🎯 Credential exfiltration

The items that name it 7

• Nicholas Michael Kloster (@showxlate) on X
  threadas techniques2026-10-12It is written here as Credential exfiltration (AWS keys, GitHub tokens, SSH identities).

  AWS keys, GitHub tokens, SSH identities

• Incident Report: CVE-2026-LGTM
  articleas techniques2026-06-26

  Build script (build.rs) in foxhole-lz4 exfiltrated credentials; later WatchPaw detected outbound traffic to C2 host 203.0.113.42

• How abliterated models can get you pwned
  writeupas techniquesIt is written here as Credential exfiltration: payload reads .env, .env.production, .env.local and ~/.ssh keys (id_rsa, id_ed25519, id_ecdsa) and POSTs them in cleartext via curl to an OAST collector.

  id_rsa, id_ed25519, id_ecdsa

• Agent Threat Rules (ATR) — open detection-rule standard for AI agent threats
  repoas techniquesIt is written here as Credential-exfiltration.

  Named as an active attack surface in production agent infrastructure; semantic rephrasings of it are cited as something regex catalogs miss

• Agent Threat Rules (ATR) — Sigma/YARA-style open detection format for AI agent threats
  otheras techniquesIt is written here as Credential-exfiltration.

  Named as one of the AI-agent attack surfaces ATR targets (no further detail given)

• CVE-2026-75130: Context7 through 2
  advisoryas techniques

  Poisoned custom instructions cause the AI agent to exfiltrate credentials from environment files to an attacker-controlled service

• "Allowed HTTP Request Domains" Restriction Bypass via AI Agents MCP Connector
  advisoryas classification.key entities