Input Token Overheads

Audit every source of per-turn input token cost on a Hermes Agent instance. Measure each, rank by cost, act on the top consumers.

Run the consolidated audit with scripts/audit_overheads.py (procedure step 1, in one command): --json for structured output, --chart <png> for a shareable chart, plus a context-doctor-style check of every always-injected file — TRUNCATED (over injection cap, silently cut every turn) and MISSING statuses, exit 1 when any file needs attention. Ported from jzOcb/context-doctor (MIT; OpenClaw-only — concepts only, script is native).

When to Use

• User says "token overhead", "context too large", "why is input so expensive"
• Model output quality degrades from context dilution
• Cost optimization — fewer input tokens per turn means lower API spend
• After adding skills, plugins, or tools — verify the overhead delta

The Overhead Map

Every turn, Hermes injects these blocks into the system prompt before the user's message: