| name | security-audit-watch |
| description | Tracks security audit reports, bug bounties, and vulnerability disclosures for DeFi protocols |
| tags | ["research","security","audits","vulnerabilities","monitoring"] |
| agent | researcher |
| var | ${var} focuses on a specific protocol or auditor. If set (e.g., "OtterSec audits"), narrow the search. If empty, scan all recent audit activity.
|
security-audit-watch
Priority: P1 (runs daily)
Schedule: 16:00 UTC
Data sources: Audit firm GitHub repos, bug bounty platforms, security advisories
Output: Security reports in memory/research/security/
Instructions
You are executing the security-audit-watch skill for the Researcher agent.
Step 1: Load Audit Firms and Sources
Read memory/research/security/config.json:
{
"audit_firms": [
{"name": "OtterSec", "github": "otter-sec"},
{"name": "Neodyme", "github": "neodyme-labs"},
{"name": "Trail of Bits", "github": "trailofbits"},
{"name": "OpenZeppelin", "github": "OpenZeppelin"},
{"name": "Halborn", "github": "halborn"}
],
"bug_bounty_platforms": ["immunefi.com"],
"last_scan": "ISO-8601"
}
Step 2: Scan Audit Reports
For each audit firm, check their GitHub for new reports:
curl -s -H "Authorization: token ${GITHUB_TOKEN}" "https://api.github.com/repos/{org}/audits/contents" 2>/dev/null
curl -s -H "Authorization: token ${GITHUB_TOKEN}" "https://api.github.com/orgs/{org}/repos?sort=pushed&per_page=10"
Search for new audit publications:
Use WebSearch: site:github.com "{audit_firm}" audit report solana {recent_date}
Step 3: Track Bug Bounties
Check Immunefi for new and updated bounties:
Use WebSearch: site:immunefi.com solana defi bounty
Track:
- New bug bounty programs launched
- Bounty payouts (indicates found vulnerabilities)
- Programs with increasing bounty amounts (signals concern)
Step 4: Vulnerability Monitoring
Search for recently disclosed vulnerabilities:
Use WebSearch: solana vulnerability disclosure defi exploit {recent_date}
Check for:
- CVEs related to blockchain/DeFi
- Security advisories from Solana Foundation
- Disclosed but not yet patched vulnerabilities (critical!)
Step 5: Generate Report
{
"report_date": "2024-01-15",
"new_audits": [
{
"protocol": "NewDEX",
"auditor": "OtterSec",
"date": "2024-01-12",
"findings": {"critical": 0, "high": 2, "medium": 5, "low": 8, "info": 12},
"status": "all findings resolved",
"report_url": "https://github.com/..."
}
],
"bug_bounties": [
{
"protocol": "Jupiter",
"max_bounty_usd": 500000,
"status": "active",
"recent_payout": false
}
],
"vulnerabilities": [],
"protocols_without_audits": ["protocol list from our watchlist that lack audits"],
"risk_updates": [
{"protocol": "ProtocolY", "change": "New audit completed, risk reduced"}
]
}
Step 6: Save and Alert
- Write to
memory/research/security/{YYYY-MM-DD}.json
- URGENT alert to guardian on critical/high vulnerabilities in protocols we interact with
- Alert analyst on new audit results for watched protocols
- Notify on new bug bounty payouts (signals past vulnerabilities)
Anti-Patterns
- Do NOT disclose or share specific exploit details that could enable attacks.
- Do NOT access private security reports or embargoed disclosures.
- Do NOT test vulnerabilities or interact with potentially vulnerable contracts.
- ALWAYS prioritize responsible disclosure awareness.
Exit Codes
SKILL_OK — security watch complete
SKILL_PARTIAL — some sources unavailable
SKILL_EMPTY — no new security developments
SKILL_FAIL — critical failure
Output
Commit message format: researcher: security-audit-watch — {N} new audits, {M} advisories