The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands like git init, uv, and python3 to initialize the project structure and run linting scripts. These operations are limited to the local environment and are consistent with the skill's purpose of managing a markdown-based knowledge vault.
[EXTERNAL_DOWNLOADS]: The skill utilizes the uv tool to manage the pyyaml dependency for its vault validation script. Both uv and pyyaml are well-known, widely trusted tools and packages in the development community.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it is designed to process external research reports and earnings call transcripts. This is mitigated by an explicit five-point human-in-the-loop (HITL) confirmation process required for data ingestion, as documented in references/ingest_sop.md. The capabilities associated with this surface include local file operations and repository management.
[SAFE]: The provided Python scripts, init_vault.py and lint-vault.py, perform legitimate project scaffolding and structural validation. The linting script correctly uses yaml.safe_load() to prevent unsafe deserialization risks.

llm-wiki-setup