Skip to content

demo.md

Latest commit

 

History

History
135 lines (97 loc) · 4.8 KB

demo.md

File metadata and controls

135 lines (97 loc) · 4.8 KB

Remote Dev Bot — Demo

See what remote-dev-bot does before you install it. This page walks through real examples from the remote-dev-bot repository itself.

What You'll See

When you trigger remote-dev-bot on an issue:

  1. Trigger — You comment /agent-resolve (or /agent-design, /agent-review)
  2. Acknowledgment — A 🚀 reaction appears on your comment
  3. Processing — A GitHub Actions workflow runs (visible in the Actions tab)
  4. Result — A PR is created (resolve), a comment is posted (design/review), or both

The examples below show this flow in action.

Example 1: Simple Resolve

The issue: Issue #33 asked for documentation of model names — a straightforward documentation task.

What happened:

  1. Someone commented /agent-resolve on the issue
  2. The agent read the issue, explored the codebase, and found where model documentation belonged
  3. The agent created PR #52 with the changes
  4. The PR was reviewed and merged

What to look at:

  • The issue description — clear, specific request
  • The PR diff — the agent's implementation
  • The PR description — the agent explains what it did and why

Takeaway: For well-defined tasks, the agent can go from issue to merged PR with minimal human intervention.

Example 2: Design Then Resolve

The issue: Issue #124 asked whether commands should be case-insensitive (to handle mobile autocorrect). This needed design analysis before implementation.

What happened:

  1. Someone commented /agent-design to get analysis first
  2. The agent posted a detailed design analysis as a comment — exploring tradeoffs, suggesting an approach
  3. A human reviewed the analysis and agreed with the recommendation
  4. Someone commented /agent-resolve to implement
  5. The agent created PR #131, which was merged

What to look at:

  • The design comment — the agent's analysis of the problem
  • The human response agreeing with the approach
  • The resulting PR — implementation matching the design

Takeaway: Use /agent-design when you want to think through a problem before committing to an implementation. The agent explores the codebase and gives you a recommendation you can accept, modify, or reject.

Example 3: Resolve With Feedback

The issue: Issue #95 asked about preventing agent loops — a security-sensitive feature.

What happened:

  1. Someone commented /agent-resolve on the issue
  2. The agent created PR #109 with an initial implementation
  3. A reviewer pointed out a regex bypass vulnerability in the implementation
  4. Someone commented /agent-resolve on the PR (not the issue) to incorporate the feedback
  5. The agent fixed the vulnerability and pushed new commits to the same PR
  6. The PR was merged

What to look at:

  • The original PR — first implementation attempt
  • The reviewer comment — identifying the security issue
  • The commit history — showing the agent's fix after feedback

Takeaway: The agent can iterate on its own PRs. Comment /agent-resolve on a PR (not the original issue) to have the agent incorporate review feedback. This is the normal code review workflow — just with an AI making the fixes.

The Three Modes

Mode Command What it does
Resolve /agent-resolve Implements changes and opens a PR
Design /agent-design Analyzes the problem and posts a comment
Review /agent-review Reviews a PR and posts feedback

All modes support model variants: /agent-resolve-claude-large, /agent-design-gpt-large, etc.

Ready to Install?

Now that you've seen what remote-dev-bot does, head back to install.md to set it up on your own repository.

After installation, see onboarding.md for hands-on exercises that walk you through using each mode on your own repo.