Everything the dashboard does, a script can do. The API lives at https://links.t-api.de/v1, speaks JSON, versions in the path, and authenticates with personal keys — the same API the macOS app and the browser extension are built on, so it covers real workflows, not a demo subset.
First link in thirty seconds
Create a key in the dashboard under API keys (the full tl_… token shows exactly once — store it like a password), then:
curl -X POST https://links.t-api.de/v1/links \
-H "Authorization: Bearer tl_your_key_here" \
-H "Content-Type: application/json" \
-d '{"destination_url": "https://example.com/release-notes", "slug": "v42"}'
List with GET /v1/links, read analytics with GET /v1/links/{link}/clicks/summary, rotate keys with POST /v1/api-keys and DELETE /v1/api-keys/{key}. The complete reference with copy-paste examples is the API guide; the machine-readable contract is OpenAPI 3.1.
Playing nice with automation
A few conventions keep scripts reliable. Every response carries API-Version and API-Latest-Version headers — check them on startup and fail loudly on major drift. Rate limits answer 429 with Retry-After: back off and retry instead of hammering. Errors are { message } (plus errors for validation), so log the message and move on.
Scope keys per use: one named key per script or CI job, revoked when the job dies. Creation and revocation both land in your activity feed and trigger a security notification, so a key minted at 3 a.m. by nobody you know is visible by breakfast. That is the whole security model — boring, explicit, and auditable — which is exactly what you want from infrastructure your deploys depend on.