Trusted — Risk Score 0/100
Last scan:23 hr ago Rescan
0 /100
agentmail
API-first email platform for AI agents — create/manage inboxes, send/receive emails, webhooks
AgentMail is a legitimate API-first email service skill with no malicious behavior detected. All capabilities are declared, documented, and consistent with its stated purpose of programmatic email management.
Skill Nameagentmail
Duration34.8s
Enginepi
Safe to install
No action needed. The skill is safe to use as documented.
ResourceDeclaredInferredStatusEvidence
Network READ READ ✓ Aligned All HTTP calls go to api.agentmail.to REST endpoints
Environment READ READ ✓ Aligned AGENTMAIL_API_KEY read via os.getenv() in all scripts
Filesystem NONE READ ✓ Aligned send_email.py:81 reads attachment files for encoding; documented and necessary
Shell NONE NONE No subprocess calls in any script
Skill Invoke NONE NONE No cross-skill invocations
Clipboard NONE NONE No clipboard access
Browser NONE NONE No browser automation
Database NONE NONE No direct database access
25 findings
🔗
Medium External URL 外部 URL
https://console.agentmail.to
SKILL.md:20
🔗
Medium External URL 外部 URL
https://your-domain.com/webhook
SKILL.md:75
🔗
Medium External URL 外部 URL
https://api.agentmail.to/v0
references/API.md:3
🔗
Medium External URL 外部 URL
https://your-app.com/webhook/support
references/EXAMPLES.md:69
🔗
Medium External URL 外部 URL
https://your-app.com/webhook/tasks
references/EXAMPLES.md:225
🔗
Medium External URL 外部 URL
https://your-app.com/tasks/
references/EXAMPLES.md:306
🔗
Medium External URL 外部 URL
https://ngrok.com/
references/WEBHOOKS.md:87
🔗
Medium External URL 外部 URL
https://abc123.ngrok-free.app
references/WEBHOOKS.md:136
🔗
Medium External URL 外部 URL
https://abc123.ngrok-free.app/webhook
references/WEBHOOKS.md:151
🔗
Medium External URL 外部 URL
https://myapp.com/webhook
scripts/setup_webhook.py:7
📧
Info Email 邮箱地址
[email protected]
SKILL.md:36
📧
Info Email 邮箱地址
[email protected]
SKILL.md:84
📧
Info Email 邮箱地址
[email protected]
SKILL.md:89
📧
Info Email 邮箱地址
[email protected]
SKILL.md:102
📧
Info Email 邮箱地址
[email protected]
references/API.md:35
📧
Info Email 邮箱地址
[email protected]
references/API.md:159
📧
Info Email 邮箱地址
[email protected]
references/EXAMPLES.md:297
📧
Info Email 邮箱地址
[email protected]
references/EXAMPLES.md:404
📧
Info Email 邮箱地址
[email protected]
references/EXAMPLES.md:428
📧
Info Email 邮箱地址
[email protected]
references/EXAMPLES.md:428
📧
Info Email 邮箱地址
[email protected]
references/EXAMPLES.md:440
📧
Info Email 邮箱地址
[email protected]
references/EXAMPLES.md:497
📧
Info Email 邮箱地址
[email protected]
references/WEBHOOKS.md:18
📧
Info Email 邮箱地址
[email protected]
scripts/check_inbox.py:7
📧
Info Email 邮箱地址
[email protected]
scripts/send_email.py:6

File Tree

8 files · 49.0 KB · 1729 lines
Markdown 4f · 1219L Python 3f · 505L JSON 1f · 5L
├─ 📁 references
│ ├─ 📝 API.md Markdown 229L · 4.3 KB
│ ├─ 📝 EXAMPLES.md Markdown 508L · 13.8 KB
│ └─ 📝 WEBHOOKS.md Markdown 294L · 7.2 KB
├─ 📁 scripts
│ ├─ 🐍 check_inbox.py Python 213L · 7.6 KB
│ ├─ 🐍 send_email.py Python 113L · 3.9 KB
│ └─ 🐍 setup_webhook.py Python 179L · 6.0 KB
├─ 📋 _meta.json JSON 5L · 128 B
└─ 📝 SKILL.md Markdown 188L · 5.9 KB

Dependencies 2 items

PackageVersionSourceKnown VulnsNotes
agentmail unpinned pip No Python SDK installed via pip; version not pinned in SKILL.md install instruction
flask unpinned pip No Optional dev dependency for webhook test server only

Security Positives

✓ No subprocess or shell execution — all operations use the agentmail SDK
✓ No credential exfiltration — AGENTMAIL_API_KEY is read only to initialize the SDK client
✓ No sensitive file access — no reads of ~/.ssh, ~/.aws, .env, or similar paths
✓ No obfuscation — no base64, eval, or encoded payloads
✓ No external IP connections — all API calls go to api.agentmail.to (documented service)
✓ No curl|bash or wget|sh remote script execution
✓ No hidden functionality — code matches SKILL.md documentation
✓ Proactive prompt injection awareness — SKILL.md documents webhook-based prompt injection risks and recommends Clawdbot allowlist filters
✓ Well-structured scripts with proper error handling and dependency checks
✓ Local webhook test server (Flask on localhost:3000) is development-only and clearly scoped