Context Pattern — Skills

Pattern: Context
Component: skills.md
Version: 1.3 | Updated: 2026-07-29

---

Skill	Path	Why required
Context	skills/context.skill.md	Trust labels, budgets, pruning
Security	skills/security.skill.md	Injection boundaries for retrieved content
Documentation	skills/documentation.skill.md	When packing docs as evidence

Related agents

• Context engineer

Skill activation triggers

Run Context whenever retrieval or packing precedes generation. Run Security when untrusted text enters the pack. Run Documentation only when packing published specs or ADRs as evidence, not chat residue.

Verification that skills worked

• Every pack segment has a trust label and size budget
• Untrusted retrieval never sits in the system-instruction channel
• Pruned items are listed so humans can restore signal deliberately

Changelog

• 1.3 (2026-07-29): Context skill activation and verification rewritten for this envelope.