Skip to content

Tools Reference

LongerAgent comes with 16 built-in tools and 8 orchestration tools. Additional tools can be added through the Skills system and MCP integration.

Built-in Tools (16)

These tools are available to the agent in every session.

File Operations

ToolDescription
read_fileRead a file's contents. Supports text files and images (PNG, JPG, GIF, WebP) on multimodal models -- the agent can directly see and analyze images.
write_fileWrite content to a file, creating it if it does not exist.
edit_fileEdit a file with targeted find-and-replace operations.
apply_patchApply a unified diff patch to one or more files.
list_dirList the contents of a directory.
globFind files matching a glob pattern (e.g., **/*.ts).
grepSearch file contents with regular expressions.
diffShow the diff between two files or a file and a string.

Shell

ToolDescription
bashRun a shell command and return its output. Has a 10-minute hard timeout and 200KB output cap per stream.
bash_backgroundRun a shell command in the background. Useful for starting dev servers or long-running processes.
bash_outputRead output from a background shell process.
kill_shellKill a running background shell process.

Testing

ToolDescription
testRun the project's test suite or a specific test file.
timeReturn the current local time, timezone, and UTC offset.

Web

ToolDescription
web_searchSearch the web. Uses the provider's native web search when available (OpenAI, Kimi, GLM), or falls back to a web search tool.
web_fetchFetch and read the content of a URL.

Orchestration Tools (8)

These tools manage sub-agents, context, and user interaction.

Sub-Agent Management

ToolDescription
spawn_agentSpawn sub-agents from a YAML call file. See Sub-Agents.
kill_agentKill one or more running sub-agents by ID.
check_statusCheck the status of running sub-agents.
waitWait for specific sub-agents to complete.

Context Management

ToolDescription
show_contextInspect the current context distribution -- how much space each segment uses. The agent uses this to decide what to summarize.
summarize_contextSurgically compress selected context segments while preserving key decisions.

User Interaction

ToolDescription
askAsk the user 1-4 structured questions with 1-4 options each. Used when the agent needs a decision before proceeding.
planManage an execution plan with tracked checkpoints.

Skills Tool

When skills are enabled, a dynamic skill tool becomes available. This tool dispatches to the active skill's instructions. Manage skills with the /skills command or by asking the agent to install new ones.

The built-in skill-manager skill enables the agent to autonomously search for, download, and install new skills.

See Skills for details.

MCP Tools

MCP servers can provide any number of additional tools. These are configured in ~/.longeragent/mcp.json and appear alongside the built-in tools automatically.

See MCP Integration for details.

Tool Safety

LongerAgent does not sandbox tool execution. The bash tool runs shell commands directly, and file tools write to disk without confirmation. This is by design for productivity, but it means you should:

  • Run LongerAgent in trusted environments.
  • Review what the agent does, especially for destructive operations.
  • Use the explorer sub-agent template for read-only investigation tasks.

Released under the MIT License.