In order to work best with the latest releases, we should:
- Use the new --trust-tools= flag, and trust the required tools (fs_read,fs_write,execute_bash, and the git tools)
- Use the --no-interactive flag
q chat --help gives:
~ q chat --help
AI assistant in your terminal
Usage: q chat [OPTIONS] [INPUT]
Arguments:
[INPUT] The first question to ask
Options:
--no-interactive Print the first response to STDOUT without interactive mode. This will fail if the prompt
requests permissions to use a tool, unless --trust-all-tools is also used
--profile <PROFILE> Context profile to use
--trust-all-tools Allows the model to use any tool to run commands without asking for confirmation
--trust-tools <TOOL_NAMES> Trust only this set of tools. Example: trust some tools:
'--trust-tools=fs_read,fs_write', trust no tools: '--trust-tools='
-v, --verbose... Increase logging verbosity
-h, --help Print help
Tools we may want to use:
- execute_bash
- fs_read
- fs_write
- internal_command
- thinking
- fetch___fetch
- git___git_add
- git___git_diff
- git___git_diff_staged
- git___git_diff_unstaged
- git___git_log
- git___git_show
- git___git_status
We should also format the shell script such that it is readable in 80 column displays.
In order to work best with the latest releases, we should:
q chat --help gives:
Tools we may want to use:
We should also format the shell script such that it is readable in 80 column displays.