pordee (พอดี) plugin
pordee
A Claude Code plugin that makes the agent answer in Thai briefly: no polite particles, hedging or greetings
- 330 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The plugin registers hooks for session start and for every message that run Node.js scripts
- The install script run through curl sets the plugin up in several agents on the machine
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/pordeeDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The plugin registers hooks for session start and for every message that run Node.js scripts
- The install script run through curl sets the plugin up in several agents on the machine
This is third-party code. Review the repository files before installing.
What it does
pordee is a Claude Code plugin, the Thai take on the Caveman idea: the agent answers briefly and spends fewer output tokens. The plugin strips polite particles (ครับ, ค่ะ, นะคะ), hedges such as "maybe" and "probably", greetings and filler connectors. The author promises a 60–75% cut in output tokens with the same content. There are two modes. In lite the agent writes full grammatical sentences without politeness or padding, like a business letter. In full it also drops function words and the nominalizing prefixes "การ-" and "ความ-" and picks short synonyms. Phrases may then be fragments following the pattern object, action, reason, next step. The mode is switched on with a slash command or a Thai keyword on its own line. Inside are two skills, pordee and pordee-stats, and two hooks: one loads the mode at session start, the other catches keywords in messages.
Who it is for. For Thai-speaking developers who talk to their agent in Thai and want to save tokens and reading time.
Good fit when
- You use Claude Code in Thai and replies come with long polite phrasing
- You want fewer output tokens in long debugging and review sessions
- You want a short business style without losing grammar: use lite
- You want the shortest format made of fragments: use full
Not a fit when
- You talk to the agent in Russian or English: the plugin is tuned for Thai
- The agent writes customer-facing text where a polite register is required
- You want to cut input tokens: the plugin compresses only the agent's output
- You need detailed explanations for beginners: in full mode a reply reads like notes
Example request
/pordee lite, then: why does this React component re-render on every render?Limitations
It works for Thai only. The token saving depends on the question: on the author's five examples lite shortens a reply by 44–60% and full by 73–84%. The install.sh or install.ps1 script finds the agents on your machine and installs the plugin into several at once, so read it before running. The hooks run on Node.js, which must be installed. Install commands for individual agents are collected in INSTALL.md. The plugin's behavior in Cursor, Windsurf, Cline and Codex was not tested for this card.
How to disable. For now: send /pordee stop or the phrase "หยุดพอดี". For good: remove the plugin or skill with the tool you installed it with and delete the pordee hooks from your settings.
FAQ
How do I turn on pordee mode in Claude Code?
Send /pordee to enable the default full mode, or /pordee lite for the gentler one. Turn it off with /pordee stop. Without slash commands, a line with only "พอดี" turns it on and "พูดปกติ" turns it off; the word must be the whole message.
How does lite differ from full?
Lite removes politeness, hedging and greetings but keeps full grammar. Full also drops function words and nominalizing prefixes and allows fragments.
How do I see token statistics?
The /pordee-stats command shows usage in the current session. With the --share flag it prints a one-line summary that is easy to copy.
Does pordee support Cursor or Codex?
The `npx skills add kerlos/pordee` command installs it into agents that support the skills registry, including Cursor, Windsurf, Cline and Codex. In a session you turn it on with /pordee, and per-agent commands are in INSTALL.md.
What does the plugin do with technical terms?
It keeps them in English. In the author's examples terms like useMemo, re-render and connection pool stay in English inside the Thai text.
Discussion
No questions or reviews yet.
Related
PluginsA plugin that measures language complexity in coding-agent replies and sends an overly abstract reply back for a rewrite
A Claude Code plugin that picks the model's effort level for every prompt and shows the prompt cache countdown
A Claude Code plugin that catches your corrections in chat and, after review, writes them into CLAUDE.md and AGENTS.md
A Claude Code output-style that makes the model write clear, natural Korean instead of broken machine Korean
A marketplace of fourteen plugins for Claude Code power users: model council, email, calendar, video digests and routine tasks
A marketplace of 17 Claude Code mods: token-spend breakdown, a rollback edit timeline, assumption tracking, a verification board and a live command feed
Comparison with similar tools
Entries that share the main search query, and how they differ.
README in short
The project's motto asks why use many words when a short answer will do. The core of the description is five before-and-after replies in Thai with estimated token savings.