advisory-mining
Skill2 starsChanged 2 months agoMITView on GitHub
Mine GitHub Security Advisories and CVE databases for incomplete fixes, finding variant vulnerabilities in patched code or similar patterns in related packages.
In this repository255 other files it gives its agents.
- ad-acl-abuseSkill
- ad-adcs-access-and-relaySkill
- ad-adcs-persistenceSkill
- ad-adcs-template-abuseSkill
- ad-ad-discoverySkill
- ad-ad-persistenceSkill
- ad-auth-coercion-relaySkill
- ad-credential-dumpingSkill
- ad-gpo-abuseSkill
- ad-kerberos-delegationSkill
- ad-kerberos-roastingSkill
- ad-kerberos-ticket-forgingSkill
- ad-pass-the-hashSkill
- ad-sccm-exploitationSkill
- ad-trust-attacksSkill
- hunt-ntlm-infoSkill
- offensive-active-directorySkill
- hunt-llm-aiSkill
- hunt-rag-vectorSkill
- offensive-ai-securitySkill
- cloud-iam-deepSkill
- hunt-cloud-misconfigSkill
- hunt-k8sSkill
- m365-entra-attackSkill
- network-container-escapesSkill
- offensive-cloudSkill
- offensive-container-escapeSkill
- offensive-k8s-attacksSkill
- okta-attackSkill
- vmware-vcenter-attackSkill
- credential-password-sprayingSkill
- hunt-brute-forceSkill
- hunt-forgot-passwordSkill
- post-exploit-credential-recoverySkill
- meme-coin-auditSkill
- offensive-crypto-attacksSkill
- web3-auditSkill
- ctfSkill
- auth-bypassSkill
- code-injection-codegenSkill
- command-injectionSkill
- cross-pollinationSkill
- cve-hunting-methodologySkill
- decompression-bombSkill
- entity-expansionSkill
- fp-checkSkill
- method-clobberingSkill
- path-traversalSkill
- prototype-pollutionSkill
- recursion-dosSkill
- redosSkill
- sandbox-escapeSkill
- target-reconSkill
- web2-vuln-classesSkill
- evasion-av-edr-evasionSkill
- offensive-anti-forensicsSkill
- offensive-edr-evasionSkill
- offensive-keylogger-archSkill
- offensive-mitigationsSkill
- offensive-waf-bypassSkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 59
Read the diff carefully
- Distinctive: line 60
Ask: what did they fix? What did they NOT fix?
- Distinctive: line 88
If the vulnerability is in a common pattern (e.g., path.join without validation), search for it in similar packages:
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Advisory Mining -- Finding Incomplete Fixes
When to Use
Looking for high-acceptance-rate findings. Incomplete fix variants have ~95% acceptance rate because:
- The vulnerability class is already acknowledged
- The fix proves the maintainer cares about security
- The variant proves the fix was insufficient
Process
Step 1: Find Recent Advisories
# GitHub Advisory API -- recent npm advisories
gh api graphql -f query='
{
securityAdvisories(first: 20, orderBy: {field: PUBLISHED_AT, direction: DESC}, ecosystem: NPM) {
nodes {
ghsaId
summary
severity
publishedAt
vulnerabilities(first: 5) {
nodes {
package { name ecosystem }
vulnerableVersionRange
firstPatchedVersion { identifier }
}
}
}
}
}'
# Search by keyword
gh api "/advisories?ecosystem=npm&keyword=injection&per_page=20"
gh api "/advisories?ecosystem=pip&keyword=traversal&per_page=20"Step 2: Read the Patch Diff
For each advisory:
- Find the fix commit (linked in the advisory or CHANGELOG)
- Read the diff carefully
- Ask: what did they fix? What did they NOT fix?
# Find security-related commits
git log --oneline --all | grep -i "security\|fix\|vuln\|CVE\|patch\|sanitize"
# Read the patch
git show <commit_hash>
git diff <before_commit>..<fix_commit>Step 3: Check for Incomplete Fix Patterns
Common incomplete fixes:
| What Was Fixed | What Was Missed |
../ blocked | ..\ not blocked (Windows) |
__proto__ filtered | constructor.prototype not filtered |
| One regex fixed | Similar regex in same file not fixed |
| One function fixed | Wrapper function calls it differently |
| Parsing fixed | Serialization has same bug |
| Validation added | Can be bypassed with encoding |
| One entry point fixed | Other entry points not covered |
| Input sanitized | Error messages leak unsanitized data |
Step 4: Search for Same Pattern in Other Packages
If the vulnerability is in a common pattern (e.g., path.join without validation), search for it in similar packages:
# Use grep.app to find same pattern across repos
# See cross-pollination skill for detailsStep 5: Verify the Variant
Apply the fp-check skill to verify the variant is real before submitting.
NVD API
# Search NVD for CVEs by keyword
curl "https://services.nvd.nist.gov/rest/json/cves/2.0?keywordSearch=xml+parser+javascript"
# Search by CPE
curl "https://services.nvd.nist.gov/rest/json/cves/2.0?cpeName=cpe:2.3:a:vendor:product:*"CVSS Guidance
Variant findings typically get:
- Same CVSS as original if the variant has same impact
- Higher CVSS if the variant bypasses the fix AND adds new impact
- Lower CVSS if the variant has additional prerequisites
References
- Diff Analysis -- How to read security patches
- GHSA API -- GitHub Advisory API usage
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

