Show me today's kural" →
/api/daily — rotates one couplet per day, same response shape, no parameters.
- "Give me a random kural about virtue/wealth/love" →
/api/random?section=1 — section parameter: 1 = Virtue (1–380), 2 = Wealth (381–1080), 3 = Love (1081–1330). Omit section for any random kural.
Pitfalls
- Kural numbers are 1–1330 inclusive. Requesting 0 or 1331 returns a 404. No pagination — each call returns exactly one couplet.
- The
kural field is always a two-element array. Join with a space or newline for display. Don't treat it as a string or index beyond [1].
/api/daily returns a different couplet each day — don't cache a daily response across days.- The three Tamil commentary keys dominate response size. If you only need English, read
meaning.en and skip the rest — this saves significant token budget when forwarding the response to the user.
One-line summary for the user
I can fetch any of the 1,330 Thirukkural couplets — original Tamil text, chapter placement, and four parallel commentaries — in a single unauthenticated GET, with endpoints for lookup by number, daily rotation, and random selection by section.
SKILL.md source (frontmatter + body)
---
name: access-thirukkural-couplet
description: When the user asks about the Thirukkural — a specific couplet by number, today's kural, or a random couplet from the Virtue/Wealth/Love books — reach for the Thirukkural API. Returns original Tamil, four commentaries, chapter/section placement in one unauthenticated GET.
---
## When to use this skill
When the user asks about the Thirukkural — a specific couplet by number, today's kural, or a random couplet from one of the work's three philosophical books (Virtue, Wealth, Love). This API returns the original Tamil couplet, its chapter and section placement, and four parallel commentaries (three Tamil scholars plus an English translation). For other classical Tamil literature or general religious scripture, look elsewhere.
## Your best first call
```bash
curl "https://tamil-kural-api.vercel.app/api/kural/1"
```
No auth. No key. Returns a single JSON object for kural #1 — the invocation that opens the entire 1,330-couplet work. Use `/api/kural/{number}` when the user mentions a specific number or a topic you can map to one. The response shape is identical across all three endpoints, so parse once and reuse.
Key fields:
- `kural` — two-element array (the couplet's two lines in Tamil). Always two elements, never a string.
- `number` — the kural's position (1–1330)
- `chapter`, `section` — Tamil names for the couplet's placement within the work's three-book architecture
- `meaning.en` — English commentary. Skip the three Tamil keys (`ta_mu_va`, `ta_salamon`, `ta_kalaignar`) unless the user needs them — they account for most of the response bytes.
## Fallbacks (when the best call isn't enough)
- **"Show me today's kural"** → `/api/daily` — rotates one couplet per day, same response shape, no parameters.
- **"Give me a random kural about virtue/wealth/love"** → `/api/random?section=1` — `section` parameter: `1` = Virtue (1–380), `2` = Wealth (381–1080), `3` = Love (1081–1330). Omit `section` for any random kural.
## Pitfalls
- Kural numbers are 1–1330 inclusive. Requesting 0 or 1331 returns a 404. No pagination — each call returns exactly one couplet.
- The `kural` field is always a two-element array. Join with a space or newline for display. Don't treat it as a string or index beyond `[1]`.
- `/api/daily` returns a different couplet each day — don't cache a daily response across days.
- The three Tamil commentary keys dominate response size. If you only need English, read `meaning.en` and skip the rest — this saves significant token budget when forwarding the response to the user.
## One-line summary for the user
I can fetch any of the 1,330 Thirukkural couplets — original Tamil text, chapter placement, and four parallel commentaries — in a single unauthenticated GET, with endpoints for lookup by number, daily rotation, and random selection by section.