chore: improve linkup guidance and skill docs

[aliou]

Summary

• tighten system prompt guidance wording for Linkup usage
• keep guidance retrieval-first and remove marketing phrasing
• improve Linkup skill doc structure for concision and triggerability
• add skills/linkup/references/examples.md for extra examples and edge-case handling
• keep SKILL.md lean and link to references for progressive disclosure

Commits

• d64d041 chore: tighten linkup guidance wording
• 9124469 docs: add linkup skill references and edge-case examples

Why the skill changes

The skill was compressed to follow skill-creator guidance (low token overhead, high-signal instructions). That improved clarity, but removed some operational detail. This update restores high-value detail without bloating SKILL.md:

• keeps core decisions in SKILL.md (tool choice, depth, query shape)
• moves richer examples and failure handling to references/examples.md
• re-adds explicit typed signatures for faster, safer tool usage

This keeps trigger metadata and main instructions short, while still giving the agent a path to deeper guidance when needed.

[aliou]

Why this skill update:

• We wanted SKILL.md to follow skill-creator guidance: concise, high-signal, low token cost.
• The previous compression improved brevity but dropped some practical guidance.
• This PR keeps SKILL.md focused on core decisions and adds a references file for deeper examples + edge cases.
• Net effect: better triggerability and maintainability, without losing operational depth when needed.