Skillget

ai-tools

skill
Security Audit
Fail
Health Warn
  • No license — Repository has no license file
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Community trust — 16 GitHub stars
Code Fail
  • rm -rf — Recursive force deletion command in ai-memory/scripts/build.sh
Permissions Pass
  • Permissions — No dangerous permissions requested
Purpose
This tool is a monorepo offering a marketplace of plugins for AI coding assistants. It provides persistent cross-session memory, safer Bash auto-approval by analyzing command ASTs, and a standalone Docker-based script for running autonomous, iterative coding loops.

Security Assessment
The security risk is Medium. The Bash auto-approval plugin is inherently sensitive; it evaluates and conditionally executes shell commands. While it uses AST-based checks to do this safely, bugs could lead to unintended command execution. The repository also contains a recursive force deletion command (`rm -rf`) inside a build script (`ai-memory/scripts/build.sh`). If handled improperly, this poses a risk of accidental local data loss. The memory plugin accesses and stores local project context, but no hardcoded secrets or dangerous system permissions were detected. Network requests appear to be limited to standard package management.

Quality Assessment
The project is actively maintained, with its last push occurring today. It features detailed documentation and has achieved a fair level of community trust with 16 GitHub stars. However, it completely lacks a standard open-source license. Without a license, strict copyright laws apply, meaning developers legally do not have permission to use, modify, or distribute the code.

Verdict
Use with caution—the tool's capabilities involve sensitive shell execution, and the complete lack of a license creates potential legal ambiguity for your project.
SUMMARY

AI Tools to be installed as skills, plugins, etc.

README.md

claude-marketplace

A monorepo for Claude Code plugins and standalone AI tooling. Install the marketplace for two practical plugins (persistent memory and safer Bash auto-approval), or use the standalone ralph-loop for autonomous multi-iteration coding.

Quickstart

Add this marketplace, then install both plugins:

/plugin marketplace add damusix/ai-tools
/plugin install ai-memory@damusix-ai-tools
/plugin install auto-approve-compound-bash@damusix-ai-tools

What you get

ai-memory

Give Claude long-term project memory across sessions, with local-first storage and tools to organize context.

  • Captures observations and synthesizes reusable memories
  • Shares one memory service across Claude Code sessions
  • Provides MCP tools for saving, searching, and organizing memory
  • Includes dashboard UI to browse and manage memories
  • Adds /remember and /forget slash commands

Docs: ai-memory/README.md

auto-approve-compound-bash (cc-auto-approve-fix source)

Auto-approve compound Bash commands safely by parsing each command segment against allow/deny rules.

  • Uses a native Go shell parser (mvdan.cc/sh/v3/syntax) for AST-based checks
  • Handles compound operators, substitutions, subshells, and nested bash/sh/zsh -c
  • Falls through safely when uncertainty is detected
  • Supports --explain, simulate, and doctor workflows
  • Ships prebuilt binaries for darwin/linux and amd64/arm64

Docs: cc-auto-approve-fix/README.md

Standalone tools

ralph-loop

An autonomous coding loop that drives AI agents (Claude, Amp, Codex, OpenCode) through iterative development cycles. Not a Claude Code plugin — ralph is a standalone zx script that runs inside a self-contained Docker environment.

  • Splits large tasks into focused iterations — one job, one commit, one status update per cycle
  • Each iteration starts with a fresh agent session, keeping context clean and decisions sharp
  • Quality gates stash broken work and log failures so the next iteration can recover
  • Streams real-time output (text, tool calls, timestamps) while the agent works
  • Includes a Dockerfile with all runtimes and tooling pre-installed — nothing to configure on the host
docker compose up -d --build
docker exec -it ralph-wiggum zsh
cd ~/my-project && ralph init

Docs: ralph-loop/README.md

Why this marketplace

  • One setup path gives you both plugins immediately
  • Plugin names are stable in the damusix-ai-tools marketplace catalog
  • Memory stays local and command auto-approval remains rule-driven

Contributing

This repo is a pnpm workspace. Install dependencies from the root:

pnpm install

Versioning

Plugin versions are managed by a CLI tool at scripts/cli/. It bumps versions, syncs all manifest files, generates changelogs from git history, and creates git tags.

Interactive mode — select plugins and bump types via prompts:

pnpm dev version

Flag mode — bump a single plugin directly:

pnpm dev version ai-memory patch
pnpm dev version auto-approve-compound-bash minor

Plugin names come from .claude-plugin/marketplace.json, not directory names. The CLI requires a clean working tree before running.

What it does per plugin:

  1. Bumps the version in the source-of-truth file (package.json or plugin.json)
  2. Syncs version across marketplace.json and plugin.json
  3. Regenerates CHANGELOG.md from conventional commits (feat, fix, breaking changes)
  4. Commits as release: <plugin>@<version> and creates a git tag

Adding a new plugin

  1. Add the plugin directory to pnpm-workspace.yaml
  2. Add an entry to .claude-plugin/marketplace.json with name, version, and source
  3. The versioning CLI discovers it automatically

License

MIT

Reviews (0)

No results found