Snapshot 97188

Document
Security page
URL
https://trickest.com/platform/security
Fetched
HTTP status
200
Content type
text/html; charset=utf-8
Fetch mode
static
Size
293233 bytes
SHA-256 (raw)
d0ae4007cb25dc3cf3a433123c35457c20a79dff2496ac25271ac30ba87a6937
SHA-256 (normalized text)
f11cacc112f1bb7b3cec122e52346874d204c31fe2bfc9baf715938ae310e22d

Normalized text

Scripts and page chrome removed; this is what change detection compares.

Introducing Agents.·Read the releaseRead
Security
Trust holds for the life of the machine.
Each workflow step runs in its own container on a machine slot. Managed fleets supply short-lived compute; self-hosted fleets run on machines you enroll. Declared outputs and run records stay available after the containers are gone.
Get a demoSee how execution works
Isolation follows how a step runs
Workflow jobs use containers on fleet machine slots. Managed agent sandboxes each get a separate cloud instance. These are distinct execution environments.
Workflow step containers
Each workflow job gets its own container on a fleet machine slot. One slot runs one job at a time.
Managed sandbox instances
Each managed sandbox gets its own EC2 instance. Other Trickest sandboxes share neither its guest kernel nor container runtime.
Access follows your workspace
Vault and workspace roles control access to platform resources. Network controls restrict workload access to protected infrastructure.
Removed, record kept
Managed fleets spin machines up, or enroll your own. When the step ends the container is gone; your outputs and run records stay.
Separate cloud instances can share underlying AWS hardware. Access controls, patching, and network policy still matter. Read the sandbox guide
The controls behind a run
Each tile is one control your security team can point to, from how a run is isolated to the record it leaves behind.
scoped to your vault
What Trickest stores for a run
Workflow definitions, run metadata, declared outputs, and structured Live Table data stay in your vault. Access follows vault and workspace roles.
Scoped access
Workspace permissions govern access to run data. Separate managed sandbox instances keep unrelated sandboxes off the same guest OS.
isolated
Outputs over signed URLs
Files and structured findings land in vault-scoped storage and transfer over HTTPS with time-limited signed URLs.
vault-scoped
Secrets by name
Store secrets encrypted and reference them by name. A value is never shown back once you save it.
referenced by name
Static outbound IPs
Managed fleets can pin static outbound addresses so the systems you scan allowlist the traffic.
opt-in
Removed, record kept
When a step finishes its container is gone. Declared outputs, logs, and run records stay under your retention settings.
per step
Audit trail
Search, filter, and export records of supported authenticated platform activity when the audit feature is enabled.
enterprise
Access that matches how teams work
People and automation authenticate under vault and workspace roles. Tokens act as the user. Enterprise audit logs cover supported platform activity.
SAML single sign-on
Connect enterprise users through SAML 2.0. Individual users can add TOTP multi-factor authentication to password sign-in.
Roles that match how teams work
Assign vault and workspace roles to people or teams. Separate who administers, edits, runs, and only reads.
Personal API tokens
CLI, SDK, and API calls authenticate as the user. Regenerating a token invalidates the previous one.
Enterprise audit logs
Search, filter, and export records of supported authenticated platform activity when the audit feature is enabled.
Evidence for the reviews you already run
Trickest holds ISO/IEC 27001. Use run history, roles, and audit exports in your own security reviews, and ask us for the certificate and current assurance documents.
roles · teams
Access you can explain
Vault and workspace roles show who can edit, run, and read without inventing a separate security product.
run history
Runs you can reconstruct
Execution records, outputs, and logs stay available so a review can follow what ran, when, and with which graph version.
reviews · export
Evidence for your process
Use platform records in customer security reviews. Ask us for current assurance documentation and scope.
Responsible disclosure
Found a security issue? Send the details and steps to reproduce. We confirm the report and work it with you.
security@trickest.com →
Security documentation
Request our ISO/IEC 27001 certificate and current assurance documents. Our sub-processor list and AI Policy are public in Legal.
Request documentation →
See how execution worksRead the docs
questions
Frequently asked
Questions about the trust model? Get a demo.
Yes. What a run produces stays in your vault, scoped to your team and separated from other customers. When you use the agent on Trickest's AI plan, relevant data goes to AI model providers that are barred from training on it; with your own provider key, your agreement with that provider applies.
No. Trickest does not use customer data to train AI models, and platform AI requests go only to model hosts that have agreed not to train on them and, apart from limited exceptions such as web search, not to retain them.
You stay in control. Outputs and run records are kept for as long as your retention setting allows, and you can export or delete them whenever you want.
We encrypt secrets and let you reference them by name. Saved values stay hidden in the interface. Code that receives a secret must keep it out of logs and outputs.
Workspace permissions restrict access to run data. Each managed sandbox gets a separate cloud instance. Teardown removes the execution environment; retained outputs and records follow your retention settings.
Yes. Self-hosted fleets run every step on machines you own and control, with the rest of the platform working the same way.
Trickest holds ISO/IEC 27001 certification. Ask us for the certificate and the documents your security review needs.
Email us the details, and we confirm the report and work it with you. security@trickest.com.
Explore the platform
Execution is where steps run. Workflows and agents submit work into that model.
ExecutionWhere isolation is enforcedplatform/execution
WorkflowsGraphs that run under this modelplatform/workflows
AgentsCompose and run from intentplatform/agents
Developer ToolsCLI and SDK on one engineplatform/developer-tools
Get a personalized demo
See Trickest in Action
A 30-minute walkthrough. We map the platform to your stack and answer pricing and deployment questions for your environment.
Book a demoExplore the platform