# Fnord
[](https://github.com/sysread/fnord/actions/workflows/run-tests.yml)
Fnord is a command line tool the builds a searchable database of your files,
using AI-generated embeddings to index and search your code base, notes, and
other (non-binary) files. Notably, it provides a conversational interface to
research within your project, answering complex questions about your code base.
## Installation
1. Install `elixir` if necessary:
```bash
# MacOS
brew install elixir
# Debian-based
sudo apt-get install elixir
```
2. Add the mix escript path to your shell's PATH:
```bash
echo 'export PATH="$HOME/.mix/escripts:$PATH"' >> ~/.bashrc
source ~/.bashrc
```
3. Install the script:
```bash
mix escript.install github sysread/fnord
```
Use the same command to reinstall. It will offer to overwrite the existing
installation.
4. Set `OPENAI_API_KEY`
Set this in your shell environment to the OpenAI project key you wish to use
for this tool. You can create a new project and get a key
[here](https://platform.openai.com/api-keys).
## Usage
### Index
The first time you run this, especially on a large codebase, it will take a
while to index everything. Subsequent runs will be faster, re-indexing only
those files which have changed since they were last indexed.
```bash
fnord index --project foo --dir /path/to/foo
```
You can **reindex** the project, forcing it to reindex all files.
```bash
fnord index --project foo --dir /path/to/foo --reindex
```
You can also watch the project for changes and reindex them as they happen
using [watchman](https://github.com/facebook/watchman). Just be sure to use
`--quiet` to suppress interactive output.
```bash
watchman-make -p '**/*' --settle 5 --run "fnord index --project $project --dir $project_root --quiet"
```
...or use the `fnord-watch` script in the [tools directory on
GitHub](https://github.com/sysread/fnord/blob/main/tools/fnord-watch).
```bash
fnord-watch -p foo -d /path/to/foo
```
### Search
Search for files in the project that match a query.
```bash
fnord search --project foo --query "some search query"
```
If you want more detail about each file matched:
```bash
fnord search --project foo --query "some search query" --detail
```
### Ask
Ask the AI assistant to answer questions about your project.
```bash
fnord ask foo "how do you run the tests for this project?"
# Pipe output to `glow` to render markdown
fnord ask foo "summarize the dependencies of this project" | glow
```
### Miscellaneous
- **List projects:** `fnord projects`
- **List files in a project:** `fnord files --project foo`
- **Show the AI-generated summary of a file:** `fnord summary --project foo --file bar`
- **Delete a project:** `fnord delete --project foo`
Note that deleting a project only deletes from the index, not the actual files.
## Tool usage
Internally, the `ask` command uses the OpenAI chat completions API to generate
a response, implementing a function tool to allow the assistant to query the
database for information.
`fnord` can be used to implement a similar tool for your own projects. While
the `ask` command severely limits the parameters that the assistant may utilize
(`query` only, with `project` being provided by the user's invocation of the
command), the following syntax includes the full set of parameters available
for the `search` command.
```json
{
"name": "search_tool",
"description": "Searches for matching files and their contents in a project.",
"parameters": {
"type": "object",
"properties": {
"project": {
"type": "string",
"description": "Project name for the search."
},
"query": {
"type": "string",
"description": "The search query string."
},
"detail": {
"type": "boolean",
"description": "Include AI-generated file summary if set to true."
},
"limit": {
"type": "integer",
"description": "Limit the number of results (default: 10)."
},
"concurrency": {
"type": "integer",
"description": "Number of concurrent threads to use for the search (default: 4)."
}
},
"required": ["project", "query"]
}
}
```
# TODO
- output formatted markdown results from `ask`
- mitigate owl's bs; this thing is chincy af
- we have GOT to come up with something bettern than Owl; it has SO many visual glitches, especially when there are many steps
- restore progress display in indexer subcommand
- nested tasks in tui
- abstract agents out into a behaviour
- support for tool calls
- support for "accumulator" agents
