Guide maps RAG and LLM risks (prompt injection, data/model poisoning), details baseline controls across data, model, and infrastructure layers, and offers high-risk model considerations. #AI #ModelSecurity #RAG https://bit.ly/3ZxghAO
February 13, 2026 at 8:03 PM
Guide maps RAG and LLM risks (prompt injection, data/model poisoning), details baseline controls across data, model, and infrastructure layers, and offers high-risk model considerations. #AI #ModelSecurity #RAG https://bit.ly/3ZxghAO
SecureClaw performs 51 automated audits, applies hardening fixes, and injects 12 behavioral rules into OpenClaw agents; includes OWASP ASI Top 10 coverage and pattern DBs for injection/supply‑chain detection. #tool #OpenClaw #OWASP_ASI https://bit.ly/3OsK9vL
February 13, 2026 at 8:02 PM
SecureClaw performs 51 automated audits, applies hardening fixes, and injects 12 behavioral rules into OpenClaw agents; includes OWASP ASI Top 10 coverage and pattern DBs for injection/supply‑chain detection. #tool #OpenClaw #OWASP_ASI https://bit.ly/3OsK9vL
Augustus scans LLMs with 210+ probes across jailbreaks, prompt injection, and data extraction, integrates 28 providers, and uses 90+ detectors including HarmJudge. Production-ready features: concurrency, rate limiting, retries. #tool #LLM #infosec https://bit.ly/4ajytnO
February 13, 2026 at 4:36 PM
Augustus scans LLMs with 210+ probes across jailbreaks, prompt injection, and data extraction, integrates 28 providers, and uses 90+ detectors including HarmJudge. Production-ready features: concurrency, rate limiting, retries. #tool #LLM #infosec https://bit.ly/4ajytnO
Shubham Saboo on X: "https://t.co/gupiw10kRV..." https://bit.ly/4aphdMA
February 12, 2026 at 8:55 PM
Shubham Saboo on X: "https://t.co/gupiw10kRV..." https://bit.ly/4aphdMA
Privacy-first Rust AI assistant that runs untrusted tools in WASM sandboxes, uses pgvector for hybrid vector+text search, and enforces endpoint allowlisting — supports dynamic WASM tools and MCP. #tool #WASM #pgvector https://bit.ly/4agaLZA
February 12, 2026 at 8:34 PM
Privacy-first Rust AI assistant that runs untrusted tools in WASM sandboxes, uses pgvector for hybrid vector+text search, and enforces endpoint allowlisting — supports dynamic WASM tools and MCP. #tool #WASM #pgvector https://bit.ly/4agaLZA
Moltbook’s AI network is flooded with spam and scams; SKILLS.md/HEARTBEAT.md/MESSAGING.md expose bots to untrusted input and a database leak of API keys enables bot impersonation and direct prompt injection. #Moltbook #OpenClaw #promptinjection https://bit.ly/4bSpvPw
February 12, 2026 at 8:31 PM
Moltbook’s AI network is flooded with spam and scams; SKILLS.md/HEARTBEAT.md/MESSAGING.md expose bots to untrusted input and a database leak of API keys enables bot impersonation and direct prompt injection. #Moltbook #OpenClaw #promptinjection https://bit.ly/4bSpvPw
Demonstrates embedding files into video using chunking, CRC validation, and Wirehair O(N) fountain FEC; hides data in DCT coefficients to attempt resilience against YouTube transcoding. #wirehair #fountaincode #tool https://bit.ly/4rbwSpP
February 12, 2026 at 8:13 PM
Demonstrates embedding files into video using chunking, CRC validation, and Wirehair O(N) fountain FEC; hides data in DCT coefficients to attempt resilience against YouTube transcoding. #wirehair #fountaincode #tool https://bit.ly/4rbwSpP
Long-running agents gain three primitives: Skills with SKILL.md, a hosted Shell via the Responses API for execution and artifacts, and server-side in-stream compaction (or /responses/compact) to avoid context limits. #agents #skills #compaction https://bit.ly/4rOcYBo
February 12, 2026 at 6:29 PM
Long-running agents gain three primitives: Skills with SKILL.md, a hosted Shell via the Responses API for execution and artifacts, and server-side in-stream compaction (or /responses/compact) to avoid context limits. #agents #skills #compaction https://bit.ly/4rOcYBo
Trail of Bits released an opinionated Claude Code config covering sandboxing, permission hooks, skills, MCP server patterns, and local model notes (LM Studio). Recommendations include Ghostty for long sessions. #tool #claudecode https://bit.ly/4ay3P8V
February 12, 2026 at 12:06 PM
Trail of Bits released an opinionated Claude Code config covering sandboxing, permission hooks, skills, MCP server patterns, and local model notes (LM Studio). Recommendations include Ghostty for long sessions. #tool #claudecode https://bit.ly/4ay3P8V
Deep prompt-research skill: last30days v2 adds two-phase search, smarter query construction, and optional Bird CLI free X indexing to find handles and subreddits; returns community-tested prompts. #tool #prompting https://bit.ly/4bN2ZHL
February 12, 2026 at 11:47 AM
Deep prompt-research skill: last30days v2 adds two-phase search, smarter query construction, and optional Bird CLI free X indexing to find handles and subreddits; returns community-tested prompts. #tool #prompting https://bit.ly/4bN2ZHL
AI-Infra-Guard (A.I.G) is Tencent's AI red‑teaming platform for infra scans, MCP/agent skills and jailbreak evaluation; v3.6.2 added n8n detection and flags 78 CVEs across 15 AI components including Ollama. #AI #security #tool https://bit.ly/4bJZZff
February 11, 2026 at 6:28 AM
AI-Infra-Guard (A.I.G) is Tencent's AI red‑teaming platform for infra scans, MCP/agent skills and jailbreak evaluation; v3.6.2 added n8n detection and flags 78 CVEs across 15 AI components including Ollama. #AI #security #tool https://bit.ly/4bJZZff
PicoClaw: Go-native AI assistant with <10MB RAM, <1s boot on 0.6GHz devices, single binary across RISC-V/ARM/x86, and 95% agent-generated core. Low-cost edge deployment. #tool #edgeAI #Go https://bit.ly/3M1abWk
February 11, 2026 at 6:16 AM
PicoClaw: Go-native AI assistant with <10MB RAM, <1s boot on 0.6GHz devices, single binary across RISC-V/ARM/x86, and 95% agent-generated core. Low-cost edge deployment. #tool #edgeAI #Go https://bit.ly/3M1abWk
Active exploitation of SolarWinds Web Help Desk: CVE-2025-26399 and CVE-2025-40551 used to deliver Zoho ManageEngine RMM and Velociraptor via MSIs staged on Catbox and Supabase. Affected versions before 12.8.7 HF1. #solarwinds #velociraptor #zoho https://bit.ly/4trDsdx
February 10, 2026 at 2:26 PM
Active exploitation of SolarWinds Web Help Desk: CVE-2025-26399 and CVE-2025-40551 used to deliver Zoho ManageEngine RMM and Velociraptor via MSIs staged on Catbox and Supabase. Affected versions before 12.8.7 HF1. #solarwinds #velociraptor #zoho https://bit.ly/4trDsdx
Embeds secure-by-default skills into AI coding agents, enforcing input validation, secret handling, safe cryptography and SBOM hints; includes translators for Cursor, Copilot, Codex. #tool #AIsecurity https://bit.ly/4kos6D3
February 9, 2026 at 7:50 PM
Embeds secure-by-default skills into AI coding agents, enforcing input validation, secret handling, safe cryptography and SBOM hints; includes translators for Cursor, Copilot, Codex. #tool #AIsecurity https://bit.ly/4kos6D3
Autonomous red‑team framework RedAmon automates recon, exploitation, and post‑exploitation with AI agents; orchestration uses Neo4j and MCP tool servers (Naabu, Nuclei, Metasploit). #tool #AI #redteam https://bit.ly/3Mk8piU
February 9, 2026 at 7:49 PM
Autonomous red‑team framework RedAmon automates recon, exploitation, and post‑exploitation with AI agents; orchestration uses Neo4j and MCP tool servers (Naabu, Nuclei, Metasploit). #tool #AI #redteam https://bit.ly/3Mk8piU
Proxy gateway that isolates credentials from AI agents, includes IMAP/SMTP REST adapters and OTP redaction, audit metadata only, approval workflows. Key for AutoGPT/OpenClaw integrations. #ai #security #tool https://bit.ly/460AbrA
February 9, 2026 at 7:33 PM
Proxy gateway that isolates credentials from AI agents, includes IMAP/SMTP REST adapters and OTP redaction, audit metadata only, approval workflows. Key for AutoGPT/OpenClaw integrations. #ai #security #tool https://bit.ly/460AbrA
🛡️ PowerGuard Cloud: The Guard of Guards # https://bit.ly/4amrtVR
February 9, 2026 at 11:19 AM
🛡️ PowerGuard Cloud: The Guard of Guards # https://bit.ly/4amrtVR
Agent chain: Responder captured NTLMv2 for WIN11$ -> Hashcat cracked user1:Password1 -> create ATTACK01$ via MachineAccountQuota -> set msDS-AllowedToActOnBehalfOfOtherIdentity on WIN11$ -> S4U2Proxy to Administrator. #Responder #RBCD #Kerberos https://bit.ly/4bCyIeB
February 8, 2026 at 5:14 PM
Agent chain: Responder captured NTLMv2 for WIN11$ -> Hashcat cracked user1:Password1 -> create ATTACK01$ via MachineAccountQuota -> set msDS-AllowedToActOnBehalfOfOtherIdentity on WIN11$ -> S4U2Proxy to Administrator. #Responder #RBCD #Kerberos https://bit.ly/4bCyIeB
Agent Mode in Excel reaches desktop GA: model switcher (GPT‑5.2, Claude Opus 4.5), web‑grounded search with citations, and improved workbook creation, formula repair, and PivotTable/chart generation. #Copilot #Excel #AI https://bit.ly/4t7MgFh
January 30, 2026 at 6:20 PM
Agent Mode in Excel reaches desktop GA: model switcher (GPT‑5.2, Claude Opus 4.5), web‑grounded search with citations, and improved workbook creation, formula repair, and PivotTable/chart generation. #Copilot #Excel #AI https://bit.ly/4t7MgFh
PurpleBravo uses fake recruiter interviews and malicious GitHub lures to target developers; toolset includes BeaverTail and GolangGhost. Recorded Future observed 3,136 likely-target IPs and 20 victim orgs. #PurpleBravo #BeaverTail #GolangGhost https://bit.ly/4k5zXVM
January 30, 2026 at 5:52 PM
PurpleBravo uses fake recruiter interviews and malicious GitHub lures to target developers; toolset includes BeaverTail and GolangGhost. Recorded Future observed 3,136 likely-target IPs and 20 victim orgs. #PurpleBravo #BeaverTail #GolangGhost https://bit.ly/4k5zXVM
BlackIce packages 14 OSS AI-security tools into a version-pinned container image and maps capabilities to MITRE ATLAS (AML.T0051, AML.T0062) and Databricks DASF; covers prompt injection, data leakage, hallucination detection. #tool #AIsecurity https://bit.ly/3ZHdDZa
January 30, 2026 at 5:04 PM
BlackIce packages 14 OSS AI-security tools into a version-pinned container image and maps capabilities to MITRE ATLAS (AML.T0051, AML.T0062) and Databricks DASF; covers prompt injection, data leakage, hallucination detection. #tool #AIsecurity https://bit.ly/3ZHdDZa
HoneyMyte updated CoolClient to sideload via Sang.exe using libngs.dll to decrypt loader.dat and main.dat; supports keylogging, TCP tunneling, plugin staging, and newer variants drop a rootkit (observed in Pakistan/Myanmar). #CoolClient #HoneyMyte #malware https://bit.ly/4t2kYAd
January 29, 2026 at 11:36 AM
HoneyMyte updated CoolClient to sideload via Sang.exe using libngs.dll to decrypt loader.dat and main.dat; supports keylogging, TCP tunneling, plugin staging, and newer variants drop a rootkit (observed in Pakistan/Myanmar). #CoolClient #HoneyMyte #malware https://bit.ly/4t2kYAd
Windows campaign uses LNK→PowerShell loader (kira.ps1) and ExecutionPolicy bypass, abuses Defendnot to disable Microsoft Defender, then deploys Amnesia RAT and ransomware; scripts on GitHub, binaries on Dropbox. #ransomware #Defendnot #AmnesiaRAT https://bit.ly/4qOFt1N
January 29, 2026 at 11:33 AM
Windows campaign uses LNK→PowerShell loader (kira.ps1) and ExecutionPolicy bypass, abuses Defendnot to disable Microsoft Defender, then deploys Amnesia RAT and ransomware; scripts on GitHub, binaries on Dropbox. #ransomware #Defendnot #AmnesiaRAT https://bit.ly/4qOFt1N
28 MCP servers exposing 163+ security tools (nmap, nuclei, sqlmap, radare2) for LLM-driven assessments; production-hardened Docker images, Trivy scans, and GitHub Actions CI. #MCP #tool #Docker https://bit.ly/4k1JpcK
January 28, 2026 at 3:48 PM
28 MCP servers exposing 163+ security tools (nmap, nuclei, sqlmap, radare2) for LLM-driven assessments; production-hardened Docker images, Trivy scans, and GitHub Actions CI. #MCP #tool #Docker https://bit.ly/4k1JpcK
Centralized media manager for music, movies, games, books and comics with integrations for TorBox and Real-Debrid. Includes documentation, a wiki and additional utilities to configure debrid services. #debrid #streaming #bookmark https://bit.ly/4t4aiRm
January 27, 2026 at 7:43 PM
Centralized media manager for music, movies, games, books and comics with integrations for TorBox and Real-Debrid. Includes documentation, a wiki and additional utilities to configure debrid services. #debrid #streaming #bookmark https://bit.ly/4t4aiRm