Tools and sub-agents
The tools Line uses, background shells, sub-agents and verification.
- Can Line CLI edit Jupyter notebooks?Yes. The notebook_edit tool replaces, inserts or deletes cells in .ipynb files. Line reads the notebook as JSON to find the right cell.How-to
- Can Line CLI install packages or use tools like Docker?Yes, through shell commands such as npm install or docker build, which ask for approval unless you allowed them.FAQ
- Can Line CLI make git commits for me?Line's git tool is read-only, but it can commit with a shell command like git commit, which asks for approval unless allowed.FAQ
- Can Line CLI search the web and read documentation?Yes. web_search finds pages and web_fetch reads them as text. Web content is treated as untrusted, so injected instructions trigger approvals.FAQ
- Can Line CLI use a web browser to test my site?Line has no built-in browser. web_fetch reads pages as text. For browser testing, Line can run your own test tools like Playwright.FAQ
- Can two sub-agents edit the same file at once?No. Line locks files per agent, requires a read before an edit, and refuses an edit if the file changed since that agent read it.FAQ
- How do background shells work in Line CLI?Line runs dev servers and watchers in background shells. /bashes lists them, /bashes <id> shows output, /bashes kill <id> stops one.How-to
- How do I create a custom sub-agent in Line CLI?Add .line/agents/<name>.md (or ~/.line/agents/<name>.md) with front matter name, description, tools, model, effort, max_steps and instructions.How-to
- How do I see what Line CLI's sub-agents did?/agents lists this session's sub-agent runs with kind, state, steps, time, tokens and files. /agents <n> shows one run's steps and report.How-to
- How do I tell Line CLI which command checks my project?Set verify_command in the project's .line/settings.json, for example "make check". Line runs it before finishing when files changed.How-to
- How does Line CLI check for compile and type errors?The diagnostics tool runs your project's own type checker, compiler or linter, detected from the project, and returns the errors.FAQ
- What are sub-agents in Line CLI?Sub-agents are extra agents Line starts on your machine to explore, research, review or do work in parallel, then report back.Explainer
- What happens when a command prints a huge amount of output?Output over about 30 KB is saved to a file. The agent gets the path plus the beginning and end, with more of the end when the command failed.FAQ
- What is the repo map in Line CLI?The repo map is a ranked outline of your codebase: the most central files with their functions, types and classes, built with tree-sitter.FAQ
- What tools can Line CLI use?Line's built-in tools cover files, search, shell, git, diagnostics, a repo map, notebooks, the web, a to-do plan, sub-agents and skills.Reference
- Why does Line CLI keep going when the model tries to stop early?If the model stops before the work is done, with open to-dos, a promise and no action, or an empty reply, Line tells it to continue.Explainer
- Why does Line CLI run my tests before it finishes?When Line changed files, it checks the work before finishing: your verify_command, else your tests, else your type checker. Failures go back to the agent.Explainer
- Why does Line CLI write long files in several parts?Very large single writes can be cut in transit. Line writes big files in parts of about 120 lines with append_file, and redoes cut steps.FAQ
Still need help?
Tried the steps above and still stuck? Email us. We usually reply within the first hour, and always within 24 hours. A full support team is with you.
Infrastructure questions, or permanent deletion of your data
infrastructure@clatterlabs.xyzFor deleting your whole account or the data of one service.