← ClaudeAtlas

harness-principleslisted

Auto-triggers on structural harness decisions: new chain primitive, new flow, remove evaluator, wrap evaluator, redesign harness, scaffolding, load-bearing, model upgrade, harness audit, refactor flow, sub-agent. Also triggers on file mentions of `src/application/chain/`, `src/application/flows/`, `src/integration/ai/providers/_engine/`. Instructs the agent to read the principles doc before proceeding.
lukas-grigis/ralphctl · ★ 10 · AI & Automation · score 77
Install: claude install-skill lukas-grigis/ralphctl
# Harness Principles — Auto-trigger This skill exists because the harness research evolves with model capability. What was load-bearing in Opus 4.5 may be overhead in Opus 4.7. Without an explicit read-first gate, structural changes happen without awareness of the research behind the current design. **Before proceeding with any structural harness change:** 1. `Read .claude/docs/HARNESS-PRINCIPLES.md` in full. 2. Identify the relevant principle section(s) by name (e.g. "§ 14 Minimal scaffolding", "§ 7 Idle watchdog"). 3. Note the current `ralphctl status` for each relevant principle (`applied` / `partial` / `gap`). 4. Proceed with the change, keeping the principle's intent visible in the diff. If you are **removing** a harness component, confirm which principle the component implements and verify the removal does not regress an `applied` row. If the removal moves a row from `applied` to `gap`, update `HARNESS-PRINCIPLES.md` as part of the same commit. If you are **adding** a new chain primitive — there are five and only five: `element` (interface), `leaf`, `sequential`, `loop`, `guard`. Push back if a proposal adds a sixth. The principle behind this constraint is § 14 (Minimal scaffolding) and the rationale is in `CLAUDE.md § Architecture`.