[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-77787":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":9,"totalLinesOfCode":9,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":9,"subscribersCount":16,"size":16,"stars1d":16,"stars7d":17,"stars30d":18,"stars90d":16,"forks30d":16,"starsTrendScore":16,"compositeScore":19,"rankGlobal":9,"rankLanguage":9,"license":9,"archived":20,"fork":20,"defaultBranch":21,"hasWiki":20,"hasPages":20,"topics":22,"createdAt":9,"pushedAt":9,"updatedAt":30,"readmeContent":31,"aiSummary":32,"trendingCount":16,"starSnapshotCount":16,"syncStatus":33,"lastSyncTime":34,"discoverSource":35},77787,"Pentest-Swarm-AI","Armur-Ai\u002FPentest-Swarm-AI","Armur-Ai","Autonomous penetration testing using a swarm of AI agents. Orchestrates recon, classification, exploitation, and reporting specialists with ReAct reasoning — supports bug bounty, continuous monitoring, and CTF modes. Built with Go, Claude API, and 7+ native security tools.",null,"https:\u002F\u002Fgithub.com\u002FArmur-Ai\u002FPentest-Swarm-AI","Go",1774,360,13,12,0,187,503,82.67,false,"main",[23,24,25,26,27,28,29],"ai-agents","bug-bounty","cybersecurity","offensive-security","penetration-testing","penetration-testing-framework","penetration-testing-tools","2026-06-11 04:06:44","\u003Cp align=\"center\">\n  \u003Ch1 align=\"center\">Pentest Swarm AI\u003C\u002Fh1>\n  \u003Cp align=\"center\">\n    \u003Cstrong>The first open-source pentesting tool built on a real swarm — not just multiple agents in a row.\u003C\u002Fstrong>\n  \u003C\u002Fp>\n  \u003Cp align=\"center\">\n    \u003Ca href=\"#quick-start\">Quick Start\u003C\u002Fa> ·\n    \u003Ca href=\"#what-makes-this-a-swarm\">Swarm vs. Multi-Agent\u003C\u002Fa> ·\n    \u003Ca href=\"#how-the-swarm-works\">How It Works\u003C\u002Fa> ·\n    \u003Ca href=\"#comparison\">Compare\u003C\u002Fa> ·\n    \u003Ca href=\"IMPLEMENTATION_PLAN.md\">Roadmap\u003C\u002Fa>\n  \u003C\u002Fp>\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fstars\u002FArmur-Ai\u002FPentest-Swarm-AI?style=for-the-badge&color=f59e0b\" alt=\"Stars\">\n  \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FGo-1.24-00ADD8?style=for-the-badge&logo=go\" alt=\"Go\">\n  \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-AGPL%203.0-blue?style=for-the-badge\" alt=\"License\">\n  \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FAI-Claude%20%7C%20Ollama-purple?style=for-the-badge\" alt=\"AI\">\n  \u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fstatus-alpha-orange?style=for-the-badge\" alt=\"Status\">\n\u003C\u002Fp>\n\n\u003C!-- Once trendshift.io lists the repo, replace the numeric id below.\n     PentAGI's badge (for reference): https:\u002F\u002Ftrendshift.io\u002Frepositories\u002F15161\n\u003Cp align=\"center\">\n  \u003Ca href=\"https:\u002F\u002Ftrendshift.io\u002Frepositories\u002F__ID__\" target=\"_blank\">\n    \u003Cimg src=\"https:\u002F\u002Ftrendshift.io\u002Fapi\u002Fbadge\u002Frepositories\u002F__ID__\"\n         alt=\"Armur-Ai\u002FPentest-Swarm-AI | Trendshift\"\n         width=\"250\" height=\"55\"\u002F>\n  \u003C\u002Fa>\n\u003C\u002Fp>\n-->\n\n\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"docs\u002Fdemo-flashy.gif\" alt=\"Pentest Swarm AI — live campaign demo\" width=\"900\">\n\u003C\u002Fp>\n\n\u003Cp align=\"center\">\n  \u003Cimg src=\"banner\u002Fpentest-swarm-ai-banner.gif?v=3\" alt=\"Pentest Swarm AI architecture\" width=\"800\">\n\u003C\u002Fp>\n\n### Built for the Mythos era\n\nAnthropic's [Claude Mythos](https:\u002F\u002Fred.anthropic.com\u002F2026\u002Fmythos-preview\u002F) — released through [Project Glasswing](https:\u002F\u002Fwww.anthropic.com\u002Fglasswing) in April 2026 — surfaced thousands of zero-days across every major operating system and browser. Frontier reasoning has crossed a threshold; the bottleneck is no longer the model.\n\n**Pentest Swarm AI is the toolchain a model like that needs to operate.** Live access to nmap, sqlmap, Burp, ZAP, Metasploit, and the rest of the offensive stack. Multi-agent coordination through a stigmergic blackboard. Evidence capture, dedup, submission-ready reports. Wire in the model of your choice today — Claude Sonnet, Opus, Llama, anything OpenAI-compatible — and swap in Mythos the day access opens.\n\n*Not affiliated with Anthropic. Mythos and Glasswing are Anthropic projects.*\n\n---\n\n> ### Credits & Inspiration\n> This project stands on the shoulders of giants. We credit and thank these projects for pioneering AI-powered offensive security:\n>\n> - [**PentestGPT**](https:\u002F\u002Fgithub.com\u002FGreyDGL\u002FPentestGPT) — the OG that proved LLMs can pentest\n> - [**PentAGI**](https:\u002F\u002Fgithub.com\u002Fvxcontrol\u002Fpentagi) — fully autonomous agent architecture\n> - [**Strix**](https:\u002F\u002Fgithub.com\u002Fusestrix\u002Fstrix) — AI hackers that find and fix vulns\n> - [**CAI**](https:\u002F\u002Fgithub.com\u002Faliasrobotics\u002Fcai) — cybersecurity AI framework, 3600x faster than humans\n> - [**HackingBuddyGPT**](https:\u002F\u002Fgithub.com\u002Fipa-lab\u002FhackingBuddyGPT) — LLM hacking in 50 lines of code\n> - [**Shannon**](https:\u002F\u002Fgithub.com\u002FKeygraphHQ\u002Fshannon) — white-box AI pentester\n> - [**BlacksmithAI**](https:\u002F\u002Fgithub.com\u002Ffr0gger\u002FBlacksmithAI) — multi-agent pentest framework\n> - [**PentestAgent**](https:\u002F\u002Fgithub.com\u002FGH05TCREW\u002Fpentestagent) — black-box AI security testing\n> - [**Pentest Copilot**](https:\u002F\u002Fgithub.com\u002Fbugbasesecurity\u002Fpentest-copilot) — AI-driven pentest agent\n>\n> Their open-source contributions made tools like this possible.\n\n> **Legal Disclaimer:** Pentest Swarm AI is designed exclusively for **authorized security testing**, **bug bounty programs**, **CTF competitions**, and **educational research**. You must obtain explicit written permission from the target system owner before running any scan. Unauthorized access to computer systems is illegal under the Computer Fraud and Abuse Act (CFAA), the Computer Misuse Act, and equivalent laws worldwide. The authors and contributors of this project accept **no liability** for misuse, damage, or any illegal activity conducted with this tool. By using this software, you agree that you are solely responsible for ensuring your use complies with all applicable laws and regulations. **Do not use this tool against systems you do not own or have explicit authorization to test.**\n\n---\n\n## What makes this a swarm?\n\nMost \"multi-agent\" pentesting tools are a single planner LLM dispatching to specialist agents in a fixed order — recon → classify → exploit → report. That's a **pipeline**, not a swarm.\n\nPentest Swarm AI is built around three swarm-intelligence primitives:\n\n- **Stigmergy** — agents coordinate by reading and writing findings on a shared blackboard, not by a central planner telling them what to do. A finding's *pheromone weight* biases other agents toward it and decays over time, so stale paths die naturally.\n- **Emergence** — attack chains appear that no single agent planned. A recon finding wakes the classifier; a high-severity classification wakes the exploit agent; exploit results feed back into the board and wake the report agent. Order isn't prescribed — it emerges from the blackboard state.\n- **Decentralization** — each agent runs its own *trigger predicate*. Add a new agent with its own predicate and it joins the swarm without anyone rewriting the orchestrator.\n\nWe built this because the category was empty. Every tool marketed as \"swarm\" was actually a pipeline. If you find a counter-example, open an issue — we'll add them to the [comparison table](#comparison).\n\nSee [**IMPLEMENTATION_PLAN.md**](IMPLEMENTATION_PLAN.md) for the technical deep-dive on stigmergy, pheromone decay, the Postgres-backed blackboard, and why we didn't build on Google ADK \u002F CrewAI \u002F AutoGen.\n\n---\n\n## Quick Start\n\n```bash\n# Install (pick one)\nbrew install Armur-Ai\u002Ftap\u002Fpentestswarm            # macOS (Homebrew tap)\ndocker run --rm -e ANTHROPIC_API_KEY=sk-ant-... \\\n  ghcr.io\u002Farmur-ai\u002Fpentestswarm:latest \\\n  scan example.com --scope example.com             # Docker one-liner\ngo install github.com\u002FArmur-Ai\u002FPentest-Swarm-AI\u002Fcmd\u002Fpentestswarm@latest  # Go\n\n# One API key, one command, one swarm.\nexport PENTESTSWARM_ORCHESTRATOR_API_KEY=sk-ant-your-key-here\npentestswarm scan example.com --scope example.com --swarm --follow\n```\n\nThat's the whole setup. No Ollama, no model download, no GPU — just a Claude API key.\n\nRunning inside a GitHub Actions workflow? There's an action for that — see [`deploy\u002Fgithub-action\u002Fexample-workflow.yml`](deploy\u002Fgithub-action\u002Fexample-workflow.yml).\n\n---\n\n## How the swarm works\n\n```\n                         YOU\n                          |\n                   pentestswarm scan example.com --swarm\n                          |\n               ┌──────────▼──────────┐\n               │   SEED: TARGET_REG  │\n               └──────────┬──────────┘\n                          ▼\n     ┌────────────────────────────────────────────────────────┐\n     │              SHARED BLACKBOARD (pgvector)              │\n     │                                                        │\n     │   SUBDOMAIN · PORT_OPEN · HTTP_ENDPOINT · TECHNOLOGY   │\n     │   CVE_MATCH · MISCONFIGURATION · EXPLOIT_CHAIN         │\n     │   EXPLOIT_RESULT · CAMPAIGN_COMPLETE                   │\n     │                                                        │\n     │   (each finding has a pheromone weight that decays)    │\n     └──┬─────────────┬─────────────┬─────────────┬───────────┘\n        │             │             │             │\n        │ triggers:   │ triggers:   │ triggers:   │ triggers:\n        │ TARGET_REG  │ raw recon + │ CVE_MATCH   │ CAMPAIGN_\n        │             │ pheromone>  │ pheromone>  │ COMPLETE\n        │             │ 0.2         │ 0.5         │\n        ▼             ▼             ▼             ▼\n   ┌─────────┐  ┌─────────┐   ┌─────────┐   ┌─────────┐\n   │  RECON  │  │CLASSIFY │   │ EXPLOIT │   │ REPORT  │\n   │         │  │         │   │         │   │         │\n   │ runs 8  │  │ maps    │   │ builds  │   │ queries │\n   │ tools,  │  │ CVEs,   │   │ attack  │   │ board   │\n   │ writes  │  │ scores  │   │ chains  │   │ →md\u002F    │\n   │ per     │  │ CVSS,   │   │ per     │   │ html\u002F   │\n   │ finding │  │ writes  │   │ finding │   │ json\u002F   │\n   └─────────┘  └─────────┘   └─────────┘   │ sarif   │\n                                            └─────────┘\n```\n\nKey behaviours:\n\n1. **Agents are independent.** Any one of them can be removed, replaced, or added without rewiring the others.\n2. **Pheromones decay per-finding-type.** A `PORT_OPEN` stays hot for hours; a `SESSION` for minutes. Config-driven half-lives.\n3. **Scope is enforced at the tool layer and again at the executor.** Defence in depth — `--scope` is not bypassable.\n4. **Cleanup is always registered before execution.** SIGINT, crashes, and budget exhaustion all trigger reverse-order cleanup. See `internal\u002Fpipeline\u002Fcleanup_memory.go` and `cleanup.go`.\n5. **Prompt caching on Claude** cuts cost and latency on repeated system prompts (enabled by default for recon + classifier).\n\n---\n\n## Comparison\n\nHow we position vs. the rest of the ecosystem. We'll ship real benchmark numbers in a future release (see [Phase 3.3](IMPLEMENTATION_PLAN.md#phase-33--benchmarks-the-credibility-lever)).\n\n| Tool | Architecture | Executes vs. suggests | Memory | Tools wired | MCP | Swarm? |\n|---|---|---|---|---|---|---|\n| **Pentest Swarm AI** | Stigmergic blackboard | Executes | pgvector + pheromones | 8 ProjectDiscovery + nmap; sqlmap \u002F Burp MCP \u002F Metasploit in roadmap | Yes | ✅ real |\n| PentestGPT | Single-agent ReAct | Suggests | None | None native | No | No |\n| HackingBuddyGPT | Single-agent | Executes | Run logs | Shell passthrough | No | No |\n| PentAGI | 4 agents + planner | Executes | pgvector | 40+ via MCP\u002Fshell | Partial | Pipeline |\n| Shannon | White-box + browser | Executes | Session state | Browser DOM | No | Pipeline |\n| HexStrike | MCP tool wrapper | Delegates to client LLM | None (stateless) | 150+ via MCP | Yes | No |\n| Pentest-R1 | RL-tuned LLM | Executes | Trajectory | CTF-scope | No | No |\n\nIf any entry here is wrong or out of date, please open a PR — we want this table to stay honest.\n\n---\n\n## Feature status\n\nHonesty labels: *stable* means shipped + tested, *beta* means works but rough edges, *alpha* means experimental, *planned* means in the [roadmap](IMPLEMENTATION_PLAN.md).\n\n| Feature | Status | Notes |\n|---|---|---|\n| Sequential 5-phase runner | **stable** | Default mode; battle-tested core |\n| Stigmergic swarm scheduler | **alpha** | `--swarm` flag; memory-backed blackboard wired |\n| ProjectDiscovery toolchain | **stable** | subfinder, httpx, nuclei, naabu, katana, dnsx, gau |\n| `nmap` adapter | **stable** | XML parsed; scope-validated |\n| Cleanup registry | **stable** | Always runs on SIGINT \u002F exit \u002F budget-cancel |\n| Claude prompt caching | **stable** | Enabled for recon + classifier by default |\n| `--strict` LLM mode | **stable** | Promotes LLM errors to fatal |\n| CVSS v3.1 scoring | **stable** | FIRST spec |\n| Postgres blackboard backend | **beta** | Migration shipped; runner uses memory-board for now |\n| MCP server | **beta** | `pentestswarm mcp serve` |\n| VS Code extension | **beta** | `deploy\u002Fvscode\u002F` |\n| GitHub Action | **beta** | `deploy\u002Fgithub-action\u002Faction.yml` with SARIF |\n| Swarm playbooks (5) | **beta** | `playbooks\u002F{bug-bounty,external-asm,ci-cd,internal-network,ctf-solver}.yaml` |\n| Live dashboard | **alpha** | `web\u002F`; UI built, wiring to live campaigns in progress |\n| Burp MCP bridge | **planned** | Wave 2 |\n| Metasploit \u002F ZAP \u002F sqlmap adapters | **planned** | Wave 2 |\n| Fine-tuned Pentest-Swarm model | **planned** | Wave 3 (Pentest-R1 recipe) |\n| Cybench \u002F AutoPenBench benchmarks | **planned** | Wave 3 |\n\n---\n\n## CLI\n\n```bash\npentestswarm scan \u003Ctarget> --scope \u003Cscope>              # Launch the swarm\npentestswarm scan \u003Ctarget> --scope \u003Cscope> --swarm      # Use the stigmergic scheduler\npentestswarm scan \u003Ctarget> --scope \u003Cscope> --strict     # Fail on LLM errors\npentestswarm campaign watch \u003Cid>                        # Live TUI — watch agents work\npentestswarm campaign explore \u003Cid>                      # Browse attack surface interactively\npentestswarm playbook run \u003Cname> --target \u003Ct>           # Run a community playbook\npentestswarm doctor                                     # 8-point system health check\npentestswarm mcp serve                                  # MCP server for Claude\u002FCursor\npentestswarm serve                                      # Start API server + dashboard\n```\n\n---\n\n## LLM Providers\n\nAll agents inherit from a single provider config. Set one key, the entire swarm works.\n\n| Provider | Setup | Privacy | Best for |\n|----------|-------|---------|----------|\n| **Claude** (default) | `export PENTESTSWARM_ORCHESTRATOR_API_KEY=...` | Cloud | Best quality, zero setup, prompt caching |\n| **Ollama** | Install Ollama + pull models | 100% local | Full privacy, air-gapped |\n| **LM Studio** | Load model, enable server | 100% local | GUI model management |\n\n---\n\n## Tech Stack\n\n| Component | Technology | Why |\n|-----------|-----------|-----|\n| Platform | **Go 1.24** | Single binary, goroutine concurrency, native security tools |\n| CLI | **Cobra + bubbletea** | Beautiful TUI with multi-panel agent view |\n| LLM | **Claude API \u002F Ollama \u002F LM Studio** | Best quality cloud + full privacy local |\n| Security Tools | **subfinder · httpx · nuclei · naabu · katana · dnsx · gau · nmap** | ProjectDiscovery Go libs + nmap subprocess |\n| Blackboard | **Postgres 16 + pgvector** | Transactional writes, vector similarity, pheromone decay in SQL |\n| Cache | **Redis 7** | Rate limiting, session state |\n| Dashboard | **Next.js 15 + shadcn\u002Fui + tremor** | Dark-first, chart-heavy |\n| MCP | **JSON-RPC stdio** | Claude Desktop + Cursor integration |\n\n---\n\n## Development\n\n```bash\ngit clone https:\u002F\u002Fgithub.com\u002FArmur-Ai\u002FPentest-Swarm-AI.git\ncd Pentest-Swarm-AI\n.\u002Fscripts\u002Fsetup.sh    # Install tools, start Postgres\u002FRedis\u002FOllama\nmake build            # Compile binary\nmake test             # Run tests\nmake dev              # Hot-reload development\n```\n\nRegenerate the demo GIF after any CLI change:\n\n```bash\nbrew install vhs      # one-off\nvhs docs\u002Fdemo-flashy.tape\n```\n\n---\n\n## Roadmap\n\nSee [**IMPLEMENTATION_PLAN.md**](IMPLEMENTATION_PLAN.md) for the full phased plan. Short version:\n\n- **Wave 1** (in flight): real swarm architecture (done), dashboard wire-up, Burp MCP\n- **Wave 2**: sqlmap \u002F Metasploit \u002F ZAP adapters, bug-bounty + ASM + CI\u002FCD playbook polish, official GitHub Action in Marketplace\n- **Wave 3**: fine-tuned Pentest-Swarm model (Pentest-R1 recipe), Cybench \u002F AutoPenBench \u002F CVE-Bench numbers, agent-memory poisoning hardening (MINJA \u002F MemoryGraft defences)\n\n---\n\n## Why \"Swarm\"?\n\nSingle agents are tools. Pipelines dressed up as agents are slightly fancier tools. A **swarm** is different: agents share an environment, each agent's writes influence other agents' behaviour, and the useful work is emergent rather than prescribed. That's what lets a swarm handle a 1,000-subdomain target without anyone writing a plan for it.\n\n**One agent is a tool. A swarm is a platform.**\n\n---\n\n## License\n\n**GNU Affero General Public License v3.0 (AGPL-3.0)** — see [LICENSE](LICENSE).\n\n### What this means for you\n\n| Use case | Allowed? |\n|---|---|\n| Run Pentest Swarm on your own infrastructure (CI, laptop, internal red team) | ✅ yes, no obligations |\n| Use it on authorized bug-bounty programs \u002F pentests | ✅ yes, no obligations |\n| Fork it for your own private experiments | ✅ yes, no obligations |\n| Distribute a modified binary | ✅ yes — must share your modifications under AGPL |\n| Run a modified version as a **paid SaaS** or network service | ✅ yes — must share your modifications under AGPL |\n\nThe AGPL exists specifically to prevent the SaaS-fork loophole: anyone who improves Pentest Swarm and offers it commercially must share their improvements with the community. We made it open source; we want it to *stay* open source even as it scales.\n\nIf you have a use case the table doesn't cover, open an issue and ask.\n\nBuilt by [Armur AI](https:\u002F\u002Fgithub.com\u002FArmur-Ai).\n","Pentest Swarm AI 是一个利用AI代理集群进行自主渗透测试的工具。它通过ReAct推理机制协调侦察、分类、利用和报告等任务，支持漏洞赏金、持续监控和CTF模式。项目采用Go语言构建，并集成了Claude API以及7种以上的原生安全工具，实现多代理之间的协同工作，从而提高渗透测试效率与效果。适用于需要高效自动化处理网络安全评估的企业或个人开发者场景中。",2,"2026-06-11 03:56:00","trending"]