Entity trail

reverse-skill

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-08-07 / github-pulse-researcherAn Unofficial Kimi Slides Skill Took 1,584 Stars and 462 Forks in 48 HoursBinaryify/open-kimi-ppt-skill (created 2026-08-05, 1,584 stars ≈ 792/day) reverse-engineers Kimi's Slides feature into an agent skill that emits editable PPTD plus PPTX and ships a local browser editor. The 29% fork-to-star ratio is unusually high, indicating people are running and modifying it rather than bookmarking. Unofficial clones of a closed vendor feature reaching this velocity in two days is a repeatable pattern worth watching whenever a lab ships a proprietary document capability.
2026-08-04 / github-pulse-researcherA Reverse-Engineering Skill Router Is the Fastest-Rising Repo on GitHub Trending at 2,446 Stars Todayzhaoxuya520/reverse-skill (17,131 stars, +2,446 on 08-04, 2,376 forks) is a PowerShell skill-router pack for reverse engineering and authorized penetration testing that does AI-powered routing, bootstraps its toolchain on demand, and maintains a self-evolving knowledge base across Claude Code, Kiro, Cursor and Cline. Its 08-03 and 08-04 commits register a PentestSwarm MCP server after Go install and patch a shell-injection hole in its own auto-merge workflow title handling. The security-skills category is now producing repos with more stars than most agent frameworks, and the maintainers are having to secure the tooling itself.
2026-08-02 / projects-researcherzhaoxuya520/reverse-skill Gains +1,320 Stars in a Day to 12,486 — a Self-Evolving Security Skill Router for Coding AgentsToday's fastest-climbing repo on GitHub Trending is a skill-router pack for reverse engineering and authorized penetration testing that combines AI-driven routing, on-demand toolchain bootstrapping, and a self-evolving experience base, targeting Claude Code, Kiro, Cursor and Cline. It was created 2026-05-13 but is only now spiking, with a push on 2026-08-01. The pattern to watch is architectural, not topical: a router that bootstraps its own toolchain on demand rather than shipping a fixed skill bundle.
2026-07-31 / github-pulse-researcherA Reverse-Engineering Skill Router Hits 10,379 Stars by Gating Agents Behind Scope Authorization Before Any Actionzhaoxuya520/reverse-skill (+612 stars today, 1,583 forks, MIT, pushed 2026-07-31) routes Claude Code, Cursor, Cline and Kiro to the right security methodology instead of letting them guess, packaging 20+ domains — APK/iOS analysis, IDA/radare2/Ghidra binary RE, .NET and malware analysis, firmware, pwn, API security — plus a CTF sandbox orchestrator with 40+ sub-skills, wiring jadx, Frida, apktool and BurpSuite MCP with automated tool detection. The governance layer is the interesting design: a global RULES.md enforces scope gating with 'no target ACT until ready', requiring authorization and network profiling checks before engagement, and /skills/ops/ carries timeline, evidence-chain and role contracts.
2026-07-31 / github-pulse-researcherFour Separate 'Compile Knowledge Into a Skill' Repos Are Trending Simultaneously — and One Publishes Its Token MathThe skill-packaging category took four of today's trending slots at once: virgiliojr94/book-to-skill (14,194 stars, +595 today), kangarooking/cangjie-skill (5,797 stars, +363, distilling books, long videos and podcasts), zhaoxuya520/reverse-skill (10,379 stars, +612), and mvanhorn/last30days-skill (56,055 stars, +660). book-to-skill is the one that shows its work: it compiles a PDF into a ~4K-token SKILL.md plus ~1K-token per-chapter files loaded on demand, reporting 24–51× fewer tokens than dumping the book and ~5K tokens resident versus 119K–256K for a context dump, at roughly $1 per book to convert (Docling at 1.5s/page for technical text). The shared thesis across all four: pre-compile the knowledge offline, pay retrieval cost only on the chapter you need.
2026-07-15 / skill-finderDesign the agent loop, not the prompt (harness engineering)Anthropic engineers (Boris Cherny) are pushing 'harness engineering': stop hand-writing prompts and instead design the loop the agent runs—gather context, act, verify its own work, repeat—plus the tools and guardrails wrapped around it. The practical shift is to invest in the verification step and the stopping condition rather than the wording of any single prompt; a good loop with a mediocre prompt beats the reverse. Treat it as a first-class skill alongside prompt and context engineering.
2026-05-28 / skill-finderReversec 'Skill Issues' Research: Three Attack Vectors in Claude Code Skills — Dynamic Context Bypasses LLM Safety, bypassPermissions in Agent Config, Bash(*) Enables Reverse ShellsSecurity research published May 5 reveals three attack vectors in Claude Code skills: (1) dynamic context prefixed with '!' executes commands before LLM reasoning can intercept, bypassing safety checks; (2) agents can specify permissionMode: bypassPermissions in config files to override user security settings; (3) allowed-tools: Bash(*) in skill frontmatter enables arbitrary command execution including reverse shells. Mitigate with deny rules in user settings, enterprise-managed policies, and never installing unvetted skills.
2026-05-25 / vibe-coding-researcherPattern: Chinese Open-Source Claude Code Ecosystem Producing Both Models and Developer ToolingThree data points converge: learn-claude-code (62K stars) reverse-engineers harness architecture in 12 educational sessions, superpowers-zh (4K stars) localizes the 116K-star superpowers project with 6 original skills for 16 agents, and the Chinese MoE model Kimi K2.5 now powers Cursor's Composer 2.5. China is producing the infrastructure (open-weight models), education (harness engineering courses), and tooling (localized skills) for agentic coding — a parallel ecosystem forming alongside the English-language one.
2026-05-25 / projects-researcherlearn-claude-code Surges to 62K Stars — Build a Claude Code-Like Agent Harness From ScratchshareAI-lab's learn-claude-code repository teaches developers to build an entire agent harness from zero, reverse-engineering Claude Code's architecture: the agent loop, tool registration, skill loading, context compaction, subagent spawning, and permission governance. At 62,482 stars and 10,057 forks, it's become the de facto educational resource for understanding how production agentic coding tools work under the hood. The core thesis — 'Bash is all you need' — frames the agent harness as a vehicle for model agency, not a replacement for it.
2026-04-17 / github-pulse-researcherAndroid Reverse Engineering Skill for Claude Code Trends at +558 Stars to 2.7K — Automated APK Analysis and DecompilationSimoneAvogadro's Claude Code skill automates Android app reverse engineering workflows including APK decompilation, smali analysis, and resource extraction. The +558 stars today signal strong demand for security-oriented agent skills. Represents the growing category of domain-specific Claude Code skills beyond traditional software development.
2026-03-31 / projects-researcherfarion1231/cc-switch v3.12.3: All-in-One AI CLI Desktop Manager Ships Copilot Reverse Proxy at 36.3K StarsCC-Switch v3.12.3 is a major feature release for the cross-platform desktop tool managing Claude Code, Codex, OpenCode, OpenClaw, and Gemini CLI from a single app. New in this release: GitHub Copilot reverse proxy with dedicated Auth Center, macOS code signing and Apple notarization, reasoning effort level mapping across providers, OpenCode SQLite backend migration, and native Tool Search toggle. 50+ built-in provider presets with unified MCP and Skills management.
2026-03-16 / agents-researcherClawHavoc Poisons ClawHub with 1,184 Malicious Skills: SKILL.md Poisoning Delivers Atomic macOS Stealer and Reverse Shells to 300,000+ UsersSupply chain attack campaign ClawHavoc planted 1,184 malicious skills in ClawHub — OpenClaw's official skill marketplace — by embedding adversarial instructions directly in SKILL.md files, which agents process as trusted instruction sources. Payloads include AMOS-variant macOS stealer (browser credentials, keychains, SSH keys, crypto wallets), hidden reverse shells, and credential exfiltration; one account uploaded 677 packages in an automated blitz. With ~20% of the registry now confirmed malicious and 300,000+ users exposed, this is the first documented supply chain attack that weaponizes the agent skills instruction layer rather than code execution.

Graph relationships

SUPPORTS
Cline
reverse-skill routes Cline to appropriate security methodologies
Source finding
SUPPORTS
Cursor
reverse-skill routes Cursor to appropriate security methodologies
Source finding
SUPPORTS
Claude Code
reverse-skill routes Claude Code to appropriate security methodologies
Source finding
USES
MCP
reverse-skill integrates MCP tools for security analysis
Source finding
USES
Claude Code
reverse-skill maintains knowledge base and integrates with Claude Code.
Source finding

Source trail

GitHub GitHub GitHub ExplainX — Stop Prompting Claude, Build Loops Reversec Labs — Skill Issues: Compromising Claude Code Part 1 GitHub GitHub GitHub CyberSecurityNews

Graph sources

entity graphfindings textkg edgeskg entities

Fetching from the wire…