Skip to main content

Slack Commands

Command syntax​

Slack command text supports named options at the start, followed by free-form prompt text.

Available options:

  • repo=owner/repo — target GitHub repository
  • model=<model-name> — LLM model to use (see /available-models for the list)
  • targetBranch=<branch> — branch Alakai will target when opening a PR
  • sourceBranch=<branch> — branch to read from (defaults to the repository's base branch)
  • ref=<branch> — alias for sourceBranch
  • prompt=<path/to/file.md> — path to an existing prompt file in the repository

Options can appear in any order, but only at the start of the command text. Everything after the last recognized option is treated as the prompt body.

/prompt​

Generates prompt markdown from a task description and replies in Slack.

/prompt Add retry logic to failed webhook delivery handling

With a repository for context:

/prompt repo=my-org/my-repo Add retry logic to failed webhook delivery handling

With a specific model:

/prompt repo=my-org/my-repo model=gpt-4.1-mini Add retry logic to failed webhook delivery handling

What we get back: a formatted prompt markdown block in Slack, ready to review or paste into a PR.

/prompt-pr​

Generates prompt markdown and opens a pull request in the target repository that saves it to prompts/.

Requires repo= and targetBranch=.

/prompt-pr repo=my-org/my-repo targetBranch=feature/retry-logic Add retry logic to failed webhook delivery handling

With a custom source branch:

/prompt-pr repo=my-org/my-repo sourceBranch=develop targetBranch=feature/retry-logic Add retry logic to failed webhook delivery handling

What we get back: the URL of the opened PR and the path to the prompt file inside the repository.

/implement​

Launches an implementation agent against a repository, either from inline text or from an existing prompt file.

From inline text:

/implement repo=my-org/my-repo Add retry logic to failed webhook delivery handling

From an existing prompt file:

/implement repo=my-org/my-repo prompt=prompts/feature__retry-logic.md

With a custom source branch:

/implement repo=my-org/my-repo sourceBranch=develop prompt=prompts/feature__retry-logic.md

What we get back: confirmation that the agent was launched, and the implementation PR URL once it is available.

/init​

Onboards a repository and enables GitHub automations.

/init repo=my-org/my-repo

What we get back:

  • A GitHub App installation link for the repository
  • Confirmation that automations are enabled
  • A PR URL if docs/alakai.md was created

/clickup-ask​

Searches a ClickUp space's documents and answers a question using an AI agent.

/clickup-ask space=Engineering query=what were the objectives for Q3-2024
  • space=<name> — the ClickUp space to search (case-insensitive)
  • query=<question> — the question to answer; everything after query= is the question

First use: if we have not connected our ClickUp account yet, Alakai replies with a personal authorization URL. Open it, complete the ClickUp login, and then run the command again — our account stays connected for future commands.

What we get back: a threaded reply in Slack with the answer and a "Sources" section linking to the ClickUp documents that were consulted.

/clickup-task​

Creates a ClickUp task directly from Slack.

/clickup-task space=Engineering list=Backlog title=Add retry logic description=Summarize the bug and add retries with exponential backoff
  • space=<name> — required
  • list=<name> — required
  • title=<title> — required
  • description=<description> — required

First use: if our ClickUp account is not connected, Alakai replies with an authorization URL. Connect once and all future commands work without prompting again.

What we get back: a Slack thread with the created task title and a direct link to the task in ClickUp.

@Alakai thread mention​

Summarizes an existing Slack thread and creates a ClickUp task from it.

Mention Alakai inside any thread:

@Alakai summarize this thread and create a task in "Engineering" - "Backlog"

Space and list names must be quoted if they contain spaces. If the channel is already mapped to a default ClickUp list by our Alakai administrator, we can omit the destination:

@Alakai summarize this thread and create a task

What we get back: a reply in the same thread with the created task title and a direct link to the ClickUp task.

/available-models​

Returns the list of LLM models we can request with model= in other commands.

/available-models

What we get back: the full list of supported models, with the default model identified.