The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill's README recommends installation via npx @tamng0905/builder-essential-skills, which targets a package repository associated with the skill's author.
[EXTERNAL_DOWNLOADS]: The tactics.md file defines search and data collection workflows that interact with several well-known academic and developer-focused APIs, including arXiv.org, Semantic Scholar, OpenAlex, deps.dev, pypistats.org, and npmjs.org. These are legitimate endpoints for the skill's stated purpose of technology and academic research.
[COMMAND_EXECUTION]: The Integrator role is granted the capability to write research artifacts and perform git commit operations. These actions are restricted to the research output directory (docs/research/) and are gated by a mandatory audit pass from an independent sub-agent.
[PROMPT_INJECTION]: The skill processes untrusted external data fetched from various web sources (e.g., blogs, forums, and code repositories). This represents an indirect prompt injection surface.
Ingestion points: Data enters through various Researcher sub-agents fetching URLs and API responses.
Boundary markers: The skill uses a HARD CONTEXT RULES preamble to limit the amount of text a researcher can fetch from a single source (e.g., quoting at most 2 sentences).
Capability inventory: The Integrator agent has the ability to write to the file system and commit to the repository.
Sanitization: The skill mitigates risks through a 'Separation of roles' contract, where Verifier and Auditor agents independently check claims and reports before they are integrated, preventing a single agent from both gathering and validating the same information.

lead-research