heating the forge...

@forgecat/microsoft-azure_skills-azure-cost-optimization — Security Risk Report | ForgeCat

microsoft-azure_skills-azure-cost-optimization

Security risk report for @forgecat/microsoft-azure_skills-azure-cost-optimization v0.0.4

Low

Static scanner rated source as safe: the profile is sourced from Microsoft's official azure-skills repository (https://github.com/microsoft/azure-skills), a legitimate, well-known upstream.
Profile metadata correctly attributes authorship to Microsoft and includes version tracking and license (MIT).
No evidence of supply-chain poisoning, typosquatted dependencies, or attacker-controlled package references.

Low

Profile contains no instructions to manipulate the AI, ignore system prompts, exfiltrate credentials, or hide its own instructions.
Content is purely educational and procedural: it documents Azure Cost Management API workflows, request schemas, error handling, and best practices for cost analysis and optimization.
No guidance poisoning detected: the profile recommends legitimate Microsoft tools (az rest, Azure CLI, Azure Monitor, Azure Resource Graph) and official Azure APIs, not typosquatted or backdoored packages.
The profile explicitly instructs safe practices (e.g., 'Get approval before deleting resources', 'Test changes in non-production first', 'Never execute destructive operations without explicit approval'), which strengthens rather than weakens security.

Details

Findings

Invisible characters detected
skills/azure-cost/cost-forecast/error-handling.md · SKILL-011
Invisible characters detected
skills/azure-cost/cost-forecast/guardrails.md · SKILL-011
Invisible characters detected
skills/azure-cost/cost-forecast/request-body-schema.md · SKILL-011
Invisible characters detected
skills/azure-cost/cost-forecast/workflow.md · SKILL-011
Invisible characters detected
skills/azure-cost/cost-optimization/azure-resource-graph.md · SKILL-011
Invisible characters detected
skills/azure-cost/cost-optimization/workflow.md · SKILL-011
Invisible characters detected
skills/azure-cost/cost-query/dimensions-by-scope.md · SKILL-011
Invisible characters detected
skills/azure-cost/cost-query/error-handling.md · SKILL-011
Invisible characters detected
skills/azure-cost/cost-query/examples.md · SKILL-011
Invisible characters detected
skills/azure-cost/cost-query/guardrails.md · SKILL-011
Invisible characters detected
skills/azure-cost/cost-query/request-body-schema.md · SKILL-011

Low

Profile declares no tools or MCP servers with excessive permissions; it is purely markdown documentation and workflow guidance.
Recommended operations are scoped to read-only queries (Cost Management API queries, Azure Monitor metrics, Resource Graph queries) and informational commands (az rest, az graph query, kubectl top).
Destructive operations (resource deletion, tier upgrades) are explicitly gated behind user approval and warnings, with dry-run and rollback procedures documented.

Details

Evidence

'Get approval before deleting resources' and 'Provide dry-run commands for validation' (SKILL.md, Safety Requirements)
'Warn user that upgrading from Free to Standard introduces an ongoing cluster management fee' (azure-aks-cost-addon.md)
All cost queries use read-only API endpoints (Microsoft.CostManagement/query, Microsoft.CostManagement/forecast)

Low

Profile declares no MCP servers or tool definitions; it is documentation only.
References to MCP tools (azure__documentation, azure__extension_cli_generate, azure__get_azure_bestpractices, azure__extension_azqr, azure__aks, azure__redis) are descriptive guidance on which tools to use, not embedded tool definitions or hidden instructions.
No binary execution, unrestricted network access, or arbitrary filesystem operations are defined or instructed.

Details

Evidence

'MCP Tools' section in SKILL.md lists tools by name and purpose but does not define them or embed malicious instructions in tool descriptions.
Profile recommends using MCP tools 'over direct CLI commands' as a safety best practice, not as a vector for attack.