Lean Claude Code Harness
Security checks across static analysis, malware telemetry, and agentic risk
Overview
Instruction-only skill that gives high-level best-practices for building a small, auditable Claude-style coding-agent harness; it requests no credentials, installs, or filesystem/network access and is internally consistent with its stated purpose.
This skill is an instruction-only best-practices guide for designing a small, auditable Claude-style coding-agent harness. It does not install code, request credentials, or declare any network endpoints. It is safe from a supply-chain perspective, but remember: the outputs are advisory — if you give an agent running this guidance access to your filesystem, network, or credentials, the agent could use those platform privileges to read or write files (the skill recommends persisting transcripts). Before using it in an automated agent, review the harness changes it proposes, and restrict the agent's runtime permissions (file writes, shell access, network access) until you have manually verified the design and verification plan.
Static analysis
No static analysis findings were reported for this release.
VirusTotal
VirusTotal findings are pending for this skill version.
Risk analysis
No visible risk-analysis findings were reported for this release.