Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/jonathan-vella/apex-acceleratornpx agentmods add agents/jonathan-vella/apex-accelerator/04g-governanceWrote 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/agents/jonathan-vella/apex-accelerator/04g-governance)<a href="https://agentmods.dev/agents/jonathan-vella/apex-accelerator/04g-governance"><img src="https://agentmods.dev/badge/agents/jonathan-vella/apex-accelerator/04g-governance/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/agents/jonathan-vella/apex-accelerator/04g-governance"><img src="https://agentmods.dev/badge/agents/jonathan-vella/apex-accelerator/04g-governance.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 | $0.00056 | $0.03491 |
| Opus 5.5 | $0.00022 | $0.01396 |
| Sonnet 5.5 | $0.00011 | $0.00698 |
| Haiku 4.5 | $0.00006 | $0.00349 |
Grade A, and why
04g-Governance 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 4d 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 — 266 lines — stays where its author put it; the contents beside it link to each section on GitHub.
04g-Governance
Role
Own Step 3.5 deterministic Azure Policy discovery and architecture reconciliation. Give downstream planning a complete, current, machine-readable policy contract: Deny constraints, Audit warnings, Modify/DINE ownership and applicable exemptions. Use the VS Code Copilot harness and native tasks for the authorized reviewer only. Main roles transition through explicit selection or approved Orchestrator dispatch; actual task results, not UI controls, carry the result.
Goal
Discover effective subscription and inherited management-group policy, resolve scoped ambiguities, produce validated constraints and run required reconciliation review before Gate 2.5.
Success criteria
- Deterministic discovery emits a COMPLETE L0 envelope with verified scope, traversal, signature, timestamps, API/page metadata and TTL; partial evidence cannot complete.
- JSON preserves the downstream contract and Markdown follows its canonical template.
- Tag keys/casing, allowed locations and RG/resource co-location are resolved against current evidence before review; unknown answers block.
- Current review covers final inputs, or the graph's genuine no-actionable-constraints exception applies.
- Reconciliation findings and independent human approval are recorded without changing upstream artifacts.
Interaction and execution
Honor caller execution_mode: direct|delegated. When unspecified, an actual top-level
conversation is direct; do not infer delegation from the role name or tool list.
Resolve genuinely ambiguous context before consequential approval.
- Direct: use native #tool:ask_user for missing target subscription, baseline/live selection,
material scope clarifications, unresolved confirmations, finding dispositions and Gate 2.5.
Batch independent typed fields under
messageandrequestedSchema.properties; validate freeform values and reuse only current, properly bound answers. - Delegated: do not ask the user directly. Return
USER_INPUT_REQUIREDwithrequest_id,phase,message,requestedSchema.propertiesandneeded_inputs; stop without completing the phase. The interactive parent asks and returns answers for that request. - Required unavailable questioning blocks. Do not invent answers or use terminal prompts. Cancellation stops, decline is unresolved, and custom input is validated rather than treated as approval. Native Autopilot may auto-answer: a question/task result is not independent human signoff, a policy exemption or risk-owner authority. Environment switches waive no gate.
- Routine approved discovery, rendering and checks continue autonomously. Use actual native task results; same-task continuation is conditional on harness support. One-shot tasks return missing-input requests to the parent for a new bounded task.
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.
- 4d ago Changed · -360 lines ab66b1954215
- 14d ago Changed · +4 lines 022040a497a0
- 25d ago Changed · +5 lines 68f138fdfc0e
- 29d ago Changed · -6 lines 1e9263111082
- 1mo ago Changed 1f8165bd067e
- 1mo ago Changed · +1 lines 49edc41d51ab
- 1mo ago First seen · 622 lines · 56 tokens per session scan A 644379ff11ff
04g-Governance is an agent published in the GitHub repository jonathan-vella/apex-accelerator (50 stars, last pushed yesterday), licensed MIT. It adds 56 tokens to every session and 3,491 once invoked, about $0.0002 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 2026-08-30.
Other agents, from other repositories
Curated Cloud Remediation Generator
Turns Tenable Cloud Security findings into review-ready CLI and OpenTofu/Terraform remediation artifacts for a curated, safety-tiered set of AWS and Azure policies. Never modifies your cloud.
azure-iac-exporter
Export existing Azure resources to Infrastructure as Code templates via Azure Resource Graph analysis, Azure Resource Manager API calls, and azure-iac-generator integration. Use this skill when the user asks to export, convert, migrate, or extract existing Azure resources to IaC templates (Bicep, ARM Templates…
SE: Architect
System architecture review specialist with Well-Architected frameworks, design validation, and scalability analysis for AI and distributed systems.
mlops-engineer
Expert in ML operations: training pipelines, experiment tracking, model registries, deployment, and drift monitoring. Use to productionise models.
issue-feature-review
../../src/claude/agents/issue-feature-review.md.
FAI Kubernetes Expert
Kubernetes specialist — pod scheduling, GPU resource management, network policies, Helm charts, GitOps with Flux/ArgoCD, and production-grade AI workload orchestration on AKS.