Intent-bound action authorization for AI agents — deterministic, per-action approval a prompt injection can't re-point, with a signed, tamper-evident audit trail.
-
Updated
Jun 11, 2026 - Python
#
confused-deputy
Here are 15 public repositories matching this topic...
Proxilion is the security layer for the agentic workforce. It turns managed AI agents into governed users by enforcing strict cryptographic boundaries on every API call to SaaS like Google Workspace, Salesforce, or Atlassian.
cryptography self-hosted reverse-proxy cybersecurity oauth-proxy human-in-the-loop zero-trust data-loss-prevention ai-security audit-logging saas-security ai-governance llm-security agentic-ai prompt-injection-defense openai-security claude-security confused-deputy managed-agents pic-protocol
-
Updated
Sep 5, 2026 - Rust
Open deterministic security tests for unsafe multi-agent handoffs and authority escalation.
python cli fuzzing privilege-escalation multi-agent-systems ai-agents red-teaming pypi-package prompt-injection llm-security agent-security security-benchmark confused-deputy
-
Updated
Sep 10, 2026 - Python
Penetration testing for AI agents — find, prove, and measure cross-server confused-deputy / prompt-injection exfiltration chains in an MCP tool mesh. Local-first, bring-your-own-model.
mcp cybersecurity penetration-testing pentest red-team ai-agents ai-security prompt-injection llm-security model-context-protocol indirect-prompt-injection confused-deputy
-
Updated
Aug 3, 2026 - Python
IDOR / Broken Object-Level Authorization in AI agent tool-calling — a hands-on vulnerable lab. An LLM tool trusts the object id it is handed and skips the ownership check, so any user reads any tenant's data.
appsec ai-agents ai-security idor security-lab broken-access-control bola llm-security tool-calling owasp-llm confused-deputy excessive-agency
-
Updated
Jul 31, 2026 - HTML
10 hands-on demos for agentic AI security: blind verification, AIBOM, eval invariants, authority confinement, recon/malware/LFI/SSRF/scan — offline only.
teaching-materials autonomous-agents vulnerability-scanning ai-security traffic-interception file-inclusion security-education supply-chain-security malware-triage network-reconnaissance llm-security aibom agentic-ai confused-deputy synthetic-fixtures blind-verification reproducible-evaluation evaluation-invariants authority-confinement evidence-backed-release
-
Updated
Sep 15, 2026 - Python
Research artifact for the JSS 2026 paper on Prompt-to-API-Call (P2A) injection attacks against LLM agents: 52 attack vectors across 5 REST backends (260 scenarios), five defenses (D1-D5), and the D5 Semantic Intent Validator. Includes the harness, attack corpus, and all result data.
ai-safety red-teaming prompt-injection llm-security llm-agents llm4se rest-api-security confused-deputy
-
Updated
Jul 16, 2026 - Python
The first benchmark environment for Sensitivity Awareness (SA) in LLMs. Evaluating how language model agents handle Role-Based Access Control (RBAC), Confused Deputy vulnerabilities, and Contextual Authorization rules.
role-based-access-control datasets access-control llm-security acl2025 agent-se confused-deputy contextual-authorization
-
Updated
Jun 9, 2026 - Python
Zero-trust capability delegation for MCP multi-agent systems. Solves the confused deputy problem with scoped JWT tokens, deterministic enforcement, and full audit trail.
-
Updated
May 11, 2026 - Python
Capability-aware risk scoring for AI coding agent tool calls. Five-dimensional decomposition, session-graph composition, repo awareness, the Agent Control Plane and Confused Deputy doctrines. Adapters for the standalone shell, Claude Code hook, MCP server, and Claude Code plugin.
python mcp ai-safety risk-scoring capability-security prompt-injection agent-security claude-code confused-deputy tool-call-firewall
-
Updated
May 4, 2026 - Python
RFC 8707 confused-deputy conformance prober and enforcing sidecar for MCP server OAuth (RFC 9068 audience binding)
-
Updated
Jun 1, 2026 - Python
The open specification for delego — a deterministic authorization & audit protocol for AI-agent actions.
-
Updated
Aug 20, 2026 - Python
A practical field manual for platform engineers on agent identity: authorization, least privilege, and audit for LLM agents as first-class principals. Shipped one chapter + runnable demo at a time.
-
Updated
Jun 1, 2026
Deliberately vulnerable 3-agent LangGraph system: 10 documented vulnerabilities with working PoCs, framework mappings and remediations.
owasp multi-agent security-research prompt-injection llm-security langgraph agentic-ai ai-red-teaming vulnerable-by-design confused-deputy
-
Updated
Sep 15, 2026 - Python
Reproducible MCP tool-poisoning (line-jumping) demo, contained by an independent plan-vs-authorize gate. Own testbed, mock server, fake secret — attacks nobody.
-
Updated
Sep 15, 2026 - Python
Add this topic to your repo
To associate your repository with the confused-deputy topic, visit your repo's landing page and select "manage topics."