What is harness-tuning?

A guide for changing Klimkit’s source files that control Codex instructions, model settings, helper agents, hooks, and bundled skills. Klimkit is a repository-managed setup for sharing these settings across machines.

Use it when changing the shared Codex setup or applying changes through Klimkit. It covers edits to guidance, configuration, agents, skills, and hooks in the source pack.

It prevents edits to generated files that may later be overwritten. It also shows where the source files live and which validation steps to run after changing them.

harness-tuning — Skill for Codex by klimentij

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent

npx skills add klimentij/klimkit --skill harness-tuning

Clone the repo

git clone --depth 1 https://github.com/klimentij/klimkit

Made for: Codex.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

README.md

[![agentmods](https://agentmods.dev/badge/skills/klimentij/klimkit/harness-tuning.svg)](https://agentmods.dev/skills/klimentij/klimkit/harness-tuning)

Your own site

<a href="https://agentmods.dev/skills/klimentij/klimkit/harness-tuning"><img src="https://agentmods.dev/badge/skills/klimentij/klimkit/harness-tuning.svg" alt="Measured on agentmods" height="20"></a>

Per session 66 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 775 The whole file, excluding the scripts and references it only reads on demand.

Security scan B 1 finding. A grade says what 26 rules found in the file — not that it is safe.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5.1	$0.00066	$0.00775
Opus 5	$0.00033	$0.00387
Sonnet 5	$0.00013	$0.00155
Haiku 4.5	$0.00007	$0.00077

Measured 7d ago against content hash 14d706876e58, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade B, and why

harness-tuning scanned grade B with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 7d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Reads agent configuration directoriesmediumAgent snooping

.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.

- `packs/codex/config.toml` projects to `~/.codex/config.toml`.

deprecated/codex-plugin/plugins/klimkit/skills/harness-tuning/SKILL.md · 43 lines

How it starts

The opening of the file, as written. The whole thing — 43 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Harness Tuning

Tune Klimkit's repo-managed Codex harness from the source pack. Do not edit generated files under ~/.codex/; those are overwritten by kk apply, kk pull, and daemon sync.

Ground Rules

Work from the Klimkit repo root.
Edit source pack files under packs/codex/.
Keep pack files hand-authored and reviewable. Avoid generated bulk output unless the user explicitly asks for a generated artifact.
After pack edits, run the relevant pack validation tests and then kk apply if the current VM should use the change immediately.
Push committed pack changes to main only when the user wants the update published. Machines with autosync enabled apply main; autosync is opt-in.

Source Map

packs/codex/AGENTS.md projects to ~/.codex/AGENTS.md.
packs/codex/config.toml projects to ~/.codex/config.toml.
packs/codex/agents/ projects to ~/.codex/agents/.
packs/codex/skills/ projects to ~/.codex/skills/.
packs/codex/hooks/ projects to ~/.codex/hooks/.

Workflow

Read the existing pack file that owns the behavior.
For implementation work, create or update the task acceptance checklist before editing source-pack files.
When integrating external advice, synthesize it into the existing pack instead of pasting template blocks; check AGENTS.md, subagents, skills, and tests for duplicate or conflicting guidance.
For UI-impacting workflow changes, require proof reports under the configured writable reports directory (solo .klimkit/reports/, team .klimkit/<operator>/reports/) with Git-tracked HTML plus screenshot/video evidence, display media as full-width sections, prefer MP4 report videos for Chrome/PWA scrubbing, and include the Tailscale-served report URL in the final handoff when available.
Make the smallest source-pack edit that solves the request.
Run uv run python -m unittest tests.test_codex_pack_validation -q.
Run broader tests when hooks, config parsing, or Klimkit install/apply behavior changed.
Run kk preview for machine-affecting projection changes when useful.
Run kk apply to make changes live on the current VM when the user wants the current VM updated now.
Before final reviewers on non-trivial implementation work, run the Reflection Gate: append a full-timestamped cross-task Reflection Log session to the configured writable reflection file (solo .klimkit/reflection.md, team .klimkit/<operator>/reflection.md), using Observations, Derived Pattern, Insight, and Next Probe by default with up to ten named sections when useful; preserve older reflection formats by appending a normalized entry when relevant; reconsider the result; and rerun impacted checks if reflection exposes a gap.
Commit and push when the change should autosync to other Klimkit machines.

Read the full file on GitHub · 43 lines

Files

What ships with it

1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

agents/openai.yaml 187 B

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

7d ago First seen · 43 lines · 66 tokens per session scan B 14d706876e58

Subscribe to this mod's changes

harness-tuning is a skill published in the GitHub repository klimentij/klimkit (7 stars, last pushed 27d ago), licensed MIT. It adds 66 tokens to every session and 775 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Building agents agent-workflows agentic-engineering ai-agents automation cli code-server

LinkedIn automation via the Linked API CLI - fetch profiles, search people and companies, send messages, manage connections, create posts, react, comment, and run Sales Navigator and custom workflows. Use when the user wants to interact with LinkedIn.

bobmatnyc/claude-mpm-skills · 51 tokens

harness-tuning

Getting it into your agent

Wrote this? Show the measurements

What it costs to keep this loaded

Grade B, and why

Reads agent configuration directoriesmediumAgent snooping

How it starts

Harness Tuning

Ground Rules

Source Map

Workflow

What ships with it

What this file has done since we first saw it

Other skills, from other repositories

linkedin

gws-modelarmor

persona-event-coordinator

persona-team-lead

persona-researcher

recipe-plan-weekly-schedule