Slash commands and skills

DAIV responds to slash commands in issue and merge or pull request comments, and some also work in the dashboard chat. Built-in commands have fixed behavior. Skills are packaged instructions that give DAIV expertise for a specific task. You invoke both the same way:

@daiv /<command> [arguments]

Use /help to list everything available where you are.

Built-in commands

These commands have fixed behavior:

Command	Scopes	Description
/help	Issues, MRs/PRs, dashboard chat	Lists all available commands and skills
/agents	Issues, MRs/PRs, dashboard chat	Lists available subagents with descriptions
/clear	Issues, MRs/PRs	Resets the conversation context to start fresh

/help

@daiv /help

Lists the slash commands and skills available in the current scope: the built-in commands, and the built-in and custom skills of the repository.

/agents

@daiv /agents

Lists the available subagents with their names and descriptions. See Subagents for details.

/clear

@daiv /clear

Starts a fresh conversation context for the current issue or merge/pull request. Use it when DAIV's context is stale or DAIV starts drifting. The earlier conversation stays in the session's history; DAIV stops seeing it.

In the dashboard chat

/help and /agents are also registered for the global scope, so they work in the dashboard chat too. Typing / in the chat composer opens a list of suggestions with those commands and the global skills. Per-repository skills do not appear there: they live in the repository's sandbox, which DAIV cannot read while the page renders.

Built-in skills

DAIV ships with these built-in skills:

Skill	Description
/plan	Explores the codebase in read-only mode and produces an implementation plan
/code-review	Reviews a merge/pull request for correctness, security, performance, structure, changed tests, and your repository's custom review rules
/security-audit	Audits code for security vulnerabilities, injection flaws, hardcoded secrets, and risks
/init	Analyzes the repository and generates or updates an AGENTS.md guidance file
/skill-creator	Guides you through creating a new custom skill

/plan

@daiv /plan implement rate limiting for the API

DAIV explores the codebase without making any changes and posts a detailed implementation plan. DAIV runs it automatically when it addresses an issue without the daiv-auto label (see Issue addressing).

/code-review

@daiv /code-review

Reviews the current merge/pull request diff for correctness, security, performance, structure, changed tests (when the diff touches test files), and your repository's custom review rules. Posts numbered findings grouped by severity, as a single report comment.

How it scores on the ReviewBench benchmark, and at what cost, is on the Benchmarks page.

/security-audit

@daiv /security-audit

Performs a dedicated security scan of the code changes, looking for injection flaws, hardcoded secrets, authentication issues, and data handling risks.

/init

@daiv /init

Analyzes the repository structure and generates an AGENTS.md file: short, repository-specific guidance for AI coding agents, covering verified commands, a repository map, invariants and pitfalls, and where changes usually go. If the file already exists, /init updates it to cut duplication and low-signal text.

/skill-creator

@daiv /skill-creator

Walks you through creating a custom skill for your repository. See Agent skills for more on custom skills.

Turn off slash commands

Slash commands are on by default. To turn them off for a repository, add this to its .daiv.yml:

slash_commands:
  enabled: false

Repository configuration lists every feature toggle.

Create a custom command

A custom slash command is a skill: creating the skill registers the command, with no separate step. To create a command named /my-command:

1. Create .agents/skills/my-command/SKILL.md, starting with YAML frontmatter.
2. In a comment, invoke it with @daiv /my-command.

The command then appears in /help next to the built-in ones. Agent skills explains how to write a skill.