feat!: use prompt library in chat buffer and better keymap support

Hopefully this release dramatically improves the usability of the
plugin.

Firstly, you can now mark specific prompts (from the prompt library) as
available to appear in the slash command completion menu in the chat
buffer via the `opts.is_slash_cmd` value. This will close #286.

Secondly, the default keymaps that came with each prompt have been
removed in place of `require("codecompanion").prompt("commit")`. This
makes it much cleaner to assign your own keymaps to prompts in the
prompt library.

README.md

@@ -230,6 +230,9 @@ vim.api.nvim_set_keymap("v", "ga", "<cmd>CodeCompanionChat Add<cr>", { noremap =
 vim.cmd([[cab cc CodeCompanion]])
 ```
 
+> [!NOTE]
+> You can also assign prompts from the library to specific mappings. See the [prompt library](#clipboard-prompt-library) section for more information.
+
 ## :gear: Configuration
 
 Before configuring the plugin, it's important to understand how it's structured.
@@ -441,6 +444,22 @@ require("codecompanion").setup({
 
 The plugin comes with a number of pre-built prompts. As per [the config](https://github.com/olimorris/codecompanion.nvim/blob/main/lua/codecompanion/config.lua), these can be called via keymaps or slash commands (via the inline assistant). These prompts have been carefully curated to mimic those in [GitHub's Copilot Chat](https://docs.github.com/en/copilot/using-github-copilot/asking-github-copilot-questions-in-your-ide). Of course, you can create your own prompts and add them to the Action Palette. Please see the [RECIPES](doc/RECIPES.md) guide for more information.
 
+**Using Keymaps**
+
+You can call a prompt from the library via a keymap using the `prompt` helper:
+
+```lua
+vim.api.nvim_set_keymap("v", "<LocalLeader>ce", "", {
+  callback = function()
+    require("codecompanion").prompt("explain")
+  end,
+  noremap = true,
+  silent = true,
+})
+```
+
+In the example above, we've set a visual keymap that will trigger the Explain prompt. Providing the `short_name` of the prompt as an argument to the helper (e.g. "commit") will resolve the strategy down to an action.
+
 ### :speech_balloon: The Chat Buffer
 
 The chat buffer is where you converse with an LLM from within Neovim. The chat buffer has been designed to be turn based, whereby you send a message and the LLM replies. Messages are segmented by H2 headers and once a message has been sent, it cannot be edited. You can also have multiple chat buffers open at the same.

doc/RECIPES.md

@@ -106,7 +106,7 @@ require("codecompanion").setup({
       opts = {
         mapping = "<LocalLeader>ce",
         modes = { "v" },
-        slash_cmd = "expert",
+        short_name = "expert",
         auto_submit = true,
         stop_context_insertion = true,
         user_prompt = true,
@@ -145,14 +145,14 @@ At first glance there's a lot of new stuff in this. Let's break it down.
 opts = {
   mapping = "<LocalLeader>ce",
   modes = { "v" },
-  slash_cmd = "expert",
+  short_name = "expert",
   auto_submit = true,
   stop_context_insertion = true,
   user_prompt = true,
 },
 ```
 
-In the `opts` table we're specifying that we only want this action to appear in the _Action Palette_ if we're in visual mode. We're also asking the chat strategy to automatically submit the prompts to the LLM via the `auto_submit = true` value. We're also telling the picker that we want to get the user's input before we action the response with `user_prompt = true`. With the `slash_cmd = "expert"` option, the user can run `:CodeCompanion /expert` from the cmdline in order to trigger this prompt. Finally, as we define a prompt to add any visually selected text to the chat buffer, we need to add the `stop_context_insertion = true` option to prevent the chat buffer from duplicating this. Remember that visually selcting text and opening a chat buffer will result in that selection from being adding as a codeblock.
+In the `opts` table we're specifying that we only want this action to appear in the _Action Palette_ if we're in visual mode. We're also asking the chat strategy to automatically submit the prompts to the LLM via the `auto_submit = true` value. We're also telling the picker that we want to get the user's input before we action the response with `user_prompt = true`. With the `short_name = "expert"` option, the user can run `:CodeCompanion /expert` from the cmdline in order to trigger this prompt. Finally, as we define a prompt to add any visually selected text to the chat buffer, we need to add the `stop_context_insertion = true` option to prevent the chat buffer from duplicating this. Remember that visually selcting text and opening a chat buffer will result in that selection from being adding as a codeblock.
 
 ### Prompt options and context
 
@@ -271,6 +271,29 @@ And to determine the visibility of actions in the palette itself:
 
 ## Other Configuration Options
 
+**Allowing a Prompt to appear as a Slash Command**
+
+It can be useful to have a prompt from the prompt library appear as a slash command in the chat buffer, like with the `Generate a Commit Message` action. This can be done by specifiying a `is_slash_cmd = true` option to the prompt:
+
+```lua
+["Generate a Commit Message"] = {
+  strategy = "chat",
+  description = "Generate a commit message",
+  opts = {
+    index = 9,
+    is_default = true,
+    is_slash_cmd = true,
+    short_name = "commit",
+    auto_submit = true,
+  },
+  prompts = {
+    -- Prompts go here
+  }
+}
+```
+
+In the chat buffer, if you type `/` you will see the value of `opts.short_name` appear in the completion menu for you to expand.
+
 **Specifying an Adapter and Model**
 
 ```lua

doc/codecompanion-recipes.txt

@@ -1,4 +1,4 @@
-*codecompanion-recipes.txt*    For NVIM v0.10.0   Last change: 2024 October 04
+*codecompanion-recipes.txt*    For NVIM v0.10.0   Last change: 2024 October 08
 
 ==============================================================================
 Table of Contents                    *codecompanion-recipes-table-of-contents*
@@ -145,7 +145,7 @@ builtin to the plugin as the _Code Advisor_ action:
           opts = {
             mapping = "<LocalLeader>ce",
             modes = { "v" },
-            slash_cmd = "expert",
+            short_name = "expert",
             auto_submit = true,
             stop_context_insertion = true,
             user_prompt = true,
@@ -185,7 +185,7 @@ PALETTE OPTIONS ~
     opts = {
       mapping = "<LocalLeader>ce",
       modes = { "v" },
-      slash_cmd = "expert",
+      short_name = "expert",
       auto_submit = true,
       stop_context_insertion = true,
       user_prompt = true,
@@ -197,7 +197,7 @@ in the _Action Palette_ if we’re in visual mode. We’re also asking the chat
 strategy to automatically submit the prompts to the LLM via the `auto_submit =
 true` value. We’re also telling the picker that we want to get the user’s
 input before we action the response with `user_prompt = true`. With the
-`slash_cmd = "expert"` option, the user can run `:CodeCompanion /expert` from
+`short_name = "expert"` option, the user can run `:CodeCompanion /expert` from
 the cmdline in order to trigger this prompt. Finally, as we define a prompt to
 add any visually selected text to the chat buffer, we need to add the
 `stop_context_insertion = true` option to prevent the chat buffer from
@@ -331,6 +331,32 @@ And to determine the visibility of actions in the palette itself:
 
 OTHER CONFIGURATION OPTIONS*codecompanion-recipes-other-configuration-options*
 
+**Allowing a Prompt to appear as a Slash Command**
+
+It can be useful to have a prompt from the prompt library appear as a slash
+command in the chat buffer, like with the `Generate a Commit Message` action.
+This can be done by specifiying a `is_slash_cmd = true` option to the prompt:
+
+>lua
+    ["Generate a Commit Message"] = {
+      strategy = "chat",
+      description = "Generate a commit message",
+      opts = {
+        index = 9,
+        is_default = true,
+        is_slash_cmd = true,
+        short_name = "commit",
+        auto_submit = true,
+      },
+      prompts = {
+        -- Prompts go here
+      }
+    }
+<
+
+In the chat buffer, if you type `/` you will see the value of `opts.short_name`
+appear in the completion menu for you to expand.
+
 **Specifying an Adapter and Model**
 
 >lua

doc/codecompanion.txt

@@ -1,4 +1,4 @@
-*codecompanion.txt*        For NVIM v0.10.0       Last change: 2024 October 04
+*codecompanion.txt*        For NVIM v0.10.0       Last change: 2024 October 07
 
 ==============================================================================
 Table of Contents                            *codecompanion-table-of-contents*
@@ -205,6 +205,9 @@ For an optimum workflow, I recommend the following keymaps:
 <
 
 
+  [!NOTE] You can also assign prompts from the library to specific mappings. See
+  the |codecompanion-prompt-library| section for more information.
+
 CONFIGURATION                                    *codecompanion-configuration*
 
 Before configuring the plugin, it’s important to understand how it’s
@@ -454,6 +457,24 @@ Chat
 Of course, you can create your own prompts and add them to the Action Palette.
 Please see the RECIPES <doc/RECIPES.md> guide for more information.
 
+**Using Keymaps**
+
+You can call a prompt from the library via a keymap using the `prompt` helper:
+
+>lua
+    vim.api.nvim_set_keymap("v", "<LocalLeader>ce", "", {
+      callback = function()
+        require("codecompanion").prompt("explain")
+      end,
+      noremap = true,
+      silent = true,
+    })
+<
+
+In the example above, we’ve set a visual keymap that will trigger the Explain
+prompt. Providing the `short_name` of the prompt as an argument to the helper
+(e.g.� "commit") will resolve the strategy down to an action.
+
 
 THE CHAT BUFFER ~
 

lua/cmp_codecompanion/slash_commands.lua

@@ -1,5 +1,6 @@
 local config = require("codecompanion").config
 local SlashCommands = require("codecompanion.strategies.chat.slash_commands")
+local strategy = require("codecompanion.strategies")
 
 local source = {}
 
@@ -20,12 +21,15 @@ function source:get_keyword_pattern()
 end
 
 function source:complete(params, callback)
-  local items = {}
   local kind = require("cmp").lsp.CompletionItemKind.Function
 
-  for name, data in pairs(config.strategies.chat.slash_commands) do
-    if name ~= "opts" then
-      table.insert(items, {
+  local slash_commands = vim
+    .iter(config.strategies.chat.slash_commands)
+    :filter(function(name)
+      return name ~= "opts"
+    end)
+    :map(function(name, data)
+      return {
         label = "/" .. name,
         kind = kind,
         detail = data.description,
@@ -34,12 +38,34 @@ function source:complete(params, callback)
           bufnr = params.context.bufnr,
           cursor = params.context.cursor,
         },
-      })
-    end
-  end
+      }
+    end)
+    :totable()
+
+  local prompts = vim
+    .iter(config.prompt_library)
+    :filter(function(_, v)
+      return v.opts and v.opts.is_slash_cmd and v.strategy == "chat"
+    end)
+    :map(function(_, v)
+      return {
+        label = "/" .. v.opts.short_name,
+        kind = kind,
+        detail = v.description,
+        config = v,
+        from_prompt_library = true,
+        context = {
+          bufnr = params.context.bufnr,
+          cursor = params.context.cursor,
+        },
+      }
+    end)
+    :totable()
+
+  local all_items = vim.tbl_extend("force", slash_commands, prompts)
 
   callback({
-    items = items,
+    items = all_items,
     isIncomplete = false,
   })
 end
@@ -50,9 +76,20 @@ end
 ---@return nil
 function source:execute(item, callback)
   vim.api.nvim_set_current_line("")
-
   item.Chat = require("codecompanion").buf_get_chat(item.context.bufnr)
-  SlashCommands:execute(item)
+
+  if item.from_prompt_library then
+    local prompts = strategy.evaluate_prompts(item.config.prompts, item.context)
+    vim.iter(prompts):each(function(prompt)
+      if prompt.role == "system" then
+        item.Chat:add_message(prompt, { visible = false })
+      elseif prompt.role == "user" then
+        item.Chat:append_to_buf(prompt)
+      end
+    end)
+  else
+    SlashCommands:execute(item)
+  end
 
   callback(item)
   vim.bo[item.context.bufnr].buflisted = false

lua/codecompanion/actions/prompt_library.lua

@@ -1,7 +1,3 @@
-local Strategy = require("codecompanion.strategies")
-local context_utils = require("codecompanion.utils.context")
-local api = vim.api
-
 local M = {}
 
 local _prompts = {}
@@ -14,6 +10,7 @@ local _prompts = {}
 function M.resolve(context, config)
   local sort_index = true
 
+  --TODO: Replace with vim.iter()
   for name, prompt in pairs(config.prompt_library) do
     if
       not config.display.action_palette.opts.show_default_prompt_library and (prompt.opts and prompt.opts.is_default)
@@ -35,7 +32,7 @@ function M.resolve(context, config)
       description = prompt.description(context)
     end
     if prompt.opts and prompt.opts.slash_cmd then
-      description = "(/" .. prompt.opts.slash_cmd .. ") " .. description
+      description = description
     end
 
     table.insert(_prompts, {
@@ -60,87 +57,4 @@ function M.resolve(context, config)
   return _prompts
 end
 
----@param desc table
----@return string|function|nil
-local function resolve_description(desc)
-  if type(desc) == "string" then
-    return desc
-  end
-  if type(desc) == "function" then
-    return desc()
-  end
-end
-
----Setup the keymap
----@param map_config table
----@param mode string
----@return nil
-local function map(map_config, mode)
-  return api.nvim_set_keymap(mode, map_config.opts.mapping, "", {
-    callback = function()
-      local context = context_utils.get(api.nvim_get_current_buf())
-
-      return Strategy.new({
-        context = context,
-        selected = map_config,
-      }):start(map_config.strategy)
-    end,
-    desc = resolve_description(map_config.description),
-    noremap = true,
-    silent = true,
-  })
-end
-
----Setup the keymaps for the prompt library
----@param config table
----@return nil
-function M.setup_keymaps(config)
-  if not config.opts.set_prompt_library_keymaps then
-    return
-  end
-
-  local prompts = config.prompt_library
-
-  for _, prompt in pairs(prompts) do
-    if prompt.opts and prompt.opts.mapping then
-      if not config.display.action_palette.opts.show_default_prompt_library and prompt.opts.is_default then
-        goto continue
-      end
-      if prompt.opts.modes and type(prompt.opts.modes) == "table" then
-        for _, mode in ipairs(prompt.opts.modes) do
-          map(prompt, mode)
-        end
-      else
-        map(prompt, "n")
-      end
-    end
-    ::continue::
-  end
-end
-
----Setup the inline slash commands for the prompt library
----@param config table
----@return table
-function M.setup_inline_slash_commands(config)
-  local slash_cmds = {}
-  local prompts = config.prompt_library
-
-  for name, prompt in pairs(prompts) do
-    if prompt.opts then
-      if not config.display.action_palette.opts.show_default_prompt_library and prompt.opts.is_default then
-        goto continue
-      end
-
-      if prompt.opts.slash_cmd then
-        prompt.name = name
-        slash_cmds[prompt.opts.slash_cmd] = prompt
-      end
-
-      ::continue::
-    end
-  end
-
-  return slash_cmds
-end
-
 return M