Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
git clone --depth 1 https://github.com/Contoso-State/red-team-agent-orchestrationWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/rules/contoso-state/red-team-agent-orchestration/01-redteam-graph)<a href="https://agentmods.dev/rules/contoso-state/red-team-agent-orchestration/01-redteam-graph"><img src="https://agentmods.dev/badge/rules/contoso-state/red-team-agent-orchestration/01-redteam-graph/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/rules/contoso-state/red-team-agent-orchestration/01-redteam-graph"><img src="https://agentmods.dev/badge/rules/contoso-state/red-team-agent-orchestration/01-redteam-graph.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00550 | $0.00550 |
| Opus 5.5 | $0.00220 | $0.00220 |
| Sonnet 5.5 | $0.00110 | $0.00110 |
| Haiku 4.5 | $0.00055 | $0.00055 |
Grade A, and why
01-redteam-graph scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 30 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Azure Red Team — Graph Orchestration Standard
Every engagement in this framework runs as ONE explicit, declarative graph. The single source of truth is graph/redteam.graph.json (redteam-azure v2.1.0): 15 nodes over a 12-specialist read-only roster, validated by tools/graph/validate-graph.mjs.
Two engines, one graph
- Dependency-free Node runner —
tools/graph/run-graph.mjs— executes simulated dispatch. The separatetools/graph/run-live.mjsuses a scoped Claude live adapter and fresh preflight; other runtimes use native agent orchestration. - First-class LangGraph target —
integrations/langgraph/— compiles the same JSON into a PythonStateGraphfor deployment.
Control flow
validate_scope (one subscription + read-only attestation) -> memory_load -> preflight_inventory -> build_security_context -> scoped Send fan-out (plan_specialists -> run_specialist) -> deterministic fan-in merge (collect_raw) -> bounded evaluator-optimizer reflection loop (evaluate, max_revisions=2, quality_threshold=0.85) -> Agent-as-a-Judge false-positive gate (judge) -> human-authorized active lanes (authorize_active interrupt) -> correlate -> report -> reflexion_debrief.
Self-improving loops (primary standard)
The evaluator-optimizer cycle, the Agent-as-a-Judge gate, and the reflexion debrief are enabled by default with evidence-gated methodology promotion and an explicit off override — tools/graph/self-improve.mjs. Cross-run learning is written ONLY to the memory/methodology/ namespace.
Immutable boundary
Read-only enforcement in guardrails/guard.mjs sits OUTSIDE the learning surface. Self-improvement can never modify guardrails/** or any guardrail namespace — that firewall is the one thing the graph cannot rewrite.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 1mo ago Changed · +12 tokens per session 693be92459ff
- 1mo ago Changed e40f0aebb9f1
- 1mo ago First seen · 30 lines · 538 tokens per session scan A e6cd3e8d8d5f
01-redteam-graph is a cursor rule published in the GitHub repository Contoso-State/red-team-agent-orchestration (6 stars, last pushed 1mo ago), licensed MIT. It adds 550 tokens to every session, about $0.0022 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other cursor rules, from other repositories
ponytail
Ponytail, lazy senior dev mode. Always pick the simplest solution that works.
angular-20
This rule provides comprehensive best practices and coding standards for Angular development, focusing on modern TypeScript, standalone components, signals, and performance optimizations.
dev-standard
Apache Superset development standards and guidelines for Cursor IDE.
control-plane-descriptors
Control plane descriptor and instance implementation patterns.
family-instance-domain-actions
Family instance domain action implementation patterns.
no-family-vocabulary-in-framework
The framework domain (packages/1-framework) carries no family- or target-specific vocabulary — types, fields, hooks, or strategy values — and, outside the CLI and the PSL parser's grammar names, no Prisma version name before 8. Enforced by the no-family-vocabulary Biome plugin plus the lint:ratchets ratchet.