Open Brain
GitHubGitHub RepoNeeds Review

elementalsouls/Claude-BugHunter

Claude-BugHunter packages security-hunting skills for Claude Code and other agent harnesses, with 83 skills and 681 disclosed-report patterns; its scope is authorized external testing.

GitHub
Open Source Open MarkdownOpen JSON
Source Summary

Claude-BugHunter describes a bundle of **83 skills** and **15 slash commands** for bug hunting and external red-team work. Its skills cover 24 vulnerability classes and draw on **681 disclosed-report patterns**, 433 of which are individually cited and auditable. The skill format can be installed across several agent harnesses; slash commands remain Claude Code-specific. For authorized security work, a solo builder could inspect its scope and selected skills as a starting point for repeatable agent procedures, including triage and evidence handling. The repository says relevant skills load by topic and includes a gate for checking engagement scope before submission. The bundle is focused on external attack surfaces. It explicitly excludes internal Active Directory attacks, post-exploitation, persistence, evasion, and several other areas; Claude runtime safeguards may also affect offensive-security requests.

Practical Implication

No separate implication has been recorded yet.

Agent-Ready Context
Claude-BugHunter describes a bundle of **83 skills** and **15 slash commands** for bug hunting and external red-team work. Its skills cover 24 vulnerability classes and draw on **681 disclosed-report patterns**, 433 of which are individually cited and auditable. The skill format can be installed across several agent harnesses; slash commands remain Claude Code-specific.
For authorized security work, a solo builder could inspect its scope and selected skills as a starting point for repeatable agent procedures, including triage and evidence handling. The repository says relevant skills load by topic and includes a gate for checking engagement scope before submission.
The bundle is focused on external attack surfaces. It explicitly excludes internal Active Directory attacks, post-exploitation, persistence, evasion, and several other areas; Claude runtime safeguards may also affect offensive-security requests.
Connected Context · Feed7 Judgment

This is a specialized, auditable security-skill bundle for authorized external attack-surface work, with explicit exclusions and scope checks. It gives the broader skill-catalog idea a concrete security use, while its cited patterns and safeguards describe structure rather than proving finding quality, safety, or portability across harnesses.

tech-leads-club/agent-skillsBoth address reusable skill catalogs, but BugHunter is a specialized security workflow while Agent Skills focuses on registry scanning and auditable installation; the latter's supply-chain controls are relevant to adopting this third-party bundle.On the Maintenance and Co-evolution of Agent Plugins: An Empirical Study of Claude Code Plugin MarketplacesThe study finds skill instructions and scripts can require coordinated maintenance. BugHunter's large bundle of skills and slash commands makes that maintenance concern pertinent, though the study does not evaluate this repository.addyosmani/agent-skillsBoth package repeatable agent procedures; BugHunter focuses on external security testing, while agent-skills emphasizes evidence gates across engineering workflows. Their scopes suggest complementary uses, not evidence that either validates the other's results.
Context Map
agentsecuritycoding#skills#coding-agents#harness-engineering