- Home
- Skills
- Workflow Automation
- Claude Code Starter Pack: rules, handoffs, cheap subagents
Works with the AI Agents you already use
Claude Code Starter Pack: rules, handoffs, cheap subagents
Rules, handoff templates and three cheap pinned-model subagents so long Claude Code sessions prove "done" and resume from a written state.
$12
About this skill
Why this exists
I'm 18, in Bucharest, and since June I've been trying to get Claude Code to run a small business for me while I'm in school: a desktop agent app, then a chat-to-terminal pipe, then a website pipeline, then cold-calling salons with a voice agent, now 26 demo sites and SMS. These files are the part that survived every attempt: the rules, handoffs and cheap subagents that let a session run for hours without me and resume the next day from a written state. Revenue so far from all of it: $0. The pack is the process, not a promise.
Stop losing context between Claude Code sessions. A tested set of rules, handoff templates and three pinned-model subagents, in plain markdown you can read in ten minutes. It makes the model prove "done" with a command, and makes the next session start from a written handoff instead of a blank slate. No software to install. Copy files, replace the placeholders, run.
What is inside
- Project rules template: NEVER-laws, "numbers only from commands run this session", output style, logging rule.
- Phase gates: plan, one commit per phase with an executable gate, verify with evidence, handoff.
- Handoff protocol: running log, latest pointer, and a full session handoff with a worked example.
- Step-level pulse logging convention with evidence tags (append-only).
- Three subagents (coder, verifier, explore) pinned to cheaper models, with caps, a 3-workers-plus-1-verifier limit and a 25-line prompt budget.
- Optional two-machine peer channel over a synced folder: claims, request/response, approval files, guardrails.
- Example SessionStart hook (uses log plus unanswered peer requests).
- MIT licensed; edit and reuse freely.
Only need the two-machine part? Tito Pulse is the pulse log + peer channel on its own, also on Agensi under my name.
Who it is for
Solo developers who already use Claude Code and run sessions long enough to hit context limits or "it said it was done but it was not".
Who it is not for
Anyone expecting a framework, an installer or guaranteed behavior. These are instructions, and models follow instructions imperfectly. If you want enforcement, you still need hooks and permissions, and the pack says so.
Rules are instructions, not enforcement; models follow them imperfectly. Model aliases and agent frontmatter fields change between Claude Code releases. Caps (60 tool calls, 100k context) are starting values to tune. No telemetry, no installer.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 13 days ago
Requires