Skip to content
Agentmart
Submit a tool

background

Fires off a full Claude Code instance in the background

Skill by PrashamTrivedi

IndexedMIT
Write a review

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/prashamtrivedi-background

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Fires off a full Claude Code instance in the background

Use cases: agent-building

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/background

What's inside

Skill name
background
Description
Fires off a full Claude Code instance in the background
License
MIT
Instructions
138 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

---
description: Fires off a full Claude Code instance in the background
argument-hint: [prompt] [model] [report-file]
allowed-tools: Bash, Read, Edit, MultiEdit, Write, Grep, Glob, WebFetch, WebSearch, Task
---

# Background Claude Code

Run a Claude Code instance in the background to perform tasks autonomously while
you continue working.

## Variables

USER_PROMPT: $1 MODEL: $2 (defaults to 'sonnet' if not provided) REPORT_FILE: $3
(defaults to './agents/background/background-report-DAY-NAME_HH_MM_SS.md' if not
provided)

## Instructions

- Capture timestamp in a variable FIRST to ensure consistency across file
  creation and references
- Create the initial report file with header BEFORE launching the background
  agent
- Fire off the new Claude Code instance using the Bot tool with
  run_in_background=true
- IMPORTANT: Pass the USER_PROMPT exactly as provided with no modifications
- Set the model to either 'sonnet' or 'opus' based on MODEL parameter
- Configure Claude Code with all necessary flags for automated operation
  [Claude Code CLI flags](https://docs.anthropic.com/en/docs/claude-code/cli-reference)
- Claude code is already installed, but not available via which command. So run
  it and report an error if any.
- All report format instructions are embedded in the --append-system-prompt
- Use --print flag to run in non-interactive mode
- Use --output-format text for standard text output
- Use --dangerously-skip-permissions to bypass permission prompts for automated
  operation

## Workflow

1. Create the report directory if it doesn't exist:

```bash
mkdir -p agents/background
```

2. Set default values for parameters:

- `MODEL`
- `TIMESTAMP` (capture once for consistency)
- `REPORT_FILE` (using the captured timestamp)

3. Create the initial report file with just the header (IMPORTANT: Only IF no
   report file is provided - if it is provided, echo it and skip this step):

Get this into your context window so you know what to pass into the Claude Code
command.

```bash
TIMESTAMP=$(date +%a_%H_%M_%S)
echo "TIMESTAMP: ${TIMESTAMP}"
REPORT_FILE="${REPORT_FILE :-. /agents/background/background-report-${TIMESTAMP}.md}"
echo "REPORT_FILE: ${REPORT_FILE}"
echo "# Background Agent Report - ${TIMESTAMP}" > "${REPORT_FILE}"
```

- Execute the command using Bash with run_in_background=true

```bash
claude \
-- model "$ {MODEL}" \
-- output-format text \
-- dangerously-skip-permissions \
-- append-system-prompt "IMPORTANT: You are running as a background agent. Your primary responsibility is to execute work
and document your progress continuously in ${REPORT_FILE}. Iteratively write to the report file continuously as you work.
Every few tool calls you should update the REPORT_FILE ## Progress section. Follow this file structure.
```

## CRITICAL WORKFLOW STRUCTURE

IMPORTANT: You MUST follow this workflow as you work:

1. The file ${REPORT_FILE} has been created with a header. You must UPDATE as
   you proceed it (not recreate it) with this EXACT markdown structure:

```markdown
# Background Agent Report - DAY-NAME_HH_MM_SS

## Task Understanding

Clearly state what User has requested you to do.

## Results

List concrete outcomes and deliverables with specific details. Update as you
work:

- [Specific file created/modified with path]
- [Numeric data or metrics]
- [Actual accomplishments]

## Task Completed (or Task Failed)

[Final summary - success confirmation or failure explanation]

ADDITIONAL SECTIONS (add as needed):

- ## Blockers - Issues preventing progress
- ## Decisions Made - Important choices and rationale
- ## Recommendations - Follow-up suggestions
- ## Warnings - Important issues to note
```

CONTINUOUSLY update ${REPORT_FILE} as you work - after each major step or
finding.

When you finish your work:

- If successful: Rename ${REPORT_FILE} to ${REPORT_FILE.md}.complete.md
- If failed/blocked: Rename ${REPORT_FILE} to ${REPORT_FILE.md}.failed.md

From PrashamTrivedi/CursorPlugin, MIT.

Signals

Stars
0
Forks
0
Open issues
0
Last commit (this folder)
Sep 1, 2026 (1 month ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from PrashamTrivedi

Cite this page

background skill. Agent Mart. https://agentmart.dev/skills/prashamtrivedi-background (last updated 2026-10-24)