Security posture
No badges here. Here's what there is instead.
Most vendor security pages are a wall of logos from auditors you've never met. Ours isn't: an access model you control, a data boundary written into the agreement, and a stack you can read on GitHub before signing anything. If that trade doesn't clear your risk process, tell us early and we both save a quarter.
One business day · no sales sequence attached
Transparency
Pack
Names every real upstream project behind every themed name. Written to be checked rather than believed.
The pack tags every claim it makes — verified, stated, planned, not built, estimate — so you can grade it rather than trust it. The stack it describes is public.
- Certifications held
- None
- No SOC 2, no ISO 27001, nothing in progress.
- Standing privileged access
- None
- Just-in-time elevation you approve and time-box.
- Published for review
- 13 sections
- Transparency pack v2026.07-1, issued 2026-07-26.
Six practices. Read them before you sign.
What we hold, what we don't, what the agents see, where your work lives, and what happens if we disappear. Ordered so the hardest answer comes first.
What We Don't Hold
Starting here, because it's the first thing a risk team asks and the last thing most vendors put in writing. We hold no third-party security certifications. If your procurement process requires one, we aren't your vendor, and your risk team is right to say so.
- No SOC 2 report. Not Type I, not Type II, not in progress
- No ISO 27001 certification
- No authority to assess, audit, or certify anyone against any framework
- No insurance certificate is implied by this page — ask and we'll send the current one
The Access Model
We never get standing privileged access to your environment. Access is something you create, scope, log, and can revoke in one command — and we demonstrate that command on day one rather than waiting for you to ask.
- A named individual account you create, under your MFA. No shared credentials
- No standing privileged access — just-in-time elevation you approve, time-boxed
- Shell history and session logs ship to a store you control and we cannot delete
- Revocation demonstrated on day one, then drilled once a quarter
What the Agents See
We run a fleet of AI agents for the repetitive work — diffing upstream releases, ranking findings, drafting write-ups. That fleet operates behind a hard data boundary, and we put that boundary in the agreement rather than just describing it on a marketing page.
- Agents see inventory data, check results, and version strings
- Agents never see your hostnames, IP ranges, configs, credentials, or log contents
- The map from a tokenized host ID to a real hostname never enters a prompt
- Anthropic is a named subprocessor — ask for the list with data classes and retention terms
Where Your Work Lives
Everything we build for you ships as Terraform, Ansible, and Compose into a repository you own, on hardware you own. We are not a platform you log into. There is nothing to migrate off of if this arrangement ends.
- Your configs live in your repo, your telemetry never leaves your environment
- We don't need domain admin, and we hold no copies of your data
- On termination you keep a perpetual license to everything deployed. No exit fee
- We publish nothing derived from your environment without your written consent
We Eat Our Own Cooking
The same standard we hold your infrastructure to, applied to the machines we work from. Short list, because a short list of real things beats a long list of aspirations.
- A dedicated workstation with full-disk encryption and endpoint detection
- Hardware-backed MFA on everything that matters, secrets in a password manager
- US-based work location, documented and attested in the engagement letter
- The public baseline is the same stack this is all built on — read it before you sign
Continuity, Honestly
We're small, and the honest answer to “what if you disappear” isn't a name on an org chart you can't verify — it's an envelope you already hold and a stack you can already read.
- A handover envelope: access inventory, revocation commands, system state, escalation contacts
- Keys held by you and your own lead. We walk you through opening it on day one
- Continuity rests on the envelope and a public stack any competent engineer can read, not on one name
- If a named second engineer is a procurement requirement, ask — you get the name and rate in writing, or a straight no
- Your agreement names your own off-hours escalation path for the hours we aren't awake
Found something? Say so.
Found a hole in our baseline, the audit tooling, or this site? That's a real contribution to everyone running it. A hardening repo that ships a bug is the worst kind of bug, and we publish the correction with a date on it.
- Report to security@opsfox.io with a description and reproduction steps
- Allow a reasonable window to investigate and fix before public disclosure
- Don't access, modify, or delete data that isn't yours while researching
- No legal action against researchers acting in good faith. Ever
- There's no paid bounty. You get credited in the advisory unless you'd rather stay anonymous
Security contactsecurity@opsfox.io
Need this in a questionnaire format?
Ask and we'll send the vendor pack: current insurance certificate, the subprocessor list with data classes, the access model, the mutual NDA, and the data handling addendum. One business day, no sales sequence attached.