The Agent Skills Directory

[SAFE]: The skill serves exclusively as a documentation and architectural guide for LangChain v1.3.15. All scripts are utility tools for version checking and documentation maintenance.
[EXTERNAL_DOWNLOADS]: The skill provides standard installation commands for official LangChain packages (e.g., langchain, langchain-openai, langchain-anthropic) via PyPI. These are recognized industry-standard libraries.
[COMMAND_EXECUTION]: Includes a maintenance script scripts/check-updates.py which performs local file integrity checks and fetches version information from the official PyPI JSON API (pypi.org). The script is restricted to reporting and does not execute arbitrary shell commands.
[DATA_EXPOSURE_AND_EXFILTRATION]: Documentation correctly illustrates best practices for secret management by using environment variables (e.g., export OPENAI_API_KEY="sk-..."). No sensitive data is hardcoded or exfiltrated.
[PROMPT_INJECTION]: No evidence of instructions intended to bypass AI safety filters or override system personas. The triggers are contextually appropriate for LangChain development tasks.
[DYNAMIC_EXECUTION]: While LangChain itself supports complex agentic workflows, this skill's instructions and provided scripts do not utilize unsafe eval(), exec(), or dynamic code loading from untrusted sources. Reference to eval() in references/05-tools.md is a documented example of a calculator tool, not a functional part of the skill's own execution logic.

langchain