SPRUT Agent Kit skill
A starter kit for ClaudeClaw: agent soul, SQLite memory, 25 ready skills, cron tasks and progress messages in Telegram.
- 66 ★ on the repository
- Last commit
- MIT
Broad access
Why this level
- The install script sets up a daemon, cron tasks and autostart, and an agent with memory gets machine access.
- At weak security levels the agent acts nearly unrestricted and communicates through Telegram.
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/sprut-agent-kitDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- The install script sets up a daemon, cron tasks and autostart, and an agent with memory gets machine access.
- At weak security levels the agent acts nearly unrestricted and communicates through Telegram.
This is third-party code. Review the repository files before installing.
What it does
A kit for ClaudeClaw, a Claude Code CLI plugin, and not for OpenClaw. The `install.sh` script checks dependencies (bun, git, claude), installs ClaudeClaw, asks for a Telegram ID and timezone and lays out the agent soul (`SOUL.md`, `AGENTS.md`) and 25 skills. It can optionally import memory and set up autostart. Memory lives in SQLite with embeddings: hybrid search (vector 0.7 and FTS5 0.3), fact decay and auto-extraction from dialogues. Skills include research, debugging, brainstorming, plans, subagents, website audit, Marp presentations, diagrams, Reddit, YouTube transcripts and posts. There is a daemon with a REST API and a web UI on port 4632, cron tasks for backups and memory cleanup and a layer protecting personal data.
Who it is for. Claude Code users who want a personal agent with memory and skills out of the box and not an empty shell, and who are ready to keep their own tokens.
Good fit when
- A personal assistant with memory that can be messaged in Telegram is needed.
- An agent must move to a new Mac with one script, soul and memory included.
- A set of working skills for research, diagrams and presentations is wanted right away.
- An author builds their own configuration and takes a ready structure template.
Not a fit when
- OpenClaw is required: the kit targets ClaudeClaw only.
- A hosted service without local scripts is required: this is not SaaS.
- Install scripts and daemons cannot run on the machine.
Example request
Install the kit, start the agent and ask it to research the last 30 days on a topic and send progress to Telegram.Limitations
The kit fits only ClaudeClaw and is incompatible with OpenClaw: the projects have different architectures. bun, git, `claude` and a Telegram bot are required, and tokens and real settings stay local with you. Security levels (locked, strict, moderate, unrestricted) are set in config: at weak levels an agent with memory and machine access acts nearly unrestricted. Some skills, such as reddit and tubescribe, call external services. The last commit is dated 29 October 2026.
How to disable. Stop the ClaudeClaw daemon and remove autostart, then delete the copied `SOUL.md`, `AGENTS.md` and skill folders from `~/.claude/skills/`. Clean the SQLite memory database and `~/.claude/claudeclaw/` configuration separately.
FAQ
How does this differ from OpenClaw?
The kit targets ClaudeClaw, a Claude Code CLI plugin, while OpenClaw is a self-hosted gateway for agents. OpenClaw kits do not fit.
What does install.sh do?
It checks bun, git and claude, installs ClaudeClaw, asks for a Telegram ID and timezone, installs the soul and 25 skills and optionally imports memory and sets up autostart.
How does memory work?
SQLite with embeddings: hybrid search (vector 0.7, FTS5 0.3), fact decay (semantic 0.01 per day, episodic 0.05) and automatic fact extraction from dialogues.
Is there a web interface?
Yes, the daemon exposes a web UI on port 4632 and a REST API for running subagents and checking their status.
How is personal data protected?
The agent stores no passports, cards or addresses, answers only the owner through a Telegram ID whitelist and runs at one of the levels from locked to unrestricted.
Discussion
No questions or reviews yet.
Related
SkillsA compact OpenClaw alternative: a personal assistant on Claude Agent SDK with agents in isolated containers and messengers added via skills
A Claude Code plugin that gives the agent a name, a personality and memory across sessions, plus chat channels.
Skills, mods, subagents and hooks for Claude Code: the agent inspects your setup and installs what fits after your approval.
ClawMetry reads the session files agents already write and shows the timeline, tool calls, tokens and cost in one dashboard.
A set of skills, slash commands and agents for Claude Code that helps you build your own skills, agents, hooks, slash commands and prompts
A Claude Code reference plus a collection of over 70 skills for development, security, data and docs, with agent configs
README in short
The README is bilingual, with a short Russian part on what is inside and for whom. The English part covers repository structure, configuration, the skill list and the portable soul architecture.