test

name	test
description	testse
safety_level	standard
guardrail_summary	["etst"]
tags	["test"]

test

Use this skill for work done through MCP tools against the target service.

When to use

• The user asks for operational work that touches the target service.
• The request is free-form, ambiguous, or safety-sensitive.
• The environment, tenant, realm, project, or profile must be resolved before mutation.

Mandatory workflow

1. Start with environment and permission discovery using the correct read-only MCP tools.
2. Normalize the user request into a short structured plan before making changes.
3. Resolve exact targets with read-only discovery tools before any mutation.
4. Execute only the minimum required mutations.
5. Run read-only verification after every mutation and compare the final state with the request.
6. Stop and ask the user whenever discovery is incomplete or the target is ambiguous.

Hard rules

• Always prefer exact identifiers over fuzzy matching.
• Never mutate if discovery data is incomplete.
• Never switch context mid-run unless the user explicitly asks and confirms it.
• Always pass required environment arguments explicitly when the MCP tool supports them.
• If this skill defines runtime policy, treat it as mandatory and assume those guardrails are enforced by the platform.
• If this skill works with service-specific MCP tools, use the original tool names in the policy, for example service_current_environment.

Reporting

• State which environment, tenant, realm, profile, or project was used.
• State which entities were discovered before mutation.
• State which mutations were applied and which were skipped.
• State which verification calls were used.
• State any ambiguity, blockers, or follow-up required.