summaryrefslogtreecommitdiffstats
path: root/src/config
AgeCommit message (Collapse)AuthorLines
2023-10-24feat: suppport vi keybindings (#148)sigoden-0/+4
2023-07-10fix: grammar/wording corrections (#135)Jackson Goode-4/+4
2023-07-10fix: minor code cleanup and pedantic lints (#134)Anthony Rubick-51/+51
2023-06-19feat: add new model gpt-3.5-turbo-16k (#126)sigoden-1/+2
2023-05-04feat: add `.copy` command (#119)sigoden-16/+1
2023-05-04feat: add `config.auto_copy` (#102)shane s-0/+18
* added config option to copy all ChatGPT output to clipboard * formatted with * updates --------- Co-authored-by: sigoden <sigoden@gmail.com>
2023-05-04fix: config file with strict perms (#118)sigoden-0/+6
2023-03-22fix: mask api_key/organization_id in info text (#93)sigoden-2/+5
2023-03-20feat: add `--dry-run` (#83)sigoden-3/+0
2023-03-20feat: check token usage in dry_run mode (#82)sigoden-11/+15
2023-03-17feat: add `config.organization_id` (#77)sigoden-12/+12
2023-03-16feat: add `config.connect_timeout` (#76)sigoden-0/+9
2023-03-16feat: support multiple models (#71)sigoden-18/+49
2023-03-13chore: optimzie role argssigoden-5/+8
2023-03-13feat: support role args (#69)sigoden-7/+45
* feat: support role args We can use role args to pass some additional arguments to the prompt. ``` - name: convert:json:yaml prompt: convert __ARG1__ below to __ARG2__ ``` `:json:yaml` is `role args`. It has two args: - arg1 `json`, it will replace __ARG1__ in prompt - arg2 `yaml`, it will replace __ARG2__ in prompt ``` 〉.role convert:json:yaml name: convert:json:yaml prompt: convert json below to yaml temperature: null 〉.role convert:yaml:json name: convert:yaml:json prompt: convert yaml below to json temperature: null ``` different role args, will generate different prompts. * small updates
2023-03-12feat: support HTTPS_PROXY and ALL_PROXY (#68)sigoden-0/+10
2023-03-11feat: add support for NO_COLOR (#67)sigoden-2/+14
2023-03-11feat: support light theme (#65)sigoden-5/+15
There are two ways to enable the light theme: - add `light_theme: true` to config.yaml - use `AICHAT_LIGHT_THEME=true` env var
2023-03-11feat: support more env vars (#63)sigoden-30/+96
- AICHAT_CONFIG_DIR: sepecify the configuration directory - AICHAT_ROLES_FILE: set the `roles.yaml` file path - AICHAT_API_KEY: set the api key
2023-03-11feat: provide `--prompt` for adding a prompt from cli (#62)sigoden-1/+1
2023-03-10feat: type `{` `[` `(` to enter multi-line editing mode (#58)sigoden-1/+1
Abandon `.editor` command, it's too complicated.
2023-03-10chore: opitmize code, move MAX_TOKENS related to config/message.rssigoden-9/+15
2023-03-10fix: forbid .clear role in middle of conversationsigoden-0/+15
2023-03-10fix: cli specify role is not applied when conversation_first set true (#57)sigoden-3/+7
close #56
2023-03-09feat: add config.conversation_first (#55)sigoden-15/+32
If set true, start a conversation immediately upon repl.
2023-03-09refactor: no need to confirm conversation when token used outsigoden-6/+8
2023-03-09refactor: optimize counting tokens (#53)sigoden-58/+42
2023-03-09feat: support two types of role prompts (#52)sigoden-103/+192
1. embeded prompt use __INPUT__ placeholder will generate one user message when send to gpt ``` - name: shell prompt: > I want you to act as a linux shell expert. Q: How to unzip a file A: unzip file.zip Q: __INPUT__ A: ``` 2. system prompt no __INPUT__ placeholder will generate on system message and one user message when send to gpt ``` - name: shell prompt: | I want you to act as a linux shell expert. I want you to answer only with bash code. Do not write explanations. ```
2023-03-09refactor: trivial updatessigoden-9/+3
- change prompt's temp role name - questionnaire of 'save chat messages' default set true - `.info` do not show role name
2023-03-09feat: add remain tokens indicator and max tokens guard (#50)sigoden-33/+50
2023-03-09feat: support conversation (#48)sigoden-0/+512