harness-guards

name	Harness Guards
description	Centralized guard catalog and auto-scan protocol for generating context-aware safety guards across plans, workflows, and artifacts. Loaded just-in-time.
source	catalog

Skill: Harness Guards

Centralized skill for generating and managing safety guards (harness comments) across all PARA artifacts that contain phased progression or protected content.

Pattern: Sidecar Skill — this skill contains NO executable logic, only data resources.

When to Load

• /plan create → Step 8.5 (before writing plan): load references/guard-catalog.md + references/scan-protocol.md
• Creating or editing artifacts with phases/progression: load references/guard-catalog.md
• Writing workflow steps that need safety gates: load references/guard-catalog.md

References

File	When	Purpose
references/guard-catalog.md	Always when this skill is triggered	Classification of all guard types with syntax and placement rules
references/scan-protocol.md	/plan create Step 8.5 or when generating guards for a new artifact	Protocol for scanning project rules/skills to generate context-aware guards

Convention: Data files live in references/ (not templates/). This follows the Sidecar Skill convention formalized in v1.7.6.3.

Key Principles

1. Skill = Data only. Guard generation logic belongs in the workflow, not here.
2. Proactive Trigger Check in agent-behavior.md enforces loading this skill before side-effects.
3. Auto-Scan Protocol reads project DNA (rules + skills indices) to produce project-specific guards instead of generic ones.
4. Progressive enforcement: If Agent consistently ignores guards → escalate to a dedicated Rule file.