# Budget Continue

> >

- **Type:** Skill
- **Install:** `agentstack add skill-enakoneschniy-claude-dev-stack-budget-continue`
- **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior
- **Seller:** [Enakoneschniy](https://agentstack.voostack.com/s/enakoneschniy)
- **Installs:** 0
- **Category:** [Productivity](https://agentstack.voostack.com/c/productivity)
- **Latest version:** 0.1.0
- **License:** MIT
- **Upstream author:** [Enakoneschniy](https://github.com/Enakoneschniy)
- **Source:** https://github.com/Enakoneschniy/claude-dev-stack/tree/main/skills/budget-continue

## Install

```sh
agentstack add skill-enakoneschniy-claude-dev-stack-budget-continue
```

Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent.

## About

# Budget Continue Skill

When the session budget warning fires, this skill presents a 4-option continuation menu and (when invoked inside a Claude Code session) can schedule the appropriate follow-up action using native Claude Code tools.

## Invocation

```
claude-dev-stack budget continue
```

Or invoke this skill directly: `/budget-continue`

## The 4 Options

### Option 1: Remind me later (CronCreate)

Creates a one-shot CronCreate reminder that fires in the current Claude Code session after a delay.

When the user selects this option inside a Claude Code session:

```
Use CronCreate to schedule a reminder:
  Title:    "Resume GSD session — budget warning"
  Prompt:   "/budget-continue"
  Schedule: "+1h"  (or user-chosen delay)
  Repeat:   false
```

### Option 2: Auto-continue locally (Desktop task)

Creates a Desktop task that fires on the local machine after a delay and runs the GSD resume workflow.

When the user selects this option inside a Claude Code session:

```
Use the Desktop task tool:
  Prompt:   "/gsd-resume-work"
  Schedule: "+1h"  (or user-chosen delay)
```

Note: The machine must be on when the task fires. STATE.md is read to determine what to resume.

### Option 3: Auto-continue in cloud (Cloud task)

Creates a Cloud task that runs on Anthropic infrastructure. Performs a fresh git clone and runs the GSD resume workflow autonomously — works even when the local machine is off.

When the user selects this option inside a Claude Code session:

```
Use the Cloud task tool:
  Repo:     (git remote origin URL — auto-detected)
  Prompt:   "/gsd-resume-work"
  Schedule: "+1h"  (or user-chosen delay)
```

Note: All state must be committed to git. GSD guarantees this via STATE.md.

### Option 4: Continue now

No scheduling action. Prints a message to continue with the current session.
Useful if the user decides to keep working and address the budget limit later.

## What the scheduled task runs

Both Desktop and Cloud tasks run `/gsd-resume-work`, which reads `.planning/STATE.md`:
- If `stopped_at` is present: resumes from that point using `resume_file` as context
- If no stopped work: runs maintenance tasks (check PRs, `/gsd-next`, `/gsd-health`)

The `loop.md` template (if installed to `.claude/loop.md`) provides the full maintenance loop instructions for scheduled tasks.

## Notes

- Hooks cannot interactively invoke Claude Code tools — hence the CLI command for manual invocation.
- When invoked as a skill inside a Claude Code session, scheduling tools (CronCreate, Desktop, Cloud) are directly available.
- The budget warning message includes: `Run: claude-dev-stack budget continue` as a reminder.

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [Enakoneschniy](https://github.com/Enakoneschniy)
- **Source:** [Enakoneschniy/claude-dev-stack](https://github.com/Enakoneschniy/claude-dev-stack)
- **License:** MIT

Install and usage instructions live in the source repository linked above.

## Pricing

- **Free** — Free

## Security capabilities

Automated source analysis of v0.1.0 — what this tool can access:

- **Network access:** no
- **Filesystem access:** no
- **Shell / process execution:** no
- **Environment & secrets:** no
- **Dynamic code execution:** no

*"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.*


## Versions

- **0.1.0** — security scan: passed — Imported from the upstream source.

## Links

- Listing page: https://agentstack.voostack.com/l/skill-enakoneschniy-claude-dev-stack-budget-continue
- Seller: https://agentstack.voostack.com/s/enakoneschniy
- Browse the marketplace: https://agentstack.voostack.com/browse

---
Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.