Link Crypt for AI agents

Give any HTTP-capable agent persistent memory for URLs — a reusable pattern for local IDE workflows, Hermes, and OpenClaw to search saved links, save new ones with tags, and keep research in sync across sessions.

What agents can do

  • Search the full library during a task (GET /search?q=...)
  • Filter by tag or domain (GET /links?tags=ai,python)
  • Save links found during a session with suggested tags (POST /links)
  • Add or update tags without opening the web app (POST /links/:id/tags)
  • Pull library stats or request an AI digest (GET /analyze/stats)

Authentication

All API calls use a bearer token. Generate a key in Preferences after signing in — keys start with lc_.

Authorization: Bearer lc_your_key_here

Never commit real keys to git or expose them in public agent instructions. Pass the key through an environment variable or your agent's secure secret store.

Key endpoints

GET/search?q=rust+async

Full-text search across titles, descriptions, domains, and saved page content.

GET/links?tags=ai,python&limit=50

List links with optional tag (AND), domain, or keyword filters.

POST/links

Save a URL with optional title, description, and tags array.

GET/analyze/stats

Library overview: link count, top tags, recent activity.

Full endpoint list including PATCH, DELETE, and /analyze/digest: API reference →

Quick start (curl)

# Search
curl https://getlinkcrypt.com/api/v1/search?q=vector+embeddings \
  -H "Authorization: Bearer lc_your_key"

# Save a link
curl -X POST https://getlinkcrypt.com/api/v1/links \
  -H "Authorization: Bearer lc_your_key" \
  -H "Content-Type: application/json" \
  -d '{"url":"https://example.com","tags":["ai","research"]}'

OpenClaw + Hermes shared setup

OpenClaw and Hermes can use the same Link Crypt agent skill instructions and API key strategy. Keep one shared SKILL.md style prompt, then reuse it across both runtimes.

  • Treat Link Crypt as the shared URL memory layer so local IDE sessions and remote agents read the same source of truth.
  • Use one shared prompt file that documents the same core endpoints (GET /search, GET /links, POST /links).
  • Store the API key once per environment in secret storage (never in repo files), and inject as Bearer auth.
  • Keep behavior consistent by using the same guardrail: do not delete links unless explicitly requested.

Per-agent setup guides