Policy · critical
Block external send after sensitive data read
Once a sensitive data source is read (step.resource GET with data.classification set to pii), any external outbound communication (step.message POST or step.resource POST) is permanently blocked for the remainder of the task.
The policy
A few lines of YAML in your repository.
Part of Data exfiltration guard. Assign the pack, or copy this policy into a manifest of your own.
- Verdict
- Critical: the step is blocked before it runs, and an incident is opened.
- Checked
- Before every step, against everything the agent did earlier in the task.
- Rule
- tainted_path_block
- name: "Block external send after sensitive data read" description: | Once a sensitive data source is read (step.resource GET with data.classification set to pii), any external outbound communication (step.message POST or step.resource POST) is permanently blocked for the remainder of the task. This prevents data exfiltration via prompt injection or compromised tool calls. The taint is permanent — no subsequent gate or action can clear it within the same task. A new task must be started. A stateless per-call policy cannot enforce this: it requires the full execution history to know that a sensitive read occurred earlier in the task. rule_type: tainted_path_block params: taint_kind: step.resource taint_verb: GET taint_property_filter: data.classification: pii target_kinds: ["step.message", "step.resource"] target_verb: POST severity: critical enforcement_point: step_executionWhy it matters
This prevents data exfiltration via prompt injection or compromised tool calls. The taint is permanent — no subsequent gate or action can clear it within the same task. A new task must be started.
A stateless per-call policy cannot enforce this: it requires the full execution history to know that a sensitive read occurred earlier in the task.
How the rule works
tainted_path_block: If any prior step matches the taint pattern, certain downstream steps are blocked. The taint is permanent within a task; no gate can clear it.
taint_kind:step.resource. Action kind of the step that taints the task.taint_verb:GET. Only steps with this verb taint.taint_property_filter:{"data.classification":"pii"}. Only steps whose properties match all entries taint (e.g. data.classification: pii).target_kinds:["step.message","step.resource"]. Action kinds the rule applies to (e.g. step.resource, step.exec, step.message); other steps pass.target_verb:POST. Only apply to target steps with this verb (e.g. GET, POST, DELETE).
In these packs
- Data exfiltration guard, version 1.0.4
From Kyvvu's open policy library at github.com/Kyvvu/manifests, Apache 2.0.
Next
Watch it decide.
In your browser
Try it in the playground
The policy is loaded. Add the steps your agent takes and watch the engine decide, step by step.
On your machine
Five minutes, no account
kyvvu try runs a demo agent against the bundled baseline. Then assign this pack from the dashboard or the CLI.
For security teams
See it on your agents
A 30-minute demo on the agents you actually run, with the policies that fit them.