
from linkly-ai-skills34
Search, browse and read local documents (PDF, MD, DOCX, TXT, HTML) via CLI or MCP with progressive search → outline → read workflows.
Linkly AI provides local-document search and progressive reading tools to index, search, outline and read files on a user's machine or curated libraries. It exposes CLI and MCP tool paths for programmatic agents, returning structured search results, outlines, grep hits, and paginated document reads.
Use this skill when a user asks to search their local documents, find files about a topic, list knowledge libraries, browse document outlines, or read a specific document. Also triggers on connection/troubleshooting phrases like 'linkly not working' or 'can not connect to linkly'.
Agents that can run shell or MCP tools (CLI-friendly agents, coding assistants with MCP tool support) — e.g., Claude Code, Cursor, agents using shell access or an MCP bridge.
Linkly AI provides local and cloud document search via CLI or MCP with a progressive search → outline → read workflow. The SKILL.md is exceptionally well-documented with clear steps, troubleshooting, and best practices. The bundled script (package.sh) is a release/packaging tool, not part of the skill's core functionality — it failed in sandbox due to no git repo, which is expected. No security concerns found; the script requires explicit user confirmation before pushing.
Well-crafted skill with thorough documentation. The only bundled script is a dev release tool, not part of the skill's user-facing functionality. No security issues detected — no creds, no exfiltration, no destructive commands. The git push in the script requires explicit 'yes' confirmation.