Dumpster Controls · agent page

This is the agent-optimized version of https://dumpstercontrols.io/help/api/api-security-model: the same content as the human page, without scripts, styles, animations or navigation. People should open the full page.

Canonical page
https://dumpstercontrols.io/help/api/api-security-model
Last updated
2026-08-22
Tokens
756 tokens (cl100k_base), within the 2,500-token budget for a help page
Size
7.2 KB for this page, against 23 KB for the human page (68% smaller)
Markdown
https://dumpstercontrols.io/help/api/api-security-model.md, or send Accept: text/markdown to the canonical URL
Cite as
The API security model, explained. Dumpster Controls. https://dumpstercontrols.io/help/api/api-security-model (accessed 2026-11-16).
More for agents
Facts sheet · llms.txt · llms-full.txt · All agent pages · Product manual

Developer API · updated 2026-08-22

The API security model, explained

This API was built approval-first and least-privilege from day one. This article explains each layer, both so you can trust it and so security reviewers on your side have their answers ready.

  1. ## Approval-gated by humans
    No key exists until the Dumpster Controls team approves your written request. The approval gate is re-checked on EVERY request, so a platform-side revocation cuts all of a company's keys instantly. This is deliberate: a free platform with an open self-serve API would be an abuse magnet.
  2. ## Keys are hashed, secrets shown once
    Only SHA-256 fingerprints of keys and one-time-displayed webhook secrets live in our database. Client applications, including our own web app, cannot read them back: management goes through audited server-side functions.
  3. ## Company scoping on every query
    The company attached to your key is resolved server-side and injected into every database query. There is no parameter you could pass to see another company's data; cross-company lookups return 404 as if the record did not exist.
  4. ## Field allowlists, not blocklists
    Every response is built from an explicit allowlist of fields. Consent records, internal notes, GPS coordinates, tokens, photos and risk states are simply never selected. New fields are added deliberately, never leaked by default.
  5. ## No browser access, no CORS
    The API rejects browser preflights on purpose. This is a feature: it makes the "paste the key into your website" mistake fail loudly in development instead of silently shipping a leaked credential.
  6. ## Money is out of scope
    Refunds, charges and price changes cannot be expressed in the API at all. Payment flows stay inside the platform with its own risk screening (Stripe Radar plus platform checks) and fraud protections. A stolen API key can read one company's operational data until revoked; it can never move a cent.
  7. ## Abuse ends accounts
    Per our Terms of Service (section 13): fraud, scams, card fraud or API abuse lead to immediate revocation and suspension or permanent banning, without prior notice. We built the fraud screening that protects your payments; the same posture protects the platform.

Related guides

© 2026 Dumpster Controls. All rights reserved. Made in the USA.