Guardians: Static verification for AI agent workflows
an implementation of the ideas in Erik Meijer's "Guardians of the Agents: Formal Verification of AI Workflows" (Communications of the ACM, January 2026) - metareflection/guardians
Opening excerpt (first ~120 words) tap to expand
Guardians Static verification for AI agent workflows. An implementation of the ideas in Erik Meijer's "Guardians of the Agents" (CACM, January 2026). The paper's thesis: the root cause of prompt injection in agentic systems is the same as SQL injection — code and data aren't separated. The fix is the same too. Instead of letting the LLM call tools one at a time and decide what to do after each result, the LLM generates a structured plan upfront using symbolic references (placeholders, not real data). A static verifier checks the plan against a security policy before any tool runs. Only verified plans execute.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at GitHub.