pluginhub CLAUDE.md

Repository instructions for pluginhub, a marketplace containing add-ons for several coding-agent command-line tools.

In plain words
What is it for?
Use them when working on plugin manifests, skills, scripts, tests, changelogs, or adaptations for Claude Code, CodeBuddy, Codex, and Kimi.
Why use it?
They explain the repository structure and conventions needed to modify plugins without breaking support for its different hosts.

Instructions file

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.

agentmods
npx agentmods add instructions/qxbyte/pluginhub/claude-md
Clone the repo
git clone --depth 1 https://github.com/qxbyte/pluginhub
Per session 7,546 This file is loaded in full into every session.
When invoked 7,546 The same file — it is already loaded in full.
Security scan B 2 findings. Scan, not verified.
Origin original No closer match found 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.

ModelPer sessionOnce invoked
Fable 5 $0.07546 $0.07546
Opus 5 $0.03773 $0.03773
Sonnet 5 $0.01509 $0.01509
Haiku 4.5 $0.00755 $0.00755

Measured yesterday against content hash 223071f3bbb2, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade B, and why

pluginhub CLAUDE.md scanned grade B with 2 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 yesterday.

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.

Reads agent configuration directoriesmediumAgent snooping

.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.

- **Claude Code and CodeBuddy are supported/verified. Codex install + skills are verified on a real host** (subdirectory install via `codex plugin marketplace add` + `codex plugin add` works; the `SessionStart` hook advi

Runs shell commandslowCapability

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

- Scripts are CLIs, not importable modules. Tests invoke them via `subprocess.run` through the `run_script` fixture in `tests/conftest.py`.
CLAUDE.md · 180 lines

How it starts

The opening of the file, as written. The whole thing — 180 lines — stays where its author put it; the contents beside it link to each section on GitHub.

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

What this repo is

A four-plugin marketplace for Claude Code / CodeBuddy / Codex / Kimi CLIs. The root is only the marketplace shell; each plugin under plugins/<name>/ is self-contained with its own plugin.json, skills, scripts, tests, and CHANGELOG. Each plugin ships four independent host manifests (.claude-plugin/ / .codebuddy-plugin/ / .codex-plugin/ / .kimi-plugin/) — see §Multi-host adaptation below; Claude Code / CodeBuddy verified, Codex verified for install + skills (SessionStart hook advisory unconfirmed), Kimi verified for local install.

  • specode (plugins/specode/, v6.5.2) — a lightweight spec-driven workflow plugin. It is not a state machine: it is a thin orchestration shell (壳) that walks a host agent through a phase pipeline (requirements → design → tasks → 执行方式 → 执行 → 验收) and, at each phase, delegates the heavy lifting: requirements to its own intake skill, design/tasks to superpowers (brainstorming / writing-plans), and the whole execution tail (执行方式 selector → execution → acceptance) to its own execute skill (/specode:execute <slug>, also manually triggerable anytime — added 6.3.0), which in turn dispatches subagent-driven-development / executing-plans / task-swarm / verification-before-completion. When superpowers is absent, specode falls back to a specode-native path (first-class, not an afterthought). It produces 4 fixed documents (requirements.md / design.md / tasks.md / implementation-log.md) under the user's specs directory. This is the plugin this CLAUDE.md documents unless stated otherwise.
  • task-swarm (plugins/task-swarm/, v0.11.0) — a standalone implementation-phase orchestrator (multi-coder fork + reviewer/validator state machine), extracted out of specode in milestone M1. It has its own CLI (scripts/task_swarm/), state machine, agents, skill (skills/swarm/, user-invocable — provides /task-swarm:swarm directly, no commands/), and CHANGELOG. specode hands off to it only via the user-chosen "委托 task-swarm" option in the 执行方式 selector, with zero import (calls task-swarm's own /task-swarm:swarm skill).
  • obsidian-wiki (plugins/obsidian-wiki/, v2.0.0) — a skills-only plugin (no commands, no hooks) for maintaining an Obsidian LLM-Wiki: a deterministic structure layer (Home tree / per-dir READMEs / partition pages via wiki-struct), content curation (wiki-curate), and a unified orchestrator (wiki-orchestrate). Generic code + per-vault config in the home-dir registry ~/.config/obsidian-wiki/ (fallback: <vault>/.wiki/config.json), zero hardcoded structure. Largely independent of the other two — its only historical tie is that the old spec→knowledge distill capability was extracted out of it into specode's distill in its own v2.0.0.
  • ragkit (plugins/ragkit/, v0.2.0) — a standalone RAG knowledge-base retrieval plugin: vector + lexical + metadata three-channel recall, RRF-fused, returning pointer cards. It can consume specode distill's knowledge-base/ output downstream; zero heavy deps (stdlib + numpy for the lexical channel). Has hooks, so it ships per-host hook files like specode.

Read the full file on GitHub · 180 lines

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.

  1. yesterday First seen · 180 lines · 7,546 tokens per session scan B 223071f3bbb2

Subscribe to this mod's changes

pluginhub CLAUDE.md is an instructions file published in the GitHub repository qxbyte/pluginhub (3 stars, last pushed 27d ago), licensed MIT. It adds 7,546 tokens to every session, about $0.0377 per session on Opus 5. A static security scan graded it B with 2 findings (reads agent configuration directories, runs shell commands). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other instructions, from other repositories