The Agent Skills Directory

[SAFE]: The skill uses a 'micro-file architecture' and a facilitator-based interaction model that requires mandatory user checkpoints ('C' to continue) at every stage of the process.
This design pattern ensures that the agent cannot proceed from discovery to research, or from research to document generation, without the user reviewing and approving the interim findings and the proposed plan.
The workflow logic is encapsulated in sequential step files (steps 01-08) and a bundled research sub-skill, providing a transparent and predictable execution path.
[SAFE]: File system operations are scoped and transparent. The skill asks the user for an output location at the start of the workflow and limits its write operations to that specific directory.
In the discovery phase (step-01), the skill scans the workspace for relevant documentation, but this is explicitly described as a search for markdown files and technical briefs to provide context for the user's specific topic.
[SAFE]: The skill relies on web search for its research functionality but implements several safety and quality guardrails.
It requires the agent to cite sources with hyperlinks for all factual claims in the generated documents (step-06, step-11 of the appendix).
The 'Verified Source Summary' requirement in the appendix ensures that any data ingested from the web is traceable and can be audited by the user.
A comprehensive validation checklist (step-07) is used to verify the integrity and consistency of the final report before completion.

trade-off-analysis