SECURITY.md

Rama main · réplica de lectura del repositorio original.

# Security Policy

## Reporting a Vulnerability

If you believe you have found a security vulnerability in any Agenta repository, please report it to us through coordinated disclosure.

**Do not** report security vulnerabilities via public GitHub issues, pull requests, or discussions.

Instead, please send an email to **security@agenta.ai**.

---

## Information to Include

Please include as much of the following as you can to help us reproduce and resolve the issue:

- Type of issue (e.g., buffer overflow, SQL injection, cross-site scripting).
- Full paths of source files related to the issue.
- The location of the affected source code (tag, branch, commit SHA, or direct URL).
- Any special configuration or environment required to reproduce.
- Step-by-step instructions to reproduce.
- Proof-of-concept or exploit code (if possible).
- Expected vs actual behaviour and potential impact.
- Your contact details and disclosure timeline preference.

---

## Our Process

- **Acknowledgement**: We will acknowledge receipt within **3 business days**.
- **Triage**: We aim to complete an initial triage within **7 calendar days** and will share severity and next steps.
- **Remediation & Disclosure**: For critical vulnerabilities we aim to release a fix or mitigation within **30 days**. For other issues, typically within **90 days**. We will coordinate any public disclosure with you.
- We will provide status updates as needed during remediation.

---

## Safe Harbor

We respect and protect good-faith security research. If you follow this policy:

- We will not initiate legal action against you for good-faith testing conducted as part of coordinated disclosure.
- Do not access, modify, or exfiltrate data beyond what is necessary to demonstrate the issue.
- Do not disrupt production services or attempt destructive actions.

---

## Scope Exclusions

The following are **out of scope**:

- Third-party services not operated by Agenta.
- Physical security attacks or social engineering of personnel.
- Low-risk informational issues without security impact (e.g., generic version banners).
- Denial-of-service attacks (**we will not accept DoS testing against production**).

---

## Recognition & Credits

If you report a valid vulnerability and want public recognition, tell us how you wish to be credited (full name, handle, company, or anonymous). Recognition is discretionary and will be coordinated with you.

---

## Emergency / Out-of-band

If email is unavailable and you need an immediate or urgent channel, contact our general line: **team@agenta.ai** (monitored during business hours). For truly critical emergencies, include “EMERGENCY / SECURITY” in the subject line of your email.

---

## Contact retention & privacy

- Report metadata will be retained for incident tracking and compliance.
- Personal data you provide will be handled according to our privacy policy.
- We will only share reporter data internally on a need-to-know basis.

---