Custom Tasks & Agentic Mode

Use roborev run to execute custom tasks with AI agents. While automatic reviews focus on commits, run lets you target specific files, ask questions, or perform targeted analysis.

roborev run "Review src/auth.go for security issues"
roborev run --wait "Find simplification opportunities in this codebase"
roborev run --agentic "Add input validation to the user controller"

Use Cases

Targeted File Reviews

Review specific files for particular concerns:

# Security review
roborev run "Review src/auth/ for security vulnerabilities"
roborev run "Check database queries in src/db/ for SQL injection"

# Simplification opportunities
roborev run "Find opportunities to simplify src/handlers/user.go"
roborev run "Identify dead code or unused functions in src/utils/"

# Refactoring candidates
roborev run "Suggest refactoring opportunities in src/api/"
roborev run "Find code duplication in src/services/"

Code Analysis

Ask questions about architecture and design:

roborev run "Explain the authentication flow in this codebase"
roborev run "What design patterns are used here?"
roborev run "Document the main entry points of this application"
roborev run "List all external API dependencies"

Custom Review Criteria

Apply specific review criteria beyond the default review:

roborev run "Review recent changes for OWASP Top 10 vulnerabilities"
roborev run "Check for proper error handling in src/api/"
roborev run "Verify all database connections are properly closed"
roborev run "Find functions exceeding 50 lines"

Making Changes (Agentic Mode)

Use --agentic to allow file modifications:

roborev run --agentic "Add comprehensive error handling to main.go"
roborev run --agentic "Refactor database layer to use connection pooling"
roborev run --agentic "Add input validation to all API endpoints"
roborev run --agentic "Convert callback-style code to async/await"

Piped Input

Pipe complex prompts or instructions from files:

echo "Add comprehensive error handling" | roborev run --agentic --wait
cat review-checklist.txt | roborev run --wait

Flags

Flag	Description
`--wait`	Wait for task to complete and show result
`--agent`	Agent to use (default: from config)
`--reasoning`	Reasoning level: `fast`, `standard`, or `thorough`
`--no-context`	Don’t include repository context in prompt
`--agentic`	Enable agentic mode (allow file edits and commands)
`--yolo`	Alias for `--agentic`
`--quiet`	Suppress output (just enqueue)

Repository Context

By default, tasks include context about the repository:

Repository name and path
Any project guidelines from .roborev.toml

Use --no-context for raw prompts without this context.

Tips

Use --wait to see results immediately in the terminal
Use --reasoning thorough for security-sensitive analysis
Combine with roborev tui to review task results later
Tasks appear in the TUI alongside commit reviews

Review vs Agentic Modes

Agents run in one of two modes depending on the task.

Mode	Tools Available	Used By
Review (default)	Read, Glob, Grep	`roborev review`, `roborev run`
Agentic	Read, Glob, Grep, Edit, Write, Bash	`roborev refine`, `roborev run --agentic`

Review Mode

Review mode is read-only. Agents can inspect code but cannot make changes. This is the safe default for automatic reviews triggered by post-commit hooks.

The agent can read files, search for patterns, and analyze code structure. It cannot edit files, create files, or run commands. No background or async operation (reviews, enqueue, or roborev run without --agentic) ever modifies your working tree.

Agentic Mode

Agentic mode allows agents to edit files and run commands. If you pass --agentic and the agent makes changes to your working tree, that’s an explicit opt-in. roborev will never do this on its own. Enable it in three ways:

Per-job:

roborev run --agentic "Refactor the error handling"
roborev run --yolo "Add input validation"   # --yolo is an alias

Per-command: the refine command automatically enables agentic mode:

roborev refine   # Always runs in agentic mode

Globally:

allow_unsafe_agents = true

Then restart the daemon:

roborev daemon restart

Agent-Specific Flags

Agent	Review Mode	Agentic Mode
Codex	`--sandbox-cmd-allowlist ""`	`--dangerously-bypass-approvals-and-sandbox`
Claude Code	Default behavior	`--dangerously-skip-permissions`
Gemini	Default behavior	`--yolo --allowed-tools`
Copilot	Default behavior	Manual approval required
Cursor	Default behavior	`--yolo`
OpenCode	Default behavior	Auto-approves in non-interactive mode

Security Considerations

When using agentic mode:

The agent can run arbitrary commands on your machine
This includes installing dependencies, running builds, etc.
Safe for your own code on trusted branches
Use isolation for untrusted code (containers, VMs)

See Auto-Fix Agentic Loop Security for detailed guidance.