What is kimi-supervision?

A set of scripts for having the local Kimi Code command-line tool do work while another agent watches and checks the results. It keeps Kimi’s session files and can recover answers when its normal output is incomplete.

Use it to start or continue Kimi sessions, supervise progress, inspect saved reports and artifacts, and independently verify the final result.

It reduces the risk of trusting a delegated task without seeing how it was done or checking its intermediate work. It also helps preserve and continue native Kimi sessions.

kimi-supervision — Skill for Codex by ltczding-gif

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 ltczding-gif/agent-supervision-skills --skill kimi-supervision

Clone the repo

git clone --depth 1 https://github.com/ltczding-gif/agent-supervision-skills

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/ltczding-gif/agent-supervision-skills/kimi-supervision/github.svg)](https://agentmods.dev/skills/ltczding-gif/agent-supervision-skills/kimi-supervision)

Your own site

<a href="https://agentmods.dev/skills/ltczding-gif/agent-supervision-skills/kimi-supervision"><img src="https://agentmods.dev/badge/skills/ltczding-gif/agent-supervision-skills/kimi-supervision/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/skills/ltczding-gif/agent-supervision-skills/kimi-supervision"><img src="https://agentmods.dev/badge/skills/ltczding-gif/agent-supervision-skills/kimi-supervision.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 51 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 7,352 The whole file, excluding the scripts and references it only reads on demand.

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

Origin 100% copy Near-identical to another mod 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	$0.00051	$0.07352
Opus 5	$0.00026	$0.03676
Sonnet 5	$0.00010	$0.01470
Haiku 4.5	$0.00005	$0.00735

Measured 12d ago against content hash 261f772aa3ec, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-12, from the pricing page.

Security

Grade A, and why

kimi-supervision scanned grade A 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 12d ago.

The scan reads SKILL.md. This mod also ships 5 executable files (scripts/common.ps1, scripts/kimi-chat.ps1, scripts/kimi-run-once.ps1, …), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

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.

Runs shell commandslowCapability

Expected in a hook, worth knowing in a rule or an instructions file.

### Anti-pattern: bypassing the supervision scripts via direct `subprocess.run`

Origin

This is a copy

100% identical to kimi-supervision — 0 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.

kimi-supervision/SKILL.md · 854 lines

How it starts

The opening of the file, as written. The whole thing — 854 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Kimi Supervision

Use this skill when you want Codex to make local Kimi Code do work, watch what happened, inspect its reports, and then independently verify the outcome.

This skill is for supervision, not blind delegation.

What this skill provides

This skill ships self-contained PowerShell scripts:

scripts/common.ps1
scripts/kimi-run-once.ps1
scripts/kimi-chat.ps1
scripts/kimi-view-session.ps1
scripts/test-kimi-supervision.ps1

The optional agents/openai.yaml file is metadata for agent environments that support skill display names, descriptions, and implicit invocation policy. The PowerShell scripts do not require it.

These scripts:

require a locally installed kimi CLI
do not depend on any hard-coded machine-specific bridge directory
prefer Kimi's native session memory
start Kimi with thinking mode enabled by default unless you pass -NoThinking
treat think-only or tool-call-only output as intermediate, not as a finished answer
keep local session artifacts for later inspection
recover answers from Kimi native files when kimi --print returns empty or incomplete stdout
default to a long native recovery window so Kimi is not judged too early

State layout

Default state root:

%LOCALAPPDATA%\kimi-supervision\

Override:

KIMI_SUPERVISION_HOME

Fallbacks if environment variables are missing:

%USERPROFILE%\.kimi-supervision
C:\kimi-supervision

Per session, this skill stores:

sessions/<name>/transcript.jsonl
sessions/<name>/last-prompt.txt
sessions/<name>/last-response.jsonl
sessions/<name>/last-response.txt
sessions/<name>/session.json

Core design

This skill uses:

Native Kimi session first
Native file recovery second
Transcript replay fallback last

Meaning:

kimi-chat.ps1 creates or reuses a real Kimi --session id
follow-up calls reuse that same native Kimi session id
if flattened stdout is empty or incomplete, the script reads Kimi's native context.jsonl
if flattened stdout contains only [THINK] or [TOOL_CALL] lines, the script still keeps waiting for final answer text
if native execution or native resume is unavailable, the script can fall back to transcript replay mode

Read the full file on GitHub · 854 lines

Files

What ships with it

6 files 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.

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.

12d ago First seen · 854 lines · 51 tokens per session scan A 261f772aa3ec

Subscribe to this mod's changes

kimi-supervision is a skill published in the GitHub repository ltczding-gif/agent-supervision-skills (2 stars, last pushed 3mo ago), licensed MIT. It adds 51 tokens to every session and 7,352 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 1 finding (runs shell commands). It is 100% identical to kimi-supervision, differing in 0 lines, and is treated as a copy.

Building agents Claude Code agent-supervision agent-tools ai-agents codex

vs-search

Search runtime and scene management: verify queries, inspect scenes, debug app readiness, and diagnose recall or scene-config issues.

volcengine/SearchCLI · 27 tokens

vs-shared

Shared SearchCLI setup: install, authenticate, run doctor, and verify the local environment.

volcengine/SearchCLI · 22 tokens

linear-graphql-skill

Operate Linear workspace issues, projects, and teams through Linear GraphQL API using UXC. Use when tasks require querying or creating issues, managing projects, or interacting with Linear workflow. Supports both Personal API Key and OAuth authentication.

holon-run/uxc · 51 tokens

kimi-supervision