Skip to content

Conversation

aaronsteers
Copy link
Contributor

@aaronsteers aaronsteers commented Oct 10, 2025

Method: full-text search for 'phase' and replace with new instructions everywhere.

Summary by CodeRabbit

  • Refactor

    • Reworked workflow from a rigid phase-based flow to a stepwise, narrated task approach and unified narration guidance across prompts.
    • Updated wording to reference tasks/jobs rather than phases.
  • Documentation

    • Removed Phase 1–3 guidance documents (initial stream read, pagination, remaining streams).
    • Removed the phase-based templates used to describe phase details.
  • Chores

    • Minor wording updates across build tooling and run guidance.

Copy link

👋 Greetings, Airbyte Team Member!

Here are some helpful tips and reminders for your convenience.

Testing This Branch via MCP

To test the changes in this specific branch with an MCP client like Claude Desktop, use the following configuration:

{
  "mcpServers": {
    "connector-builder-mcp-dev": {
      "command": "uvx",
      "args": ["--from", "git+https://github.com/airbytehq/connector-builder-mcp.git@aj/fix/better-agent-coordination", "connector-builder-mcp"]
    }
  }
}

Testing This Branch via CLI

You can test this version of the MCP Server using the following CLI snippet:

# Run the CLI from this branch:
uvx 'git+https://github.com/airbytehq/connector-builder-mcp.git@aj/fix/better-agent-coordination#egg=airbyte-connector-builder-mcp' --help

PR Slash Commands

Airbyte Maintainers can execute the following slash commands on your PR:

  • /autofix - Fixes most formatting and linting issues
  • /poe <command> - Runs any poe command in the uv virtual environment
  • /poe build-connector prompt="Star Wars API" - Run the connector builder using the Star Wars API.

📝 Edit this welcome message.

Copy link

coderabbitai bot commented Oct 10, 2025

📝 Walkthrough

Walkthrough

This PR removes three phase-guidance markdown files and deletes Pydantic phase data models. It replaces explicit phase-oriented prompts with a stepwise narrated task approach via INTERNAL_MONOLOGUE_GUIDANCE and updates several docstrings to use job/task wording. No runtime control-flow or algorithmic behavior changes.

Changes

Cohort / File(s) Summary
Docs: Remove phase guidance
connector_builder_agents/prompts/phase-1-stream-read.md, connector_builder_agents/prompts/phase-2-pagination.md, connector_builder_agents/prompts/phase-3-remaining-streams.md
Deleted Phase 1–3 markdown guidance files.
Prompt composition: Stepwise narration
connector_builder_agents/src/guidance.py
Replaced explicit phase workflow with stepwise narrated task approach; added INTERNAL_MONOLOGUE_GUIDANCE and injected it into manager & developer prompt assembly; adjusted wording for progress and exit criteria.
Agents: Docstring wording
connector_builder_agents/src/agents.py
Updated public function docstrings to remove phase-centric language; no signature or behavior changes.
Phases data models: Removal
connector_builder_agents/src/phases.py
Removed exported Pydantic models Phase1Data, Phase2Data, Phase3Data and related module content.
Runner: Docstring wording
connector_builder_agents/src/run.py
Updated run_manager_developer_build docstring to drop “3-phase” reference; no logic changes.
Tools: Docstring wording
connector_builder_agents/src/tools.py
Updated tool docstrings from “phase” to “job”; no logic changes.

Sequence Diagram(s)

sequenceDiagram
  autonumber
  participant U as User
  participant M as Manager Agent
  participant D as Developer Agent
  participant T as Tools (mark_job_success/failed)
  note over M,D: INTERNAL_MONOLOGUE_GUIDANCE included in both prompts

  U->>M: Start connector build request
  M->>M: Break work into small steps and narrate plan
  loop For each step
    M->>D: Assign next task with context + narration cues
    D->>D: Execute step and narrate progress
    alt Success
      D->>T: mark_job_success
      T-->>D: Ack
      D-->>M: Step results
    else Failure/blocked
      D->>T: mark_job_failed
      T-->>D: Ack
      D-->>M: Error/needs info
    end
    M->>M: Decide next step or finish
  end
  M-->>U: Build outcome and artifacts
Loading

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

Possibly related PRs

Pre-merge checks and finishing touches

✅ Passed checks (3 passed)
Check name Status Explanation
Description Check ✅ Passed Check skipped - CodeRabbit’s high-level summary is enabled.
Title Check ✅ Passed The title succinctly captures the two primary changes in the pull request—enhancing agent coordination and removing references to the three-phase workflow—and aligns directly with the modifications made throughout the codebase and documentation. It is concise, specific, and avoids superfluous details or vague language.
Docstring Coverage ✅ Passed Docstring coverage is 100.00% which is sufficient. The required threshold is 80.00%.
✨ Finishing touches
  • 📝 Generate docstrings
🧪 Generate unit tests (beta)
  • Create PR with unit tests
  • Post copyable unit tests in a comment
  • Commit unit tests in branch aj/fix/better-agent-coordination

📜 Recent review details

Configuration used: CodeRabbit UI

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between f6d5149 and d81baf7.

📒 Files selected for processing (1)
  • connector_builder_agents/src/run.py (1 hunks)
🚧 Files skipped from review as they are similar to previous changes (1)
  • connector_builder_agents/src/run.py

Comment @coderabbitai help to get the list of available commands and usage tips.

@aaronsteers aaronsteers changed the title Aj/fix/better agent coordination feat: improved agent coordination and removal of '3 phases' guidance Oct 10, 2025
Copy link
Contributor

@Copilot Copilot AI left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull Request Overview

This PR refactors the connector building system by removing the rigid "3 phases" approach and replacing it with a more flexible task-based workflow. The change aims to improve agent coordination by allowing for more adaptive and incremental development steps.

  • Removed the entire phases.py module containing Phase1Data, Phase2Data, and Phase3Data models
  • Updated documentation and prompts to use "job" and "task" terminology instead of "phase"
  • Replaced structured phase guidance with flexible step-by-step instructions

Reviewed Changes

Copilot reviewed 8 out of 8 changed files in this pull request and generated no comments.

Show a summary per file
File Description
connector_builder_agents/src/tools.py Updated function docstrings to reference "job" instead of "phase"
connector_builder_agents/src/run.py Removed "3-phase" reference from function docstring
connector_builder_agents/src/phases.py Completely removed file containing phase data models
connector_builder_agents/src/guidance.py Replaced phase-based workflow with flexible task-based instructions
connector_builder_agents/src/agents.py Updated agent descriptions to remove phase references
connector_builder_agents/prompts/phase-3-remaining-streams.md Removed phase-specific prompt file
connector_builder_agents/prompts/phase-2-pagination.md Removed phase-specific prompt file
connector_builder_agents/prompts/phase-1-stream-read.md Removed phase-specific prompt file

Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.

Copy link

github-actions bot commented Oct 10, 2025

PyTest Results (Fast)

0 tests  ±0   0 ✅ ±0   0s ⏱️ ±0s
0 suites ±0   0 💤 ±0 
0 files   ±0   0 ❌ ±0 

Results for commit d81baf7. ± Comparison against base commit 45b1b8c.

♻️ This comment has been updated with latest results.

@aaronsteers
Copy link
Contributor Author

aaronsteers commented Oct 12, 2025

/build-connector

🛠️ Building connector...

Link to job logs.

🟦 Poe command evals run --connector source-jsonplaceholder completed successfully.

✳️ Show/Hide Summary Output

Evaluation Run Summary

Experiment: RXhwZXJpbWVudDo1Mw==

Run Date: 2025-10-12 05:04 UTC

Configuration Value
developer_model gpt-5
manager_model gpt-5
readiness_eval_model gpt-4o

Experiment Stats

Evaluator Mean Min Max
readiness_eval ❌ 0.00 ❌ 0.00 ❌ 0.00
streams_eval ✅ 1.00 ✅ 1.00 ✅ 1.00
Overall 🟡 0.50
Score Level Count Percentage
🟡 Partial (0.5-0.74) 1 100%

Per-Connector Results

Comparing to prior experiment: RXhwZXJpbWVudDo0OA==

Connector Duration readiness_eval streams_eval Overall
source-jsonplaceholder 5m 10s ❌ 0.00 (→) ✅ 1.00 (→) 🟡 0.50 (→)

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

enhancement New feature or request

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant