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.
npx agentmods add skills/underpass-ai/kmp/kmp-setupnpx skills add underpass-ai/kmp --skill kmp-setupgit clone --depth 1 https://github.com/underpass-ai/kmpWrote 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.
[](https://agentmods.dev/skills/underpass-ai/kmp/kmp-setup)<a href="https://agentmods.dev/skills/underpass-ai/kmp/kmp-setup"><img src="https://agentmods.dev/badge/skills/underpass-ai/kmp/kmp-setup.svg" alt="Measured on agentmods" height="20"></a>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 | $0.00033 | $0.00785 |
| Opus 5 | $0.00016 | $0.00392 |
| Sonnet 5 | $0.00007 | $0.00157 |
| Haiku 4.5 | $0.00003 | $0.00078 |
Grade A, and why
kmp-setup scanned grade A with 0 findings 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 today.
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.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
KMP setup
Resolve the plugin root as two directories above this SKILL.md. Diagnose
first with <plugin-root>/scripts/kmp-doctor.sh; it is a thin adapter into
kmp-mcp doctor, whose Rust lifecycle diagnosis owns host inventory, engine
proof, effective wiring and plugin-tree parity.
When a session notice or version comparison says a newer release exists, run
<plugin-root>/scripts/kmp-update.sh. That inventories both native hosts,
updates every installed KMP plugin, and installs the checksummed engine from
the same release. Do not update only one host or one half.
Once the release is proved and every host points at it, the update removes the
superseded versions each host's plugin cache had kept, leaving the installed
release and the one before it for rollback. The receipt names what went under
plugin_caches; report it rather than letting sixty megabytes disappear
quietly.
For an enabled Codex plugin, the plugin owns MCP. Install or update the engine
with <plugin-root>/scripts/kmp-install-binary.sh, but do not add a global
mcp_servers.kmp registration, copied prompts, or an AGENTS snippet. If a
global registration also exists, report the collision and remove it only as
the explicit ownership repair for the requested setup.
Standalone Codex wiring is retired. The native plugin is the single MCP owner;
reject --standalone and diagnose any old global registration as a collision.
Show the active agent policy with kmp-mcp config. It reports two settings.
memory routing decides whether an agent enters KMP unasked. The default is
on request: memory is called when the user asks for KMP, when a kmp skill or
command runs, or when the project opts in. Change it only when the user asks
for always-on recall, with kmp-mcp config memory-routing always;
on-request returns to the default. Never turn it on as part of an install.
There is no language setting to configure. A semantic question is asked in
English with the user's own words passed as asked_as, and the answer is
given in the user's language with stored evidence, refs, relation why and
source metadata preserved byte-for-byte. kmp-mcp doctor and kmp-mcp info
name the lexical-bridge table a store would read, or say there is none and
what that means for Ask; report that line when the user asks about
languages. An older config file may still carry an ask_fallback_languages
line; it is ignored and the doctor says so. Questions in Chinese, Japanese or
Thai are not segmented by word yet; storage remains byte-exact. Upgrades must
leave the policy file intact.
Setup and update must not select or write a memory store. Guide sync is a separate, explicit data operation. Run it only when the user asks to install or refresh the guides:
KMP_MCP_BIN=<path-to-kmp-mcp> <plugin-root>/scripts/kmp-guide-sync.sh sync
It writes guide:kmp-agent and guide:kmp into the store selected by that
command. In a project, that is project memory and its commit-native bundle.
Name this effect before running it. Exact reruns add no events; a later plugin
version supersedes only changed guide entries.
Finish by rerunning <plugin-root>/scripts/kmp-doctor.sh. A running Codex
session needs one restart to load changed skills or MCP wiring.
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.
- today Changed · +1 lines b4eff70ca363
- yesterday Changed · +8 lines aa4cbe002a14
- 4d ago First seen · 58 lines · 33 tokens per session scan A 02037a226284
kmp-setup is a skill published in the GitHub repository underpass-ai/kmp (0 stars, last pushed today), licensed Apache-2.0. It adds 33 tokens to every session and 785 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
i18n-localization
Internationalization and localization patterns. Detecting hardcoded strings, managing translations, locale files, RTL support.
chinese-documentation
中文文档排版参考——中英文空格、全半角标点、术语保留、链接格式、中文文案排版指北约定。仅在用户显式 /chinese-documentation 时调用，不要根据上下文自动触发。.
seedance-vocab-ja
This skill should be used when the user asks for Japanese Seedance 2.0 prompt wording, Japanese cinematic vocabulary, or translation of camera, lighting, action, VFX, audio, and production terms into Japanese.
harden
Improve interface resilience through better error handling, i18n support, text overflow handling, and edge case management. Makes interfaces robust and production-ready.
asc-subscription-localization
Bulk-localize subscription, subscription-group, and in-app purchase display names across App Store locales using asc, including API 4.4.1 version-scoped v2 resources. Use when filling or updating subscription/IAP names and descriptions without App Store Connect UI work.
intlayer-preact
Integrates Intlayer internationalization with Preact applications. Use when the user asks to "setup Preact i18n", create a new translated component, use the "useIntlayer" hook in Preact, or configure providers.