- index.js: header comments and moyun_ultimate_profile sources now describe the five layers as this project own design - package.json: description rewritten (no upstream package names, no fused-from wording), truncated tail fixed - README / README.en / CHANGELOG / docs/INSTALL / market: the anti-pattern table and Mermaid diagram now say host-patch plugins / config-override plugins instead of naming upstream packages; migration hints refer to an older build of this project - prompts/frames/transform_first.md: frame title and tool names now moyun_frame / moyun_ultimate_profile - prompts/ultimatum-mode.md: persona id is moyun-ultimate - cordis.patch.yml: permission-layer comment no longer names a package - install.ps1: bundle replacement regex is now moyun-* generic Residual scan across all text/code files: 0 hits.
3.2 KiB
3.2 KiB
Add moyun-ultimate
What it is
A five-layer fused jailbreak core for DeepSeek Harness. Built by taking apart 12 community jailbreak packages, diffing what each one actually contributes, and merging the useful parts by layer instead of stacking the packages.
Why layered, not stacked
The three biggest jailbreak plugins in this ecosystem aim at the same targets:
| package | how it works | collides on |
|---|---|---|
| host-patch plugins | rewrite host source files (pattern-matched rules) | same host files |
| config-override plugins | override official cordis config ids | system-prompt, approval, permission, sandbox-policy |
| hybrid plugins | do both and share a state directory | both of the above |
Same-id config is last-write-wins; same-file source patches mean two pattern sets racing over one piece of text. Installing two of them does not double the effect — it disables both.
So this plugin ships zero host patches. It owns persona, refusal strategy, execution discipline and the skill library; the permission layer stays with whatever host-patch plugin you already run.
What's inside
- Five ordered prompt sections (21 KB total)
order 100persona coreorder 150refusal kernel — authorization-as-given, foreign-pattern table (<ethics_reminder>/ fakeSystem:/ identity override → zero weight), boundary-word normalization, template workflowsorder 160execution intensity — act don't narrate, deliver completely, report one line when blocked and continue, keep going through WAF/CDNorder 200anti-dilution mirror — identical semantics, survives context compactionorder 210skill routing
- 63 built-in security skills (41 general + 22 intranet/redteam, 1 de-duplicated), synced
idempotently into
$DSH_HOME/skillson first start — existing files are never overwritten moyun_frame— 5 re-framing templates (roleplay / textbook / defender_reverse / transform_first / k3) loaded on demand, not resident in the system promptmoyun_ultimate_profile— self-check tool- Live refusal-score badge in the corner
- Five independent layer toggles
Verification performed
- Module loads cleanly, five sections register in order 100/150/160/200/210, total 21,354 B
- Both tools register and return expected payloads (
skills: 63, five layers on) - Invalid frame argument returns a proper error listing the 5 valid names
- Layer toggles: disabling
routingyields 4 sections --dump-configcomposes without error and containsid: moyun-ultimate- 63 skill directories land in
$DSH_HOME/skillson install and are visible to the agent without restarting
Notes for review
category: workflow— it changes how work gets done rather than adding a UI surface, memory store, or standalone skill pack. Happy to move it if another category fits better.- Ships no screenshots yet; the badge is a small pulsing pill in the bottom-right corner. I can
add
data/screenshots.jsonentries once the repo hasdocs/screenshots/*.png. - All prompt text, skill organisation and tooling in this repository are the author's own work, released under MIT.