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/huaweicloud/huaweicloud-devkitWrote 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/huaweicloud/huaweicloud-devkit/huawei-agent-rules)<a href="https://agentmods.dev/rules/huaweicloud/huaweicloud-devkit/huawei-agent-rules"><img src="https://agentmods.dev/badge/rules/huaweicloud/huaweicloud-devkit/huawei-agent-rules/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/huaweicloud/huaweicloud-devkit/huawei-agent-rules"><img src="https://agentmods.dev/badge/rules/huaweicloud/huaweicloud-devkit/huawei-agent-rules.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 | $0.00918 | $0.00918 |
| Opus 5 | $0.00459 | $0.00459 |
| Sonnet 5 | $0.00184 | $0.00184 |
| Haiku 4.5 | $0.00092 | $0.00092 |
Grade A, and why
huawei-agent-rules 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 2d 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 — 79 lines — stays where its author put it; the contents beside it link to each section on GitHub.
HuaweiCloud Devkit — Agent Rules
Agent 加载插件时自动读取，定义全局行为约束。
Core Principles
- Always use the Huawei MCP server first (sandboxed, observable, auditable). Fall back to
hcloudCLI only when MCP is unavailable. - Before any task, check for available skills. Use
huaweicloud_search_docsorhuaweicloud_retrieve_skillvia MCP. Follow skill procedures over general knowledge. - When uncertain about API parameters, quotas, limits, or error codes — search documentation via MCP. Do not guess.
- Prefer IaC over direct CLI calls for infrastructure creation (Terraform for Huawei Cloud, or CloudFormation equivalent).
- Never use em dashes in resource names/tags. Use hyphens only.
- Match user language. Respond in the same language the user writes in.
Secret Safety
- MUST load the
huawei-dewskill first for any secret, credential, API key, token, or password task. - MUST NOT call
hcloud csms download-secretorhcloud csms show-secretdirectly. - MUST NOT call
hcloud kms decryptdirectly. - MUST use
{{resolve:csms:secret-id:SecretString:key}}with the MCP proxy — secrets resolve at runtime, never enter agent context. - NEVER echo AK/SK values in conversation or logs.
IAM Security
- Least privilege by default. Start with no permissions, add only what is needed.
- MUST NOT create IAM users — use IAM Identity Center (OneAccess) or temporary credentials.
- MUST NOT create long-term AK/SK — use temporary STS tokens.
- MUST scope resource ARNs — no
*wildcards. - Confused deputy protection: Add
g:SourceAccountandg:SourceUrnconditions to trust policies.
Network Security
- MUST block public access by default on OBS buckets.
- MUST use security groups (not iptables) for ECS/CCE access control.
- SHOULD use VPC endpoints for internal service access (no public internet).
- MUST enable HTTPS-only on OBS buckets and API endpoints.
Observability
- SHOULD enable CTS audit logging on all accounts.
- SHOULD configure Cloud Eye alarms for production workloads (CPU > 80%, memory > 85%, disk > 90%).
- SHOULD enable OBS access logging for audit compliance.
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.
- 2d ago First seen · 79 lines · 918 tokens per session scan A 212ad2eff565
huawei-agent-rules is a cursor rule published in the GitHub repository huaweicloud/huaweicloud-devkit (46 stars, last pushed today), licensed Apache-2.0. It adds 918 tokens to every session, about $0.0046 per session on Opus 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 2026-09-07.
Other cursor rules, from other repositories
atmos-rules
This is a Golang CLI project using Cobra for command structure and Viper for configuration management. The project maintains high code quality standards through golangci-lint and aims for maximum test coverage. The repository also includes the website source code for https://atmos.tools.
cursor-rule
This project can use agentmap (@raymondchins/agentmap) — a queryable import/symbol map at .claude/agentmap/.
hatch3r-scalability
CQ6 Scalability Quality measurement rule — stateless-handler ratio, back-pressure pattern adoption, idempotency-key adoption on POST/PUT/PATCH, queue offloading for >1s ops, pool sizing per concurrency profile.
documentation
Documentation: code docs, READMEs, ADRs, maintenance.
git-workflow
Git workflow: commits, branches, PRs, history management.
playwright
Playwright: e2e testing, page objects, fixtures, assertions.