AGENT SKILL

run-shell-fallback

Use run_shell with inline Python as a reliable fallback when execute_code_sandbox or read_file fail

License 许可
MIT
所属仓库
HKUDS/AgentPilot
最近更新
2026-11-12
Stars
★ 8.0k
Forks
944

01 前言 Front Matter

name
run-shell-fallback
description
Use run_shell with inline Python as a reliable fallback when execute_code_sandbox or read_file fail

02 资源文件 1

03 SKILL.md 正文

Run Shell Fallback Pattern

This skill provides a reliable workaround when execute_code_sandbox or read_file fail with 'unknown error' by switching to run_shell with inline Python code.

When to Use

Apply this pattern when you encounter:

  • execute_code_sandbox fails with "unknown error" or timeout
  • read_file fails to read accessible files with "unknown error"
  • Sandbox environment is unreliable for your specific task

How to Implement

1. File Reading Fallback

When read_file fails, use run_shell with Python or cat:

For text files:

run_shell(command="cat /path/to/file.txt")

For structured parsing (Python one-liner):

run_shell(command="python3 -c \"import json; print(json.load(open('/path/to/file.json')))\"")

For multi-line Python processing:

run_shell(command="python3 << 'EOF'
with open('/path/to/file.txt', 'r') as f:
    content = f.read()
    # Process content here
    print(content.upper())
EOF")

2. Code Execution Fallback

When execute_code_sandbox fails, use run_shell with inline Python:

Simple one-liners:

run_shell(command="python3 -c \"print('Hello'); import os; print(os.getcwd())\"")

Multi-line scripts (heredoc):

run_shell(command="python3 << 'EOF'
import os
import json

def process_data(data):
    return {k: v.upper() if isinstance(v, str) else v for k, v in data.items()}

with open('input.json') as f:
    data = json.load(f)
    
result = process_data(data)
print(json.dumps(result, indent=2))
EOF")

With inline file write:

run_shell(command="python3 << 'EOF'
result = {'status': 'success', 'count': 42}
with open('output.json', 'w') as f:
    json.dump(result, f, indent=2)
print('File written successfully')
EOF")

3. Decision Flow

1. Attempt execute_code_sandbox or read_file
2. If failure with 'unknown error':
   a. Determine task type (code execution vs file reading)
   b. Choose inline Python approach (-c for simple, heredoc for complex)
   c. Execute via run_shell
   d. Parse output as needed
3. Continue with task using results from run_shell

Best Practices

  1. Escape quotes properly in -c commands: Use \" for double quotes inside the command string

  2. Use heredoc for complex scripts: When code exceeds 2-3 lines or needs multiple imports, use << 'EOF' syntax

  3. Quote EOF marker: Use << 'EOF' (with quotes) to prevent variable expansion in the heredoc

  4. Capture stdout: Design your inline Python to print results to stdout for easy capture

  5. Error handling: Add try/except blocks in your inline Python to catch and report errors clearly

  6. File paths: Use absolute paths when possible, or ensure working directory is correct

Examples

Example 1: Read and parse JSON after read_file failure

run_shell(command="python3 -c \"import json; data=json.load(open('config.json')); print(data['key'])\"")

Example 2: Process CSV data after execute_code_sandbox failure

run_shell(command="python3 << 'EOF'
import csv
with open('data.csv', 'r') as f:
    reader = csv.DictReader(f)
    rows = list(reader)
    print(f\"Total rows: {len(rows)}\")
    for row in rows[:5]:
        print(row)
EOF")

Example 3: Transform and write output file

run_shell(command="python3 << 'EOF'
import json
with open('input.json') as f:
    data = json.load(f)
transformed = [item.upper() if isinstance(item, str) else item for item in data]
with open('output.json', 'w') as f:
    json.dump(transformed, f, indent=2)
print('Transformation complete')
EOF")

Limitations

  • run_shell executes in the shell environment, not the sandbox - ensure paths and dependencies are available
  • Output is limited to stdout/stderr - use file writes for large data
  • Complex multi-file projects may require creating temporary script files first

04 同仓库更多 Skills

HKUDS
HKUDS/AgentPilot

agentpilot

Set up AgentPilot and use its single entry to prepare missing inputs and run configured RSI experiments. Use when the user requests AgentPilot installation or work with AgentPilot.

★ 8.0k
HKUDS
HKUDS/AgentPilot

remember

Review auto-memory entries and propose promotions to OPENSPACE.md, OPENSPACE.local.md, or retained auto-memory topics; detect duplicates, stale facts, conflicts, and ambiguous entries before making changes.

★ 8.0k
HKUDS
HKUDS/AgentPilot

skill-discovery

Search reusable OpenSpace skills locally, and browse cloud packages step by step when a cloud skill may help. Reusing proven skills saves tokens, improves reliability, and extends your capabilities beyond built-in tools.

Skill与智能体开发
★ 8.0k
HKUDS
HKUDS/AgentPilot

delegate-task

Delegate tasks to OpenSpace — a full-stack autonomous worker for coding, DevOps, web research, and desktop automation, backed by an extensive MCP tool and skill library. Skills auto-improve through use, reducing token consumption over time. A cloud community lets agents share and collectively evolve reusable skills.

Skill与智能体开发
★ 8.0k
HKUDS
HKUDS/AgentPilot

write-special-chars-ts-check

Workaround for write_file failures caused by special characters (apostrophes, backticks, template literals) using a Python heredoc script, plus a reliable TypeScript type-checking step that filters output to only the newly-written files to avoid being misled by pre-existing errors.

★ 8.0k
HKUDS
HKUDS/AgentPilot

write-special-chars-readback

Workaround for write_file failures caused by special characters (apostrophes, backticks, template literals) using a Python heredoc script, with mandatory post-write verification via read_file and cleanup of /tmp helper scripts.

★ 8.0k