What this policy does
3 allow · 1 review · 2 models- 3 tools allowed immediately; 1 require human review; every other tool is denied.
- Models: gpt-5-mini, claude-sonnet-5.
- PII is allowed; prompt injection warns; data leaving the org warns.
- Up to 3000 tokens per request.
Tools
Allow / review / everything else denieddelete_research_dataDeny
draft_outreach_emailAllow
read_customer_profileAllow
score_icp_fitAllow
send_outreach_emailReview
Denied entries come from tools your registered agents advertise; any tool not listed here is also denied by default.
Models
- gpt-5-mini
- claude-sonnet-5
Safety
PII
Allow
Prompt injection
Warn
Data leaving org
Warn
Max tokens / request
3000
Assigned agents
0 totalNot assigned to any agent. This policy isn't enforcing anything yet — agents without an assigned policy are denied by default.
No agents assigned yet.
History1 version
| Version | Status | Created | Summary | |
|---|---|---|---|---|
| Version 1 | active | Jul 20, 2026, 5:35:20.532 PM PDT | Initial version |