Delegate
Skill chinayin/gox-claude-plugins/plugins/token-thrift/skills/delegate
Team coding-standards plugins for Claude Code (Agent Skills marketplace) · 面向 Claude Code 的团队代码规范插件(技能 marketplace)
npx -y skills add chinayin/gox-claude-plugins --skill delegateAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Token-saving subagent routing. Use when facing token-heavy, read-heavy, conclusion-only work (reading Lark/Feishu in full or writing to it, wide code searches, log/long-document analysis) to decide whether to offload the work to a cheaper-model subagent and which tier to use.
SKILL.md
1.8 KB, 378 tokens by cl100k_base, as published. Nobody here has run it
Token-saving three-tier split
The main agent (Opus) does only orchestration and decisions with the user. Offload token-heavy work; keep raw material out of the main context.
Which tier
| Kind of work | Dispatch to | Model | Why |
|---|---|---|---|
| Read / compare / verify / search (conclusion only) | cheap-reader | Haiku | Cheapest ($1/$5); raw material is read once and discarded |
| Correctness-sensitive writes (Lark/Feishu XML/blocks, reliable persistence) | careful-writer | Sonnet | Write errors are costly; Sonnet is only 40–60% of Opus |
| Orchestration / decisions | main agent | Opus | Keep the main context lean |
A read that needs subtle judgment may be promoted to Sonnet; a purely mechanical append may drop to Haiku.
When it's worth offloading
Rule of thumb: offload when the throwaway material the subagent must process is ≳ 3,000 tokens. Because:
- Skill loading (e.g.
lark-*SKILL + references, often tens of thousands of tokens) moves from Opus pricing to the cheaper model's pricing. - Raw material stays out of the main context, avoiding it being re-billed on every subsequent turn (the biggest lever).
If the task is tiny, or the main thread will need the raw material again, keep it inline instead.
How to dispatch
Use the Task/Agent tool with subagent_type set to cheap-reader or careful-writer. Have it return only a compact conclusion; if the result is large, have it write a scratchpad file and return the path. Keep the Opus main context lean.
Gives 0 of the 12 instructions most docs writing skills give in 378 tokens
Counted across 1,637 of the 3,044 authors here whose files we hold, read 2026-08-06
- announce the skill at startin 54 of 1637, across 21 files
- convert legacy doc files before editingin 45 of 1637, across 7 files
- predict questions readers might askin 42 of 1637, across 3 files
- Generate clarifying questions for initial contextin 42 of 1637, across 3 files
- Create document scaffold with placeholder textin 42 of 1637, across 3 files
- Brainstorm content options for each sectionin 42 of 1637, across 3 files
- Test document with fresh context-less instancein 42 of 1637, across 3 files
- ask interview questions one at a timein 42 of 1637, across 26 files
- include exact file paths in every taskin 42 of 1637, across 15 files
- Apply surgical edits during refinementin 41 of 1637, across 2 files
- Offer structured workflow or freeformin 40 of 1637, across 1 file
- Ask for document meta-contextin 40 of 1637, across 1 file
Said here and by no other author read
- offload token-heavy work to cheaper subagents
- keep raw material out of the main context
- dispatch conclusion-only reads to cheap-reader
- dispatch correctness-sensitive writes to careful-writer
- offload tasks processing over 3000 tokens
- keep tiny tasks inline
Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.