Borrowing it
Nothing to install: this file belongs to liuyingxuvka/Khaos-Brain. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/liuyingxuvka/Khaos-Brain/main/.agents/skills/khaos-brain-update/SKILL.mdgit clone --depth 1 https://github.com/liuyingxuvka/Khaos-BrainWrote 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/liuyingxuvka/khaos-brain/khaos-brain-update)<a href="https://agentmods.dev/skills/liuyingxuvka/khaos-brain/khaos-brain-update"><img src="https://agentmods.dev/badge/skills/liuyingxuvka/khaos-brain/khaos-brain-update/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.
<a href="https://agentmods.dev/skills/liuyingxuvka/khaos-brain/khaos-brain-update"><img src="https://agentmods.dev/badge/skills/liuyingxuvka/khaos-brain/khaos-brain-update.svg" alt="Reviewed on agentmods" width="80" 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 5.1 | $0.00047 | $0.01034 |
| Opus 5 | $0.00023 | $0.00517 |
| Sonnet 5 | $0.00009 | $0.00207 |
| Haiku 4.5 | $0.00005 | $0.00103 |
Grade A, and why
khaos-brain-update 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 12d 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.
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.
How it starts
The opening of the file, as written. The whole thing — 50 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Khaos Brain Update
This is the sole manual software-update route. Use it only for an explicit user request in the current conversation. It updates repository software and installer-managed Codex integration; it does not maintain cards and it does not replace Sleep or Dream.
Use when
- The user directly asks AI in this conversation to update Khaos Brain.
- The configured Git upstream is available and the requested action is a safe fast-forward update.
Do not use when
- The desktop UI merely reports that an upstream version is available.
- A user clicks or hovers the read-only version status.
- A scheduled task, installer check, background retry, startup path, aggregate assurance run, or persisted file tries to trigger an update.
- The current conversation contains no explicit update request.
There is no scheduled automation, no persisted authorization, no prepared state, no compatibility alias, and no fallback update entrypoint. The exact retired khaos-brain-system-update task must remain absent.
Recovery boundary
The update must still work when retrieval, desktop settings, organization state, or the UI is unhealthy. Preserve .local/, private/history/candidate/outbox KB state, organization caches and Skill bundles, exchange ledgers, user-created untracked files, and user-owned Codex configuration outside managed paths.
Apply contract
- Run only
python scripts/run_khaos_brain_manual_update.py --explicit-user-request --json. This target-owned entrypoint rejects before run-state creation, Git mutation, installer execution, or automation-state mutation when current-request authorization is absent. - The manual check fetches and compares only the exact configured upstream.
no-updateis the only successful terminal no-op. Missing upstream, fetch failure, dirty tracked work, local-ahead or diverged topology, an open Khaos Brain UI, a previous failed update, concurrent execution, or any unknown state remains blocked and unfinished. - Use Git fast-forward only. Never reset, force checkout, rebase, move tags, overwrite dirty user/peer work, or guess another remote branch.
- Snapshot the exact two surviving composite automation statuses and independent
user_pausedvalues, then pause those two survivors before mutation. Retired split/update tasks are never restored. - Run the transactional installer through
python scripts/install_codex_kb.py --json. It must run the versioned maintenance migration, settle old lifecycle debt, publish direct-to-current LogicGuard authority, prove zero retired authority residuals, retire exactkb-architect-pass,kb-architect, andkhaos-brain-system-updatemanaged surfaces, and preserve similarly named user assets. Every expensive assurance owner binds exact source, data, toolchain, environment, and installed-projection inputs; unchanged owners reuse one immutable terminal-success receipt and only affected owners execute. Late input drift replans only its mapped owners and never triggers an unconditional second aggregate campaign. Upgrade-attempt currentness is one bounded hash-boundHEAD.jsonplus its exact current projection; an installation check must never enumerate attempt history, use an install-manifest attempt as fallback authority, or launch migration, model, retrieval, pytest, resume, or assurance subprocesses. - Keep both survivors paused while the target-owned entrypoint builds the exact restoration plan and performs the deferred installation check.
- Only after every target-owned hard gate passes, apply that exact plan, read back both statuses and
user_pausedvalues, run the zero-execution read-only installation currentness check, then mark CURRENT. - Any native, migration, installation, restoration, readback, or final-check failure marks FAILED and must keep surviving automations paused. A timeout is incomplete until the complete owned process tree is terminated and zero descendants are confirmed.
What ships with it
4 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.
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 · 50 lines · 47 tokens per session scan A fec7407ce034
khaos-brain-update is a skill published in the GitHub repository liuyingxuvka/Khaos-Brain (37 stars, last pushed 1mo ago), licensed MIT. It adds 47 tokens to every session and 1,034 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 2026-08-30.
Other skills, from other repositories
dws
A command-line guide for DingTalk, a workplace platform with chat, documents, calendars, approvals, tasks, and other business tools.
tmeet-skill
A command-line tool for Tencent Meeting, a video-conferencing service. It covers sign-in, meetings, recordings, transcripts, reports, invitations, and controls during a meeting.
tencent-docs
A skill for creating, reading, editing, searching, and managing Tencent Docs, an online document and spreadsheet service.
lark-shared
A shared set of rules for using lark-cli, a command-line tool for accessing Lark, the collaboration platform also known as Feishu. It covers setup, login, identity selection, permissions, JSON output, and authorization errors.
government-writing
A skill for writing formal Chinese government documents, including notices and opinions, in the required official format.
package-author
A packaging guide for turning scripts, skills, or an MCP service into a standard plugin package. MCP is a way for an AI assistant to connect to external tools.