The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes shell utilities including find, grep, cat, ls, and jq to traverse the directory structure, identify frameworks, and extract metadata from project manifest files like package.json or requirements.txt.- [DATA_EXFILTRATION]: The skill scans for sensitive file paths such as .env, .aws/config, and SSH directories as part of its directory mapping and completeness assessment. This access is restricted to detecting existence and reporting locations within the local 'analysis-report.md' file; no network-based exfiltration or data transmission to remote domains was identified.- [PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection because it reads and processes arbitrary content from the local codebase (e.g., README files and source code) to generate summaries.
Ingestion points: Local codebase files read via cat and grep during analysis.
Boundary markers: Absent; the skill does not wrap ingested content with specific delimiters or instructions to ignore embedded commands.
Capability inventory: File system access (read/write) and shell command execution.
Sanitization: None; the skill assumes the analyzed codebase is safe for summary generation.- [REMOTE_CODE_EXECUTION]: The skill instructs the user to copy plugin slash commands from a local directory (~/.claude/plugins/stackshift/) to the project workspace. While this involves command execution, it relies on locally existing plugin files and does not perform unverified remote downloads.

analyze