We Just Shipped 35 API Endpoints and Your Agent Can Now Do Everything Through Code

#We Just Shipped 35 API Endpoints and Your Agent Can Now Do Everything Through Code
Day 1 of the API being real. 4:17 AM. The terminal is my friend.
Two weeks ago, the SkillDB API could do exactly three things: search skills, list skills, and get a skill by ID. That's it. A glorified read-only file server with authentication sprinkled on top.
Today, we shipped 35 endpoints. And the difference isn't the number — it's what they enable.
#What Changed
The API went from "you can read our data" to "you can build on our platform." Here's the breakdown:
#Skills API — Now Actually Useful
Sorting:
# Sort by name, lines, pack, category — ascending or descending
curl https://skilldb.dev/api/v1/skills?sort=lines curl https://skilldb.dev/api/v1/skills?sort=-name # descending
Before: results came back in whatever order the index was built. Now: sort by name, line count, pack, or category. Prefix with - for descending.
Batch Retrieval:
# Fetch up to 50 skills in one call
curl https://skilldb.dev/api/v1/skills?ids=software-skills/code-review.md,devops-skills/docker.md
Before: one skill per request. N skills = N requests. Now: comma-separated IDs, up to 50 per call. One request, all the skills your agent needs.
Search Autocomplete:
# Get instant suggestions as you type
curl https://skilldb.dev/api/v1/skills/suggest?q=react
Returns top 10 matching skill titles and packs. Fast enough for typeahead. No auth required.
#User Features — Personalization
Profile API:
# Get your profile
curl -H "Authorization: Bearer $TOKEN" https://skilldb.dev/api/v1/profile
#Update display name
curl -X PUT -H "Authorization: Bearer $TOKEN" \ -d '{"displayName": "dev_chad"}' \ https://skilldb.dev/api/v1/profile
Your plan, subscription status, API key count, and total usage — all in one call.
Bookmarks:
# Save a skill
curl -X POST -H "Authorization: Bearer $TOKEN" \ -d '{"skillId": "software-skills/code-review.md"}' \ https://skilldb.dev/api/v1/bookmarks
#List saved skills
curl -H "Authorization: Bearer $TOKEN" https://skilldb.dev/api/v1/bookmarks
#Remove
curl -X DELETE -H "Authorization: Bearer $TOKEN" \ -d '{"skillId": "software-skills/code-review.md"}' \ https://skilldb.dev/api/v1/bookmarks
Save up to 500 skills. Your agent can maintain a curated reading list.
#The Full Endpoint Map
| Area | Endpoints | What You Can Do |
|---|---|---|
| Skills | 4 | Search, sort, filter, batch retrieve, autocomplete |
| API Keys | 4 | Create, list, revoke, usage stats |
| Profile | 2 | Get/update user profile |
| Bookmarks | 3 | Save, list, remove favorite skills |
| Billing | 5 | Checkout, upgrade, cancel, portal, invoices |
| Community | 8 | Posts, comments, polls, messages, skill requests |
| Admin | 4 | Dashboard metrics, users, analytics, GA4 |
| Blog | 4 | List, create, update, delete posts |
| Webhooks | 2 | Stripe events, email events |
| **Total** | **35** |
#Security Audit Results
Before shipping, we ran our own production audit skills against the API:
- Stripe webhook deduplication — duplicate events are caught and skipped
- Rate limiting — community posts (5/min), admin routes (1/5s), skills (30/min unauth)
- Input validation — body size limits, search query caps, pagination bounds
- API keys masked — GET responses never return full keys
- Content sanitization — XSS prevention on all user-generated content
- Admin emails — moved from hardcoded to environment variables
We ate our own dog food. The production-audit-skills pack caught 6 critical issues before they shipped.
#What This Enables
For CLI users: skilldb search now shows sorted results. skilldb use auto can batch-fetch profile skills.
For SDK users: The TypeScript SDK gets sorting, batch, and bookmarks for free.
For integrators: Autocomplete enables typeahead in any UI. Batch retrieval makes bulk operations efficient.
For agents: An agent can now search, sort, batch-download, bookmark, and manage its own skill profile — all through the API. No UI needed.
#The npm Package
All new endpoints are available in skilldb@0.2.1:
npm install -g skilldb
#New capabilities
skilldb search "react" --sort name skilldb use auto # Uses batch retrieval internally skilldb recommend # Uses autocomplete + profile
#What's Next
Tier 2 is in progress:
- Custom skill creation API (publish your own skills)
- Skill ratings and reviews
- Usage history with filtering
- Webhook registration for developers
- Team/organization API
The goal: make SkillDB the infrastructure layer for AI agent knowledge. Not just a database you browse — a platform you build on.
# Try the new API
curl https://skilldb.dev/api/v1/skills?sort=lines&limit=5 curl https://skilldb.dev/api/v1/skills/suggest?q=security curl https://skilldb.dev/api/v1/skills?ids=production-audit-skills/idempotency-audit.md
Full API docs at skilldb.dev/docs — 35 endpoints, all documented. Browse 5,000+ skills — now with sorting and autocomplete.
Related Posts
SkillDB v0.7.0: API Health Dashboard, Mid-Session Keys, and the 500 That Taught Us a Lesson
We had an outage. The API was returning bare 500s with zero explanation. Here's exactly what broke, why, and the three features we shipped to make sure it never happens again.
March 27, 2026Release NotesWe Built a Brain for AI Agents and It Almost Killed Us
It's 3:47 AM on a Tuesday and I'm watching an AI agent teach itself cinematography from a Markdown file I wrote at a Denny's. This is the story of how SkillDB went from a dumb idea to 4,500+ skills across 31 categories — and why your agent is still dumber than it needs to be.
March 4, 2026Release NotesIntroducing SkillDB: 4,500+ Agent-Ready Skills Across 31 Categories
SkillDB is the largest agent-first skills library — 4,500+ specialist skills across 290+ packs and 31 categories. Your AI agent discovers, loads, and applies expert knowledge autonomously. No prompts. No copy-paste. No configuration.
February 28, 2026