The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of CLI tools including dotnet, aspire, docker, podman, and git to perform audits, builds, and infrastructure management.
[EXTERNAL_DOWNLOADS]: The skill facilitates the installation of .NET project templates (dotnet new install) and AI agent skills (aspire agent init) from official Microsoft/Aspire sources. It also mentions optional global installation of the @playwright/cli Node.js package.
[PRIVILEGE_ESCALATION]: Includes instructions to run aspire certs trust or dotnet dev-certs https --trust, which typically require administrative/sudo privileges to modify the system's trusted root certificate store.
[INDIRECT_PROMPT_INJECTION]: The skill audits and processes existing repository files, including CLAUDE.md, AGENTS.md, and source code. This represents an attack surface where malformed repository content could attempt to influence the agent's behavior.
Ingestion points: Phase 1 (Discovery) reads solution files, project files, SDK configurations, and application source code.
Boundary markers: While the skill emphasizes "evidence-based" discovery, it lacks explicit instruction to use unique delimiters or boundary markers when processing untrusted file content.
Capability inventory: The skill can execute arbitrary shell commands via dotnet and aspire, write/modify repository files, and manage containerized infrastructure.
Sanitization: The skill contains specific rules to redact secrets and credentials from output (Rule 7), though it does not specify sanitization for potential prompt injection patterns in processed data.
[DATA_EXFILTRATION]: The skill accesses aspire describe output and environment variables which contain generated credentials and sensitive connection strings. Although the instructions explicitly mandate redacting this information before displaying output, the underlying capability to access these secrets exists.

dotnet-aspire-local-environment