docs: sync KB — claude-plugins-marketplace, permission-manager-classifier-development
All checks were successful
Reindex Knowledge Base / reindex (push) Successful in 5s

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
Cal Corum 2026-03-18 23:23:00 -05:00
parent fa11cd0431
commit b192b3ca47
2 changed files with 268 additions and 0 deletions

View File

@ -0,0 +1,130 @@
---
title: "Claude Plugins Personal Marketplace"
description: "Reference for the cal/claude-plugins Gitea repo — personal Claude Code plugin marketplace with 20 plugins (10 agents, 10 skills), including installation, adding new plugins, and maintenance."
type: reference
domain: development
tags: [claude-code, plugins, gitea, marketplace]
---
# Claude Plugins Personal Marketplace
**Repo:** `cal/claude-plugins` on Gitea ([https://git.manticorum.com/cal/claude-plugins](https://git.manticorum.com/cal/claude-plugins))
**Local clone:** `/mnt/NV2/Development/claude-plugins/`
**License:** AGPL-3.0
## What It Is
A self-hosted Claude Code plugin marketplace containing 20 plugins (10 agents, 10 skills) extracted from local `~/.claude/` definitions. Registered via `claude plugin marketplace add https://git.manticorum.com/cal/claude-plugins.git` (stored in `~/.claude/settings.json` under `extraKnownMarketplaces`).
## Plugin Inventory
### Agents (10)
| Name | Description |
|------|-------------|
| architect | PRD creation, system design, technical specs |
| claude-researcher | Web research with multi-query decomposition |
| designer | UX/UI design review, visual design, accessibility |
| engineer | Code implementation, debugging, optimization, testing |
| issue-worker | Autonomous Gitea issue fixer — creates PRs |
| pentester | Penetration testing and vulnerability assessments |
| pr-reviewer | Automated Gitea PR reviewer |
| swarm-coder | Implementation agent for orchestrated swarms |
| swarm-reviewer | Code reviewer for orchestrated swarms |
| swarm-validator | Spec validator for orchestrated swarms |
### Skills (10)
| Name | Description |
|------|-------------|
| backlog | Surface next Gitea issue or TODO to work on |
| create-scheduled-task | Manage headless Claude scheduled tasks on systemd timers |
| json-pretty | JSON prettifier CLI tool |
| optimise-claude | Guide for writing effective CLAUDE.md files |
| playwright-cli | Browser automation for testing and screenshots |
| project-plan | Generate PROJECT_PLAN.json for task tracking |
| resume-tailoring | Tailored resume generation with company research |
| save-doc | Save documentation to KB with proper frontmatter |
| youtube-transcriber | Transcribe YouTube videos via OpenAI GPT-4o-transcribe |
| z-image | Local GPU image generation with Z-Image Turbo |
## Installation
```bash
# Marketplace is already registered. To install a plugin:
claude plugin install <name>@cal-claude-plugins --scope user
# Enable/disable:
claude plugin enable <name>@cal-claude-plugins
claude plugin disable <name>@cal-claude-plugins
# Update marketplace cache (pulls latest from Gitea):
claude plugin marketplace update cal-claude-plugins
```
## Adding a New Plugin
### Directory Structure
Each plugin lives under `plugins/<name>/` with this layout:
```
plugins/<name>/
.claude-plugin/
plugin.json # name, description, version
skills/<name>/ # for skill plugins
SKILL.md # skill definition with frontmatter
agents/ # for agent plugins
<name>.md # agent definition
```
### plugin.json
```json
{
"name": "my-plugin",
"description": "One-line description of what it does.",
"version": "1.0.0"
}
```
### SKILL.md Frontmatter
```yaml
---
allowed-tools: Read,Write,Edit,Glob,Grep,Bash
description: What this skill does
user-invocable: true
---
```
### Register in marketplace.json
Add an entry to `.claude-plugin/marketplace.json` in the `plugins` array:
```json
{
"name": "my-plugin",
"source": "./plugins/my-plugin",
"description": "One-line description."
}
```
### Commit and Push
```bash
cd /mnt/NV2/Development/claude-plugins
git add plugins/my-plugin .claude-plugin/marketplace.json
git commit -m "feat: add my-plugin skill"
git push origin main
```
Then update the cache: `claude plugin marketplace update cal-claude-plugins`
## Important Notes
- **SKILL.md `name:` must match directory name** — a mismatch causes the skill to not load (e.g., the `optimise-claude` plugin had `name: claude-optimised` which broke it)
- **SSH remote** — uses `git@git.manticorum.com:cal/claude-plugins.git` via SSH config alias (`~/.ssh/config` maps `git.manticorum.com``10.10.0.225` as user `git`)
- **No secrets** — repo is public; plugins are just prompt definitions, no credentials
- **Plugins requiring MCP servers**`backlog`, `issue-worker`, and `pr-reviewer` require `gitea-mcp`; they install fine without it but those tool calls won't execute
- **sync-kb does NOT apply here** — this is a standalone repo, not part of claude-home. Changes require manual commit/push

View File

@ -0,0 +1,138 @@
---
title: "Writing Classifiers for permission-manager (agent-toolkit)"
description: "Guide to adding new command classifiers to the St0nefish/agent-toolkit permission-manager plugin for Claude Code, covering project structure, conventions, testing, and PR workflow."
type: guide
domain: development
tags: [claude-code, permissions, agent-toolkit, bash, classifier]
---
# Writing Classifiers for permission-manager
## Overview
The `permission-manager@agent-toolkit` plugin (St0nefish/agent-toolkit) provides a `cmd-gate` PreToolUse hook that classifies Bash commands before execution. Classifiers are bash scripts in `plugins-claude/permission-manager/scripts/classifiers/` that decide whether commands should be auto-allowed, require user approval, or be denied.
## Project Structure
```
plugins-claude/permission-manager/
scripts/
cmd-gate.sh # Hook entry point
lib-classify.sh # Classification framework + dispatcher
classifiers/
cargo.sh # cargo/rust
docker.sh # docker/compose
find.sh # find (deny -delete/-exec rm)
git.sh # git (with protected branch logic)
gh.sh # GitHub CLI
gradle.sh # gradle/gradlew
jvm-tools.sh # java/mvn
npm.sh # npm/node/pnpm/yarn/npx
pip.sh # pip/python/poetry/pyenv
read-only-tools.sh # cat, ls, grep, jq, ps, etc.
tea.sh # Gitea CLI
uv.sh # uv/uvx (added 2026-03-18)
tests/permission-manager/
test-classify.sh # Main test harness (618+ tests)
```
## Classification Decisions
Three possible outcomes:
- **`allow`** — auto-approve, no prompt (read-only or safe local operations)
- **`ask`** — prompt user for approval (destructive, remote, or elevated operations)
- **`deny`** — block the command (e.g., `find -delete`)
- **passthrough**`return 0` without calling allow/ask/deny; defers to Claude Code's built-in permission system
## Classifier Conventions
### File Header
```bash
# shellcheck shell=bash
# shellcheck source=../lib-classify.sh
```
### Function Naming
- Entry point: `check_<tool>()` (e.g., `check_uv`, `check_docker`)
- Subcommand extractor: `extract_<tool>_subcommand()` (for tools with global flags)
### Entry Guard Pattern
Simple tools (cargo, npm, pip) use `awk` + `case`:
```bash
local first_token
first_token=$(echo "$command" | awk '{print $1}')
case "$first_token" in
uv) ;;
*) return 0 ;;
esac
```
Complex tools (docker, git) use `perl` guard:
```bash
echo "$command" | perl -ne '$f=1,last if /^\s*docker(\s|$)/; END{exit !$f}' || return 0
```
### Reason Message Style
- Read-only: `"<tool> <subcmd> is read-only"`
- Local ops: `"<tool> <subcmd> is a local build/dev operation"`
- Ask: `"<tool> <subcmd> modifies <what>"`
- Deny: `"<tool> <subcmd> is not allowed"`
### Classification Philosophy (from existing classifiers)
| Category | Decision | Examples |
|----------|----------|---------|
| Read-only inspection | allow | `git status`, `docker ps`, `pip list` |
| Local build/install | allow | `npm install`, `pip install`, `cargo build` |
| Global tool install | ask | `cargo install`, `uv tool install` |
| Package uninstall | ask | `pip uninstall`, `uv pip uninstall` |
| Publish/deploy | ask | `npm publish`, `cargo publish`, `uv publish` |
| Execute arbitrary packages | passthrough | `npx`, `uvx`, `uv run --with` |
| Destructive operations | deny | `find -delete` |
## Wiring a New Classifier
In `lib-classify.sh`, add to `classify_single_command()`:
```bash
check_<tool>
[[ "$CLASSIFY_MATCHED" -eq 1 ]] && return 0
```
Place it logically near related classifiers (e.g., `check_uv` after `check_pip`).
## Testing
### Test Format
```bash
run_test_both <expected> "<command>" ["label"]
```
`run_test_both` runs the test in both Claude and Copilot modes. Expected values: `allow`, `ask`, `deny`, `none` (passthrough).
### Test Section Structure
```bash
# ===== ALLOW: <tool> read-only =====
echo "── <tool> read-only ──"
run_test_both allow "<command>"
# ===== ALLOW: <tool> local build/dev =====
echo "── <tool> local build/dev ──"
...
```
### Running Tests
```bash
bash tests/permission-manager/test-classify.sh # Full suite
bash tests/permission-manager/test-classify.sh uv # Filter by keyword
```
## PR Workflow
1. Fork `St0nefish/agent-toolkit` on GitHub
2. Branch from `master` (not `main`)
3. Add classifier, wire it, update tests
4. Run full test suite — zero regressions required
5. PR via `gh pr create --repo St0nefish/agent-toolkit --base master`
## Reference PR
- [PR #38: feat: add uv/uvx classifier](https://github.com/St0nefish/agent-toolkit/pull/38) — full uv/uvx classifier with 102 tests, submitted 2026-03-18