calibrate-comms

Calibrate how the vault's agents write to you. A multi-phase skill that profiles your communication preferences across nine axes (density, sequence, modality, abstraction, tradeoff appetite, detail, jargon, tone, and how you brief the agent), grounds them in validated psychometrics as priors, confirms them with sample-reaction tests, and compiles them into operational writing directives persisted to CLAUDE.md so every agent (Claude + Codex) obeys them. Re-runnable. Use when the user wants the agent to match their communication style, fix recurring "too long" / "too terse" / "WTF does this mean" feedback systemically, set up or re-run their communication profile, or asks how you should be writing to them.