agent-security

Deep Dives and Frameworks

Implementation playbooks, operator patterns, and durable analysis.

Agent Sandboxes Need Egress Budgets, Not Trust Prompts
RAG Cost Attacks Turn Retrieval Into a Budget Risk

Signals, Maps, and Watch Lists

Production-oriented analysis, benchmarks, and market/system intelligence.

Agent Sandboxes Need Egress Budgets, Not Trust Prompts
RAG Cost Attacks Turn Retrieval Into a Budget Risk

External tools

Execution tooling is separate

Swarm Signal keeps the analysis layer. Use BoredTools for reusable production templates and trackers.

Open BoredTools Open Budget Tracker

Signal Failure Briefs Evidence-first framing

Agent Sandboxes Need Egress Budgets, Not Trust Prompts

The live risk in agent security is shifting from "will the model say something unsafe?" to "what can the harness actually touch after the model decides?"...

Signal Failure Briefs Evidence-first framing

RAG Cost Attacks Turn Retrieval Into a Budget Risk

A June 2026 paper on retrieval-augmented inference cost attacks reports a failure mode that many RAG teams are not testing: poisoned external documents...