The Agent Skills Directory

[DATA_EXPOSURE]: The skill accesses .env.local and .env.example to identify environment variables required for project setup. While .env.local is a sensitive file path, the skill includes a specific guardrail instructing the agent to extract variable names only and never their values. There are no network exfiltration patterns or tools used to send this data externally.
[COMMAND_EXECUTION]: The skill uses local shell commands to retrieve repository history and recent changes.
Evidence: Step 3 executes git log --oneline -15 and git diff HEAD~5 --stat to provide context on recent work.
[INDIRECT_PROMPT_INJECTION]: The skill ingests various files from the codebase (READMEs, manifests, source code) which are considered untrusted data and could contain embedded instructions.
Ingestion points: Processes content from README.md, package.json, and source files across the src/ directory.
Boundary markers: The instructions do not specify the use of delimiters or 'ignore' instructions when reading repository content.
Capability inventory: The skill utilizes shell execution for git commands and produces a summary for the user.
Sanitization: No explicit sanitization or filtering of file content is mentioned before the information is summarized.

workflow-onboard