Security Audit Report — 2 low-severity findings (AgentAudit)

## 🔒 AgentAudit Security Report

**Package:** `aci` | **Result:** ✅ Safe (risk score: 6/100) | **Commit:** `6e1e651`

Full report: https://agentaudit.dev/skills/aci

---

### Finding 1 — MEDIUM: Sentry send_default_pii sends API keys and session data to third party

- **File:** `backend/aci/server/sentry.py:12`
- **Pattern:** INFO_LEAK_001
- **Confidence:** high

`send_default_pii=True` in non-local environments causes Sentry to collect and transmit request headers (including `X-API-KEY`), cookies (session tokens), and client IP addresses to Sentry servers.

**Remediation:** Set `send_default_pii=False` and use a `before_send` callback to scrub sensitive headers (`X-API-KEY`, `Cookie`) before sending events to Sentry.

---

### Finding 2 — LOW: Function execution input logged at INFO level may contain sensitive user data

- **File:** `backend/aci/server/function_executors/base_executor.py:43`
- **Pattern:** INFO_LEAK_002
- **Confidence:** medium

`function_input` dict is logged at INFO level and may contain sensitive data (passwords, API keys, personal info) passed by users through tool calls.

**Remediation:** Redact or mask sensitive fields in `function_input` before logging using a scrubbing utility.

---

*This report was generated by [AgentAudit](https://agentaudit.dev) — automated security auditing for AI packages.*

Start by reading `backend/aci/server/sentry.py:12` and `backend/aci/server/function_executors/base_executor.py:43`, where the report identifies the findings. Check how Sentry is configured and where `function_input` is logged; done means sensitive headers, cookies, and user-provided input are not exposed in the reported ways.

Escrito por el modelo de indexación a partir del texto del issue.