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/GoogleCloudPlatform/gemini-cloud-assist-mcpWrote 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/commands/googlecloudplatform/gemini-cloud-assist-mcp/preferences)<a href="https://agentmods.dev/commands/googlecloudplatform/gemini-cloud-assist-mcp/preferences"><img src="https://agentmods.dev/badge/commands/googlecloudplatform/gemini-cloud-assist-mcp/preferences/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/commands/googlecloudplatform/gemini-cloud-assist-mcp/preferences"><img src="https://agentmods.dev/badge/commands/googlecloudplatform/gemini-cloud-assist-mcp/preferences.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 | $0.00017 | $0.00890 |
| Opus 5.5 | $0.00007 | $0.00356 |
| Sonnet 5.5 | $0.00003 | $0.00178 |
| Haiku 4.5 | $0.00002 | $0.00089 |
Grade A, and why
preferences 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 yesterday.
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 — 57 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Maintain the user's GCP operating preferences in ~/.config/cloud-operator/preferences.md. This file will be read before the first ask_cloud_operator call in a new session and passed as context, so it should be concise, factual, and written as directives.
If $ARGUMENTS is non-empty
Treat it as a single preference to remember (like /remember). Read the existing file, add or update the matching entry under the right heading (replace contradicting entries rather than appending duplicates), write it back, and confirm in one line. Then sync it to the operator (below).
Otherwise — bootstrap interview
- Read the existing file if present and show a short summary; ask whether to update or start fresh.
- Look for context the user already has, and propose it rather than asking from scratch:
gcloud config list --format=json(account, default project, region/zone)- IaC files in the current directory (e.g.
*.tf, provider/backend blocks) for tool, provider version, state backend, naming patterns - Any GCP-related lines in the user's CLAUDE.md files or memory
- Ask about the remaining topics in two or three short batches (not one question at a time). Accept "skip" for anything:
- Scope: organization / folder, default project (used as
projectforask_cloud_operator), other allowed projects, environments (dev/staging/prod) and how to tell them apart, operator agent ID only if they want to pin one instead of their personal default agent. Agent IDs are per user and per project: find yours in thecontextIdof your sessions (search_operator_sessions, or any operator response):projects/<p>/locations/global/agents/<AGENT_ID>/sessions/…. Save it aslocations/global/agents/<AGENT_ID>together with its project: sessions name the project by number, so the full name fails whenprojectis the project ID - Location: default region/zone, data residency constraints
- Conventions: resource naming, required labels/tags, networking (shared VPC, subnets, private IPs only)
- Tooling: Terraform vs gcloud vs console, module sources, state backend, CI/CD
- Change policy: what needs explicit approval (e.g. anything in prod, IAM changes, deletes), dry-run/plan first, change windows
- Security & compliance: least-privilege IAM, CMEK, VPC-SC, org policies, no public IPs/buckets
- Cost: budgets, preferred machine types, committed-use or spot preferences
- Communication: level of detail, whether to show commands/plans before acting
- Scope: organization / folder, default project (used as
- Write the file in this shape (omit empty sections):
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.
- yesterday First seen · 57 lines · 17 tokens per session scan A 907c622ce719
preferences is a command published in the GitHub repository GoogleCloudPlatform/gemini-cloud-assist-mcp (69 stars, last pushed 2d ago), licensed Apache-2.0. It adds 17 tokens to every session and 890 once invoked, about $0.0001 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 2026-10-18.
Other commands, from other repositories
deploy-schema
Deploy Sanity schema to the Content Lake with verification.
wasm-split
Command "wasm-split" from getsentry/toolkit, covering examples, add a build id to a module, in place, and print it, capture the build id for a later upload, also drop function names from the shipped binary and point browsers at a companion served from a cdn.
checklist
Generate a custom checklist for the current feature based on user requirements.
clarify
Identify underspecified areas in the current feature spec by asking up to 5 highly targeted clarification questions and encoding answers back into the spec.
specify
Create or update the feature specification from a natural language feature description.
analyze
Perform a non-destructive cross-artifact consistency and quality analysis across spec.md, plan.md, and tasks.md after task generation.