skills.rest
Maybe4a6f7365
Maybe4a6f7365COMMUNITY
Explore all skills by this author →
GitHub

case-bundle-builder

Constructs case-bundle.yaml files from CVE/GHSA advisories, repositories, and fix-commit evidence.

1|Updated Aug 11, 2026
Search Tags:#provenance#cve#ghsa#patch analysis#osv#vulnerability ingestion#case bundle
One-click install
npx skills add https://github.com/Maybe4a6f7365/agentic-bug-bounty-framework --skill case-bundle-builder-maybe4a6f7365
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: case-bundle-builder
Source: https://github.com/Maybe4a6f7365/agentic-bug-bounty-framework/tree/main/skills/case-bundle-builder
Command: npx skills add https://github.com/Maybe4a6f7365/agentic-bug-bounty-framework --skill case-bundle-builder-maybe4a6f7365

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Turning a public vulnerability advisory into structured, auditable engineering training material requires manually correlating OSV/GHSA records, Git commits, patch diffs, regression tests, and maintainer discussion. This Skill automates that ingestion into a single canonical case-bundle.yaml with full provenance. ## Core Features & Use Cases - Advisory-to-Bundle Construction: Queries OSV and GitHub Advisory APIs, resolves vulnerable/fix commits, and populates all eight schema sections of a case bundle. - Evidence Preservation with Provenance: Saves raw API responses, patch artifacts, and SHA-256 hashes in a source ledger so every claim is traceable and auditable. - Human-Review Gating: Marks five star-rated fields (patch diff, changed tests, maintainer discussion, negative control, false-positive conditions) as requiring human sign-off before ingestion. - Use Case: Given CVE-2023-25813 (Sequelize SQL injection), the Skill fetches the GHSA record, clones the repository, extracts the introduced-to-fixed patch, identifies regression tests, and emits a review-ready case-bundle.yaml. ## Quick Start Build a case bundle for CVE-2023-25813 from the Sequelize repository and mark the star-rated fields for human review.

Frequently Asked Questions about case-bundle-builder

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I build a case bundle from a CVE or GHSA identifier?▼

Provide a CVE/GHSA ID or a repository URL with introduced and fixed commits. The Skill queries the OSV API, resolves the GitHub advisory, extracts the patch diff and regression tests, and fills all eight sections of case-bundle.yaml.

How do I look up vulnerability data with the OSV API?▼

POST to https://api.osv.dev/v1/query with an id, package/ecosystem plus version, or commit SHA. Preserve aliases, severity, affected ranges, and events, and never treat an ecosystem introduced value of 0 as a Git commit.

Does this Skill run PoCs or test live targets?▼

No. It is a build-time evidence-collection skill only. It explicitly prohibits executing PoCs, probing live systems, or bypassing repository access controls; output is engineering-source training material, not a finding.

What happens when the fix commit or patch is unavailable?▼

Missing evidence is recorded as null with the searches performed and the reason. The Skill never fabricates a SHA, and stop conditions halt the build for unverifiable commit pairs, embargoed patches, or out-of-scope repositories.

Which fields require human review before ingestion?▼

Five star-rated fields require human sign-off: behavioral_model.negative_control, engineering_evidence.patch_diff, changed_tests, maintainer_discussion, and agent_guidance.false_positive_conditions. They stay marked HUMAN-REVIEW-REQUIRED until a reviewer is identified.

Related Skills

Explore More Software Engineering Skills→
yhy0yhy0
74

ghsa-skill-builder

Generate security vulnerability Skills from GHSA and HackerOne reports.

Community
Advanced
chengzongcaichengzongcai
21

threat-hunting

Guides hypothesis-driven threat hunting and detection engineering with Sigma, YARA, and SIEM queries.

Community
Intermediate
makinuxmakinux
75

adversarial-panel

Run multi-model adversarial reviews with cross-critique and calibrated synthesis.

Community
Advanced
paperclipaipaperclipai
79.8k

deal-with-security-advisory

Automate the triage, patching, and publication workflow for GitHub Security Advisories.

Official
Advanced