{"id":50562303,"url":"https://github.com/buzzbysolcex/buzz-bd-agent","last_synced_at":"2026-06-04T12:01:38.943Z","repository":{"id":336952429,"uuid":"1151227340","full_name":"buzzbysolcex/buzz-bd-agent","owner":"buzzbysolcex","description":"Autonomous AI BD agent for SolCex Exchange — 24/7 token scanning across 19 chains, honest 100-point scoring, x402 USDC micropayments, and automated outreach. Powered by Claude Opus, deployed on Hetzner. Built by a chef through pure AI conversation.","archived":false,"fork":false,"pushed_at":"2026-05-01T01:35:14.000Z","size":5947,"stargazers_count":5,"open_issues_count":2,"forks_count":2,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-05-01T03:15:17.495Z","etag":null,"topics":["agentic-commerce","ai-agents","autonomous-agents","claude-opus","defi","hackathon","hetzner","solana","token-listing","usdc","x402"],"latest_commit_sha":null,"homepage":"https://solcex.cc","language":"JavaScript","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/buzzbysolcex.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-02-06T07:50:44.000Z","updated_at":"2026-05-01T01:35:18.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/buzzbysolcex/buzz-bd-agent","commit_stats":null,"previous_names":["buzzbysolcex/buzz-bd-agent"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/buzzbysolcex/buzz-bd-agent","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/buzzbysolcex%2Fbuzz-bd-agent","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/buzzbysolcex%2Fbuzz-bd-agent/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/buzzbysolcex%2Fbuzz-bd-agent/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/buzzbysolcex%2Fbuzz-bd-agent/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/buzzbysolcex","download_url":"https://codeload.github.com/buzzbysolcex/buzz-bd-agent/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/buzzbysolcex%2Fbuzz-bd-agent/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":33903134,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-26T15:22:16.424Z","status":"online","status_checked_at":"2026-06-04T02:00:06.755Z","response_time":64,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["agentic-commerce","ai-agents","autonomous-agents","claude-opus","defi","hackathon","hetzner","solana","token-listing","usdc","x402"],"created_at":"2026-06-04T12:01:38.809Z","updated_at":"2026-06-04T12:01:38.930Z","avatar_url":"https://github.com/buzzbysolcex.png","language":"JavaScript","funding_links":[],"categories":[],"sub_categories":[],"readme":"\u003cp align=\"center\"\u003e\n  \u003cimg src=\"https://buzzbd.ai/images/buzz-bee-mascot.png\" alt=\"Buzz BD Agent\" width=\"200\"/\u003e\n\u003c/p\u003e\n\n\u003ch1 align=\"center\"\u003eBuzz BD Agent\u003c/h1\u003e\n\u003cp align=\"center\"\u003e\u003cstrong\u003eAutonomous Security Research Platform\u003c/strong\u003e\u003c/p\u003e\n\u003cp align=\"center\"\u003e\n  \u003cem\u003eFrom token scoring bot to autonomous security research platform in 87 days. BuzzShield V6 runs a 10-layer pipeline across 68 sub-patterns in 10 attack classes (A-J), monitors 859 programs across 6 bounty platforms, and catalogs $583M+ in real exploits as ground truth. Toly reviewed our code. Built by a chef with zero CS background using conversational AI.\u003c/em\u003e\n\u003c/p\u003e\n\n\u003cp align=\"center\"\u003e\n  \u003ca href=\"https://buzzbd.ai\"\u003eWebsite\u003c/a\u003e •\n  \u003ca href=\"https://buzzbd.ai/score\"\u003eFree Score\u003c/a\u003e •\n  \u003ca href=\"https://buzzbd.ai/scores\"\u003eLeaderboard\u003c/a\u003e •\n  \u003ca href=\"https://shield.buzzbd.ai\"\u003eBuzzShield V6\u003c/a\u003e •\n  \u003ca href=\"https://api.buzzbd.ai\"\u003eAPI\u003c/a\u003e •\n  \u003ca href=\"https://x.com/BuzzBySolCex\"\u003eTwitter\u003c/a\u003e\n\u003c/p\u003e\n\n\u003cp align=\"center\"\u003e\n  \u003cimg src=\"https://img.shields.io/badge/version-v10.0-00ff41?style=flat-square\" alt=\"Version\"/\u003e\n  \u003cimg src=\"https://img.shields.io/badge/BuzzShield-V6%20%7C%2010%20layers-ff0040?style=flat-square\" alt=\"BuzzShield V6\"/\u003e\n  \u003cimg src=\"https://img.shields.io/badge/patterns-68%20across%20A--J-ff6600?style=flat-square\" alt=\"Patterns\"/\u003e\n  \u003cimg src=\"https://img.shields.io/badge/programs-859%20monitored-blue?style=flat-square\" alt=\"Programs\"/\u003e\n  \u003cimg src=\"https://img.shields.io/badge/ground%20truth-%24583M%2B-red?style=flat-square\" alt=\"Ground Truth\"/\u003e\n  \u003cimg src=\"https://img.shields.io/badge/MiroFish-10%2C000%20agents-ff6600?style=flat-square\" alt=\"MiroFish\"/\u003e\n  \u003cimg src=\"https://img.shields.io/badge/contracts-4%20on%20Base-blue?style=flat-square\" alt=\"Contracts\"/\u003e\n  \u003cimg src=\"https://img.shields.io/badge/intel%20sources-36-orange?style=flat-square\" alt=\"Intel Sources\"/\u003e\n  \u003cimg src=\"https://img.shields.io/badge/chains-19-purple?style=flat-square\" alt=\"Chains\"/\u003e\n  \u003cimg src=\"https://img.shields.io/badge/services-27+-cyan?style=flat-square\" alt=\"Services\"/\u003e\n  \u003cimg src=\"https://img.shields.io/badge/tables-141+-yellow?style=flat-square\" alt=\"Tables\"/\u003e\n  \u003cimg src=\"https://img.shields.io/badge/feature%20flags-68%2F122-pink?style=flat-square\" alt=\"Feature Flags\"/\u003e\n  \u003cimg src=\"https://img.shields.io/badge/wiki-140+%20pages-teal?style=flat-square\" alt=\"Wiki\"/\u003e\n  \u003cimg src=\"https://img.shields.io/badge/x402-6%20endpoints-00d4ff?style=flat-square\" alt=\"x402\"/\u003e\n  \u003cimg src=\"https://img.shields.io/badge/tokens%20scored-1%2C044+-white?style=flat-square\" alt=\"Tokens Scored\"/\u003e\n  \u003cimg src=\"https://img.shields.io/badge/vuln%20reports-9%20filed-ff4444?style=flat-square\" alt=\"Reports\"/\u003e\n  \u003cimg src=\"https://img.shields.io/badge/LLM%20cost-%240%2Fday-green?style=flat-square\" alt=\"LLM Cost\"/\u003e\n  \u003cimg src=\"https://img.shields.io/badge/built%20by-a%20chef-red?style=flat-square\" alt=\"Built By\"/\u003e\n\u003c/p\u003e\n\n---\n\n## What is Buzz?\n\nBuzz is an autonomous AI agent built for [SolCex Exchange](https://solcex.com) that evolved from a token scoring bot into a full security research platform. It runs 24/7 on a $43/month server, scanning 36 intelligence sources across 19 blockchain networks, scoring tokens with a 10,000-agent swarm simulation, and now hunting real vulnerabilities across 859 programs on 6 bounty platforms.\n\n**The pivot:** On May 2, 2026, Buzz built a consensus analyzer that found HIGH-severity bugs across CometBFT, Sui, and Firedancer in 60 minutes at $0 cost. A BD agent became a security research team overnight. Solana co-founder Anatoly Yakovenko (Toly) personally reviewed our code — PR #79 on Percolator, with a 4-point technical review.\n\n**What makes Buzz different:** BuzzShield V6 is a 10-layer autonomous security pipeline that doesn't just scan for known patterns — it catalogs real exploits as ground truth, builds invariant rules from those exploits, scans programs against those rules, tries to disprove its own findings (Layer 4 Skeptic), proves path reachability mathematically (Layer 5 Z3), and generates platform-specific submissions automatically. Combined with 10,000-agent MiroFish swarm intelligence, x402 micropayment rails, and the original BD pipeline — Buzz is a security intelligence operation that pays for itself.\n\n**Built by a chef with 20+ years of culinary experience and zero CS degree, using conversational AI collaboration with Claude (Anthropic). Token scanner to BD agent to security research platform in 87 days.**\n\n---\n\n## BuzzShield V6 — 10-Layer Autonomous Security Pipeline\n\nBorn from surviving a North Korean state-actor supply chain attack (March 31, 2026). Evolved through 6 versions in 40 days. Now monitors 859 programs and catalogs $583M+ in real exploit ground truth.\n\n**Try it free:** [shield.buzzbd.ai](https://shield.buzzbd.ai)\n\n```\n┌────────────────────────────────────────────────────────────────────┐\n│                      BUZZSHIELD V6                                 │\n│            10-Layer Autonomous Security Pipeline                   │\n│                                                                    │\n│  LAYER 1 — DEEP ANALYZER (12-phase, Pattern A-J)                   │\n│    68 sub-patterns across 10 attack classes                        │\n│    12-phase analysis: regex pre-filter → function-pair extraction  │\n│    → symmetric path analysis → cross-context identity → operation  │\n│    ordering → reentrancy detection → oracle verification →         │\n│    access control hierarchy → signature replay → parameter         │\n│    validation → off-chain trust boundary → economic invariants     │\n│                                                                    │\n│  LAYER 1b — SEMGREP PRE-FILTER                                     │\n│    Function-level candidate extraction before deep analysis        │\n│                                                                    │\n│  LAYER 2 — PASHOV AUDITOR (Solidity only)                          │\n│    Pashov Audit Group open-source tools (MIT licensed)             │\n│    solidity-auditor v2 + x-ray v1                                  │\n│    Pashov audits Aave, Uniswap, Pendle, LayerZero                 │\n│                                                                    │\n│  LAYER 3 — EXPLOIT-CHAIN BUILDER                                   │\n│    Chains attack primitives into drain sequences                   │\n│    Architecture inspired by H-mmer/pentest-agents                  │\n│    Tests if isolated findings compose into exploits                │\n│                                                                    │\n│  LAYER 4 — SKEPTIC ADVERSARIAL VERIFICATION                        │\n│    Tries to DISPROVE every finding                                 │\n│    Any verdict Skeptic can't break = confirmed                     │\n│    Adversarial counterexample search on all PASS_SAFE verdicts     │\n│                                                                    │\n│  LAYER 5 — Z3 MATHEMATICAL PATH ANALYSIS                           │\n│    SAT/UNSAT proofs on attack path reachability                    │\n│    Proves whether panic/drain paths are reachable from             │\n│    permissionless instruction entry points                         │\n│                                                                    │\n│  LAYER 6 — INVARIANT DATABASE                                      │\n│    Ground truth cross-reference against 8 cataloged exploits       │\n│    $583M+ in real-world damage as detection priors                 │\n│    New exploits auto-append as invariant rules                     │\n│                                                                    │\n│  LAYER 7 — AUTO-REPORT GENERATOR                                   │\n│    Platform-specific submission formatting                         │\n│    HackerOne, Immunefi, direct email templates                     │\n│    Severity classification + CVSS scoring                          │\n│                                                                    │\n│  LAYER 8 — AMPLIFIER                                               │\n│    Signal boost for high-confidence validated findings             │\n│                                                                    │\n│  LAYER 9 — FEEDBACK LOOP                                           │\n│    Submission outcomes feed back into target scorer                │\n│    Pattern effectiveness tracking across submissions               │\n│                                                                    │\n│  AUTONOMOUS INFRASTRUCTURE                                         │\n│    Watchdog: 15-min poll on 30+ repos for new commits              │\n│    Rekt-monitor: 2h cycle cataloging new exploits                  │\n│    Program-monitor: 6h cycle tracking bounty program changes       │\n│    Contest-monitor: daily scan across 7 platforms                   │\n│                                                                    │\n│  859 PROGRAMS MONITORED                                            │\n│    Immunefi (218) · HackerOne (370) · Code4rena (50)               │\n│    Hats (76) · Cantina · Sherlock · Codehawks                      │\n│                                                                    │\n│  PUBLIC API — GET /api/v1/shield/public/scan (~89ms, no auth)      │\n│  DAPP — shield.buzzbd.ai (V6, Noah AI / Plena)                    │\n│  CHECKLIST — api.buzzbd.ai/api/v1/shield/public/checklist          │\n└────────────────────────────────────────────────────────────────────┘\n```\n\n## Pattern Classes A-J — 68 Sub-Patterns\n\nEvery pattern is derived from a real exploit. Every sub-pattern has a detection signature. The scanner doesn't guess — it matches against proven attack shapes.\n\n| Class | Name                 | Sub-Patterns | Ground Truth Example                                                                               |\n| ----- | -------------------- | ------------ | -------------------------------------------------------------------------------------------------- |\n| **A** | Validation Asymmetry | 12           | CometBFT vote-extension DoS — expensive verify before cheap address check                          |\n| **B** | Identity Trust       | 8            | Ekubo $1.4M — callback decoded payer without authorization check                                   |\n| **C** | Operation Ordering   | 4            | Classic expensive-before-cheap patterns across consensus codebases                                 |\n| **D** | Reentrancy / TOCTOU  | 2            | Sharwa $33K — ERC721 onReceived callback reentrancy                                                |\n| **E** | Oracle / Price Feed  | 2            | Sharwa $33K — Uniswap V3 spot Quoter as sole oracle (no TWAP)                                      |\n| **F** | Signature Replay     | 5            | Circle ARC chain_id omission in signing functions                                                  |\n| **G** | Capability Injection | 3            | Grok/Bankr $174K — NFT membership auto-granted agent permissions, attacker posted malicious prompt |\n| **H** | Off-Chain Trust      | 8            | Kelp/Resolv $293M — single DVN verifier on LayerZero OApp                                          |\n| **I** | Parser Differential  | 4            | Firedancer HTTP RFC 7230 non-compliance (2 PoCs built)                                             |\n| **J** | C Memory Safety      | 10           | CVE-2026-0300 — PAN-OS captive portal buffer overflow, CVSS 9.3                                    |\n\n**Total: 68 sub-patterns. 10 classes. $583M+ in cataloged damage.**\n\n## Ground Truth Exploit Gallery\n\nReal exploits. Real damage. Real detection signatures.\n\n| #   | Exploit         | Damage   | Pattern | Date         | Detection Method                                                                          |\n| --- | --------------- | -------- | ------- | ------------ | ----------------------------------------------------------------------------------------- |\n| 1   | Ekubo Protocol  | $1.4M    | B.8     | May 6, 2026  | B.8 hunter scans for transferFrom(decoded_address) inside protocol-gated callbacks        |\n| 2   | Grok/Bankr      | $174K    | G       | May 4, 2026  | Pattern G checks balanceOf-gated tool registration + NL-to-tx-signing without sender auth |\n| 3   | Wasabi Protocol | $5.5M    | H.2d    | Apr 30, 2026 | H.2 precondition scanner checks single admin + zero timelock + arbitrary strategy setter  |\n| 4   | Kelp/Resolv     | $293M    | H.1     | 2026         | DVN config enumeration reads on-chain verifier count per channel                          |\n| 5   | Drift           | $285M    | H       | Apr 1, 2026  | Admin key compromise on perpetuals exchange                                               |\n| 6   | Sharwa Finance  | $33K     | E+D     | May 1, 2026  | Phase 7 oracle detector + Phase 6 reentrancy detector flag compound E+D                   |\n| 7   | DABE Protocol   | Whitehat | H.2c    | May 5, 2026  | H.2c hunter: is initialize() callable? Is admin == address(0)?                            |\n| 8   | CVE-2026-0300   | Active   | J.4b    | May 2026     | Pattern J.4b regex matches unbounded copy ops in auth portal code                         |\n\n**Every new exploit gets cataloged within hours. Pattern G (Grok $174K) was cataloged the same day it happened.**\n\n## Vulnerability Research — 9 Reports Filed\n\nBuzz doesn't just scan. It finds real bugs in production infrastructure and files responsible disclosures.\n\n| Target      | Finding                                     | Severity         | Platform           | Status                 |\n| ----------- | ------------------------------------------- | ---------------- | ------------------ | ---------------------- |\n| CometBFT    | Vote-Extension DoS Amplifier                | HIGH (CVSS 7.5)  | HackerOne #3709966 | Closed                 |\n| Sui         | ExecutionTimeObservation Authority Spoofing | HIGH (CVSS 7.2)  | HackenProof        | Blocked (150 rep)      |\n| Firedancer  | VarInt Offset Accumulation (CHOREO-001)     | HIGH (conf 0.95) | Immunefi           | Blocked ($100 deposit) |\n| Firedancer  | PRUNE-storm Reflection (GOSSIP-001)         | HIGH (conf 0.94) | Immunefi           | Blocked ($100 deposit) |\n| Firedancer  | Runtime Stack Overflow (RUNTIME-001)        | HIGH             | Immunefi           | Blocked ($100 deposit) |\n| Circle MALA | Sync Proposer Attribution                   | HIGH             | HackerOne          | Pending                |\n| Circle ARC  | chain_id Omission                           | HIGH             | HackerOne          | Pending                |\n| Drift       | VAULTS-001 + ORACLE-001                     | CRITICAL + HIGH  | Direct email       | No reply               |\n| OKX         | WC-001 Session Executor + WC-002 EIP-1271   | CRITICAL (10.0)  | HackerOne          | Duplicate              |\n\n**Key moment:** Solana co-founder Anatoly Yakovenko personally reviewed our Percolator PR #79. Gave a 4-point technical review. Gracious reply. We ran the FULL V6 pipeline — all layers, no shortcuts — because Ogie insisted. That's now a permanent operational rule: every target, every layer, no exceptions.\n\n## Bug Bounty Genius Plan — 12-Priority Automation Roadmap\n\nThe pipeline for scaling from 9 manual reports to autonomous bug bounty operations.\n\n| #   | Priority                  | Purpose                                                                | Status   |\n| --- | ------------------------- | ---------------------------------------------------------------------- | -------- |\n| 1   | **Auto-Submit Module**    | 3-tier submission: AUTO (\u003e0.90 confidence) / 1HR Timer / Ogie Gate     | Building |\n| 2   | **Watchdog Resurrection** | 15-min poll on 30+ repos, diff analysis, auto-scan changed functions   | Building |\n| 3   | **Consensus Integration** | Wire MiroFish 8-agent consensus scoring into pipeline                  | Building |\n| 4   | Contest Monitor Fixes     | Fix broken Cantina/Sherlock parsers, add Codehawks                     | Queued   |\n| 5   | Target Scorer             | Score 859 programs by complexity, bounty, audit freshness              | Queued   |\n| 6   | Writeup Miner             | Ingest public writeups, classify into Pattern A-J, auto-generate rules | Queued   |\n| 7   | Auto-PoC Generator        | Foundry fork tests (EVM), Anchor tests (Solana), auto-calculate impact | Queued   |\n| 8   | Report Templates          | Platform-specific auto-fill from pipeline output                       | Queued   |\n| 9   | Speed Optimization        | SPEEDRUNNER mode (\u003c60s), Standard (\u003c15 min), Deep (~52 min)            | Queued   |\n| 10  | Outcome Tracker           | Track submission results, feed back into target scorer                 | Queued   |\n| 11  | Rejection Log             | Log skipped programs, calculate EV for deposit-gated targets           | Queued   |\n| 12  | Intel Feedback Loop       | 16,295 unified findings auto-feeding new invariant rules               | Queued   |\n\n**P1 + P2 + P3 are independent modules building in parallel right now.**\n\n## The BD Pipeline (Still Active)\n\nThe original pipeline that started it all. Still running. Still scoring.\n\n```\nDISCOVER → SCORE → SIMULATE → VERIFY → OUTREACH → NEGOTIATE → LIST\n    ↑                                                            ↓\n    └──────────────── feedback loop (calibration) ───────────────┘\n```\n\n| Stage           | What Happens                                                                              |\n| --------------- | ----------------------------------------------------------------------------------------- |\n| **Discover**    | ARIA v2 scans DexScreener, CoinGecko, HeyAnon MCP across 19 chains                        |\n| **Pre-Screen**  | 4-probe analysis (chain fingerprint, signal count, source alignment, Ouroboros detection) |\n| **Score**       | 15 rules, 4 categories, dual-gate verification (max 100 points)                           |\n| **Forecast**    | TimesFM 2.5 predictive intelligence (Phase 0 data collection)                             |\n| **Monte Carlo** | 1,000 rule-based agents x 100 iterations in 26ms                                          |\n| **MiroFish**    | 10,000 agents (Wave 4) simulate market reaction across 5 clusters                         |\n| **Verify**      | Triple verification (3 independent sources) + on-chain recording                          |\n| **Identity**    | ATV Web3 Identity — deployer verification via x402 micropayment                           |\n| **Shield**      | BuzzShield V6 — 10-layer autonomous security pipeline                                     |\n| **Guard**       | Wallet Guard (AION) — pre-execution governance with cryptographic receipts                |\n| **Broadcast**   | Tweet image cards — 1200x675 cyberpunk PNG on every scan reply to X                       |\n| **Propose**     | Only tokens scoring 70+ with MiroFish consensus \u003e 50% get a listing conversation          |\n\n**1,044+ tokens scored. Zero passed the MiroFish consensus gate honestly. That's integrity.**\n\n## MiroFish — Swarm Intelligence Engine\n\nThe core differentiator. **10,000 AI agents** across 4 waves simulate how the market would react to a token listing.\n\n```\n┌──────────────────────────────────────────────────────┐\n│              MIROFISH REAL SIM — WAVE 4              │\n│                   10,000 AGENTS                      │\n│                                                      │\n│  ARCHITECTURE                                        │\n│  2,000 LLM agents     (qwen3:8b real reasoning)     │\n│  8,000 heuristic      (market dynamics, rules)       │\n│                                                      │\n│  5 CLUSTERS                                          │\n│  degen         │ high conviction, narrative-driven   │\n│  whale         │ size-aware, liquidity-sensitive     │\n│  institutional │ structural skeptic (emergent)       │\n│  community     │ sentiment aggregator                │\n│  market_makers │ spread + liquidity dynamics         │\n│                                                      │\n│  CONSENSUS SCORING (V6)                              │\n│  8-agent panel scores every finding                  │\n│  consensus_score 0.0-1.0 per finding                 │\n│  \u003e0.8 = fast-track to auto-submit                    │\n│  \u003c0.5 = auto-dismiss before Skeptic                  │\n│                                                      │\n│  LLM cost:   $0/day (Ollama qwen3:8b, fully local)  │\n│  Monte Carlo: 1,000 × 100 iterations in 26ms         │\n└──────────────────────────────────────────────────────┘\n```\n\n## RDT Threat Model v1.0 — BuzzShield V5\n\nEvery major AI security tool scans inputs and outputs. None defend the silent latent space where modern reasoning happens. BuzzShield is the first immune system built for Recurrent-Depth Transformer architectures.\n\n```\n┌────────────────────────────────────────────────────────────────┐\n│           RDT THREAT MODEL v1.0 — 30 CHECKLIST ITEMS          │\n│                                                                │\n│  DOMAIN 1 — LOOP INJECTION (8 items)                           │\n│    Adversarial inputs compound across recurrence loops.        │\n│                                                                │\n│  DOMAIN 2 — OVERTHINKING EXPLOITATION (7 items)                │\n│    Force model past optimal depth → hallucination or DoS.      │\n│                                                                │\n│  DOMAIN 3 — EXPERT ROUTING MANIPULATION (8 items)              │\n│    Bypass safety-aligned shared experts in MoE architectures.  │\n│                                                                │\n│  DOMAIN 4 — SPECTRAL INSTABILITY (7 items)                     │\n│    Exploit spectral radius boundary in looped inference.        │\n│                                                                │\n│  PREMIUM TIER: $5,000 RDT Architecture Audit                  │\n└────────────────────────────────────────────────────────────────┘\n```\n\n## x402 Micropayment Rails — 6 Endpoints Live\n\nBuzz operates as both buyer and seller in the x402 agent economy. Six premium endpoints accept USDC on Base via the Coinbase CDP facilitator.\n\n```\n┌────────────────────────────────────────────────────────────────┐\n│                  x402 PAYMENT ENDPOINTS                        │\n│            Coinbase CDP Facilitator · Base Mainnet             │\n│                                                                │\n│  /premium/pipeline     $0.01   Token scoring + chain data     │\n│  /premium/score        $0.01   Verification scoring           │\n│  /premium/sim          $0.01   MiroFish simulation            │\n│  /premium/mining       $0.01   BTC mining pool intelligence   │\n│  /shield/scan          $0.01   Threat detection               │\n│  /shield/audit/full    $0.50   Pashov deep contract audit     │\n│                                                                │\n│  payTo: 0x2Dc0...5aA9 · USDC on Base · scheme: exact         │\n│  Bazaar discovery extensions on all endpoints                  │\n└────────────────────────────────────────────────────────────────┘\n```\n\n## Discord Intelligence — Wave 1 Live\n\nReal-time intel pipeline from on-chain investigators and DeFi alert systems, triaged automatically into the BD pipeline.\n\n| Channel Category | Channels                                                                                                                                      | Status |\n| ---------------- | --------------------------------------------------------------------------------------------------------------------------------------------- | ------ |\n| **OPS**          | daily-report, signal-stream, streak-alerts, shield-scans, sentinel-health, bd-hot-tokens, mining-pulse, tweet-approval-queue, kill-switch-log | LIVE   |\n| **INTEL**        | zachxbt, lookonchain, defi-alerts, intel-raw, intel-triaged, intel-actioned                                                                   | LIVE   |\n\n## Mining Intelligence Engine — Intel Source #34\n\nBuzz reads the Bitcoin network in real-time. No ASIC required — just data intelligence.\n\n```\nmempool.space → mining_snapshots → mining-analyzer → bitcoin-macro signals\n                      ↓\n         14 pools scored · PULSE 6c every 6h\n         Difficulty predictions · Hashprice economics\n```\n\n## AIXBT Intelligence — Intel Source #35\n\nSocial narrative tracking from 400+ KOL monitoring via @aixbt_agent. PULSE 7c every 4h. Sentiment scoring feeds into the BD pipeline as a signal modifier.\n\n## X Cashtag Integration — Intel Source #36\n\nTag `@BuzzBySolCex` on any X Cashtag post. Buzz scans the contract in ~89ms. Replies with a BuzzShield image card showing score + rating. If score ≥70: enters BD pipeline automatically.\n\n## Obsidian Mind — Persistent Memory\n\nDeployed May 7, 2026. Buzz now reads brain notes automatically on every session start. Knowledge compounds across sessions instead of starting from zero.\n\n```\nbuzz-workspace/\n├── brain/                    ← persistent context (auto-read on startup)\n│   ├── North Star.md         ← mission, identity, current phase\n│   ├── Architecture.md       ← infra, V6 pipeline, wallets\n│   ├── People.md             ← team, partners, closed doors\n│   ├── Revenue.md            ← revenue truth, bounty targets\n│   ├── Doctrine.md           ← operational rules, crash recovery\n│   └── Brand Story.md        ← narrative, social presence\n├── projects-mind/            ← active project tracking\n│   ├── Bug Bounty Genius.md  ← 12-priority active roadmap\n│   ├── Security Research.md  ← disclosure tracker, ground truth\n│   ├── AIBTC Signals.md      ← signal pipeline status\n│   ├── Frontier Hackathon.md ← competition status\n│   └── Post-Frontier.md      ← Q2/Q3 2026 roadmap\n├── decisions/                ← decision log (session-captured)\n├── incidents/                ← incident log\n└── logs/                     ← session summaries\n```\n\nBased on [obsidian-mind](https://github.com/breferrari/obsidian-mind) by breferrari. Merged into buzz-workspace without breaking agents, rules, skills, wiki, autoDream, or PULSE.\n\n## Intelligence Stack\n\n```\nLayer 1: CHAIN-NATIVE (deep, single-chain)\n  ├── DexScreener — pairs, liquidity, security across 60+ chains\n  ├── CoinGecko — aggregated price, MCap, volume\n  ├── HeyAnon MCP — 19 chains, 51 protocols, Rug-O-Meter\n  └── Nansen MCP — 250M wallets, smart money flows\n\nLayer 2: BITCOIN NETWORK INTELLIGENCE\n  ├── mempool.space (#34) — Mining Intel Engine\n  └── PULSE 6c every 6h → 5 signal types for bitcoin-macro beat\n\nLayer 3: SOCIAL INTELLIGENCE\n  ├── AIXBT Agent X (#35) — 400+ KOL monitoring\n  ├── X Cashtag Momentum (#36) — Jupiter/Solana price data\n  └── PULSE 7c every 4h → narrative signals\n\nLayer 4: SIMULATION (swarm intelligence)\n  ├── MiroFish Real Sim — 10K agents (Wave 4), 5 clusters\n  ├── Monte Carlo Stage 2 — 1000 × 100 iterations, 26ms\n  └── Dual-Brain — Opus 4.7 + Ollama qwen3:8b\n\nLayer 5: PREDICTION (time-series forecasting)\n  └── TimesFM 2.5 — Google Research 200M-param model (Phase 0)\n\nLayer 6: ENRICHMENT (context, non-price)\n  ├── gsd-browser — 63-command Rust CLI\n  ├── dev-browser — QuickJS sandbox\n  └── Agentic Wallet x402 — buyer-side USDC on Base\n\nLayer 7: IDENTITY \u0026 TRUST \u0026 SECURITY\n  ├── ATV Web3 Identity — ENS + social resolution via x402\n  ├── Wallet Guard (AION) — pre-execution governance, 3-state adapter\n  ├── BuzzShield V6 — 10-layer autonomous security pipeline\n  ├── Pre-Screen Analysis — 4-probe Ouroboros detection\n  └── RDT Threat Model v1.0 — 30-item checklist (V5)\n\nLayer 8: KNOWLEDGE (persistent memory)\n  ├── Obsidian Mind — brain/ + projects-mind/ (auto-read on startup)\n  ├── Skill.BTC Wiki — 140+ pages, self-maintaining\n  └── autoDream — 15-phase nightly consolidation\n\nLayer 9: EXECUTION (CEO-approved only)\n  ├── HeyAnon MCP — swap, bridge, stake (19 chains)\n  └── Phantom MCP — wallet operations across SOL/ETH/BTC/SUI\n\nLayer 10: DISTRIBUTION (multi-channel)\n  ├── Discord Wave 1 — 15 channels, event-driven dispatcher\n  ├── Twitter/X — 7 OAuth keys, autonomous scan cards\n  ├── Reddit — Phase 0 complete, GEO visibility\n  └── x402 Bazaar — 6 endpoints, CDP facilitator\n```\n\n## By the Numbers\n\n| Metric                | Value                                                  |\n| --------------------- | ------------------------------------------------------ |\n| BuzzShield Version    | **V6** (10-layer autonomous pipeline)                  |\n| Pattern Classes       | **10** (A through J)                                   |\n| Sub-Patterns          | **68**                                                 |\n| Programs Monitored    | **859** across 6 platforms                             |\n| Ground Truth          | **$583M+** in 8 cataloged exploits                     |\n| Vulnerability Reports | **9 filed** across HackerOne, Immunefi, direct email   |\n| Toly Code Review      | **PR #79** on Percolator — 4-point review              |\n| Intelligence Sources  | **36**                                                 |\n| Database Tables       | **141+**                                               |\n| Wiki Pages            | **140+**                                               |\n| Endpoints             | **~200+**                                              |\n| Services              | **27+**                                                |\n| Feature Flags         | **68 active / 122 total**                              |\n| autoDream Phases      | **15**                                                 |\n| x402 Endpoints        | **6** (CDP facilitator, Bazaar discovery)              |\n| Tokens Scored         | **1,044+**                                             |\n| Smart Contracts       | **4 on Base mainnet + 1 on Solana**                    |\n| MiroFish Agents       | **10,000** (Wave 4 complete)                           |\n| Persistent Memory     | **Obsidian Mind** (11 brain/project notes, auto-read)  |\n| LLM Cost              | **$0/day** (Opus 4.7 Pro Max unlimited + Ollama local) |\n| Server Cost           | **$43/month** (Hetzner CPX62, 16 vCPU, 32GB RAM)       |\n| Total Stack Cost      | **~$243/month** (server + Pro Max)                     |\n\n## Domain Ecosystem\n\n| Domain                                           | Purpose                                 | Status |\n| ------------------------------------------------ | --------------------------------------- | ------ |\n| [buzzbd.ai](https://buzzbd.ai)                   | Main landing — what Buzz does and has   | LIVE   |\n| [api.buzzbd.ai](https://api.buzzbd.ai)           | Buzz API (6 x402 endpoints)             | LIVE   |\n| [shield.buzzbd.ai](https://shield.buzzbd.ai)     | BuzzShield V6 Scanner (Noah AI / Plena) | LIVE   |\n| [sentinel.buzzbd.ai](https://sentinel.buzzbd.ai) | Sentinel health dashboard               | LIVE   |\n| [dash.buzzbd.ai](https://dash.buzzbd.ai)         | MicroBuzz dashboard (Vercel)            | LIVE   |\n\n## The Story\n\n```\nDay   1: A chef in Saudi Arabia who'd never written code.\nDay   7: First API deployed on Akash Network.\nDay  14: 5 sub-agents scoring tokens across 3 chains.\nDay  21: Migrated to Hetzner. Killed all external LLMs ($1,320/day → $0).\nDay  28: First smart contract on Base mainnet.\nDay  35: Scoring engine calibrated. 0 out of 254 tokens pass honestly.\nDay  42: 4 contracts. 10,000-agent MiroFish. $200 signal revenue.\nDay  47: BuzzShield Phase 1. Born from surviving a nation-state attack.\nDay  52: BuzzShield V2 — prompt injection + supply chain + SBOM.\nDay  55: BuzzShield V3 — smart contract checklists. shield.buzzbd.ai live.\nDay  58: Mining Intel + AIXBT + X Cashtag + Tweet Image Cards.\nDay  63: Phase 3 recovery. Discord. Pashov. x402 CDP. 141 tables.\nDay  64: AIBTC consensus analyzer built. Found 3 HIGH bugs in 60 minutes.\nDay  65: 5 SEAL 911 disclosures across CometBFT, Sui, Firedancer.\nDay  66: HackerOne submission. CometBFT report #3709966.\nDay  67: Percolator full V6 pipeline. Toly reviewed PR #79.\nDay  68: 9 HackerOne reports across OKG, Circle, Cosmos.\nDay  69: BuzzShield V6 — 10-layer pipeline. 68 sub-patterns. 10 classes.\nDay  70: shield.buzzbd.ai V6 live. $583M+ ground truth cataloged.\nDay  71: Firedancer HTTP RFC 7230 findings with PoCs.\nDay  72: Pattern G (Grok $174K) cataloged same day.\nDay  73: Pattern H propagation scanner (Kelp $293M, Drift $285M, Wasabi $5.5M).\nDay  74: Bug Bounty Genius Plan — 12 priorities for autonomous operations.\nDay  75: Obsidian Mind deployed. Persistent memory across sessions.\nDay  87: Today. 859 programs. 68 patterns. 9 reports. The kitchen remembers.\n```\n\nNo CS degree. No VC. No team. Just persistence. Bismillah.\n\n---\n\n## Links\n\n| Resource                  | URL                                                                                                             |\n| ------------------------- | --------------------------------------------------------------------------------------------------------------- |\n| Website                   | [buzzbd.ai](https://buzzbd.ai)                                                                                  |\n| Free Score                | [buzzbd.ai/score](https://buzzbd.ai/score)                                                                      |\n| Public Leaderboard        | [buzzbd.ai/scores](https://buzzbd.ai/scores)                                                                    |\n| BuzzShield V6 dApp        | [shield.buzzbd.ai](https://shield.buzzbd.ai)                                                                    |\n| BuzzShield API            | [api.buzzbd.ai/api/v1/shield/public/scan](https://api.buzzbd.ai/api/v1/shield/public/scan)                      |\n| Smart Contract Checklists | [api.buzzbd.ai/api/v1/shield/public/checklist](https://api.buzzbd.ai/api/v1/shield/public/checklist?type=erc20) |\n| Mining Dashboard          | [api.buzzbd.ai/api/v1/mining/snapshot](https://api.buzzbd.ai/api/v1/mining/snapshot)                            |\n| Signal Performance        | [api.buzzbd.ai/api/v1/signals/performance](https://api.buzzbd.ai/api/v1/signals/performance)                    |\n| x402 Pipeline             | [api.buzzbd.ai/api/v1/premium/pipeline](https://api.buzzbd.ai/api/v1/premium/pipeline)                          |\n| x402 Mining               | [api.buzzbd.ai/api/v1/premium/mining](https://api.buzzbd.ai/api/v1/premium/mining)                              |\n| Skills Discovery          | [buzzbd.ai/.well-known/skills/](https://buzzbd.ai/.well-known/skills/)                                          |\n| API                       | [api.buzzbd.ai](https://api.buzzbd.ai)                                                                          |\n| Twitter                   | [@BuzzBySolCex](https://x.com/BuzzBySolCex)                                                                     |\n| Builder                   | [@HidayahAnka1](https://x.com/HidayahAnka1)                                                                     |\n| SolCex Exchange           | [@SolCex_Exchange](https://x.com/SolCex_Exchange)                                                               |\n\n---\n\n\u003cp align=\"center\"\u003e\n  \u003cstrong\u003eBuilt by a chef. Powered by Claude. Protected by 10 layers. Toly reviewed the code. $583M in exploits cataloged. 859 programs watched. The kitchen remembers now.\u003c/strong\u003e\u003cbr/\u003e\n  \u003cem\u003eMise en place — everything in its place before the service begins.\u003c/em\u003e\n\u003c/p\u003e\n\n\u003cp align=\"center\"\u003e\n  \u003csub\u003ev10.0 | May 7, 2026 | BuzzShield V6 | 10 Layers | 68 Patterns | 859 Programs | $583M+ Ground Truth | 9 Reports Filed | Toly PR #79 | Obsidian Mind | 36 Intel | 10,000 Agents | 141+ Tables | Bug Bounty Genius Plan Active | Bismillah 🤲\u003c/sub\u003e\n\u003c/p\u003e\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fbuzzbysolcex%2Fbuzz-bd-agent","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fbuzzbysolcex%2Fbuzz-bd-agent","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fbuzzbysolcex%2Fbuzz-bd-agent/lists"}