fix(backend): standardize microservice host/port configuration

[majdyz]

Summary

This PR fixes inconsistencies in microservice host and port configuration defaults in settings.py and .env.default.

Changes Made

1. Agent Generator Host Configuration

   • Changed agentgenerator_host default from "" (empty) to "localhost"
   • Makes it consistent with other client services: rabbitmq_host, redis_host, clamav_service_host
   • Fixes the service_not_configured error when only AGENTGENERATOR_PORT is set

2. Agent Generator Port Configuration

   • Changed agentgenerator_port default from 8000 to 8009
   • Avoids conflict with Kong gateway (which uses port 8000)
   • Follows the sequential port allocation pattern for internal services (8001-8008)

3. Environment File Updates

   • Added AGENTGENERATOR_HOST=localhost to .env.default for clarity
   • Added documentation comment explaining the Agent Generator service

Configuration Pattern Analysis

Client Services (connect TO external services):

• Use "localhost" as default host
• Examples: rabbitmq_host, redis_host, clamav_service_host, agentgenerator_host

Server Services (LISTEN on all interfaces):

• Use "0.0.0.0" as default host
• Examples: websocket_server_host, agent_api_host

Port Allocation:

• 8001: websocket_server
• 8002: execution_manager
• 8003: execution_scheduler
• 8005: database_api
• 8006: agent_api
• 8007: notification_service
• 8008: copilot_executor
• 8009: agentgenerator (NEW - avoids 8000 Kong conflict)

Impact

• Breaking Change: Minimal - only affects users who haven't explicitly set AGENTGENERATOR_HOST or AGENTGENERATOR_PORT
• Bug Fix: Resolves "service_not_configured" error when trying to use Agent Generator with only port configured
• Consistency: Aligns with established patterns for other microservice configurations

Testing

• ✅ All existing tests pass (24/24 in test/agent_generator/test_service.py)
• ✅ Pre-commit hooks pass
• Configuration logic properly handles new defaults

Motivation

User reported getting service_not_configured error even after setting AGENTGENERATOR_PORT=8009. Investigation revealed:

1. The check bool(settings.config.agentgenerator_host) returns False when host is empty string
2. Port 8000 conflicts with Kong gateway
3. Inconsistent patterns across microservice configurations

This PR standardizes the configuration to follow established patterns and avoid conflicts.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

📜 Recent review details

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Disabled knowledge base sources:

• Linear integration is disabled

You can enable these sources in your CodeRabbit configuration.

📥 Commits

Reviewing files that changed from the base of the PR and between e40c8c7 and 11cfd87.

📒 Files selected for processing (2)

• autogpt_platform/backend/.env.default
• autogpt_platform/backend/backend/util/settings.py

🧰 Additional context used

📓 Path-based instructions (7)

autogpt_platform/backend/.env*

📄 CodeRabbit inference engine (autogpt_platform/CLAUDE.md)

Backend configuration: Use backend/.env.default for default values (tracked in git) and backend/.env for user-specific overrides (gitignored)

Files:

• autogpt_platform/backend/.env.default

autogpt_platform/**/.env*

📄 CodeRabbit inference engine (autogpt_platform/CLAUDE.md)

Platform configuration: Use .env.default for Supabase/shared defaults (tracked in git) and .env for user-specific overrides (gitignored)

Files:

• autogpt_platform/backend/.env.default

autogpt_platform/backend/**/{.env.default,.env}

📄 CodeRabbit inference engine (autogpt_platform/backend/CLAUDE.md)

Use .env.default for default environment configuration and .env for user overrides in the backend

Files:

• autogpt_platform/backend/.env.default

autogpt_platform/backend/**/*.py

📄 CodeRabbit inference engine (.github/copilot-instructions.md)

autogpt_platform/backend/**/*.py: Use Python 3.11 (required; managed by Poetry via pyproject.toml) for backend development
Always run 'poetry run format' (Black + isort) before linting in backend development
Always run 'poetry run lint' (ruff) after formatting in backend development

Files:

• autogpt_platform/backend/backend/util/settings.py

autogpt_platform/backend/**/*.{py,txt}

📄 CodeRabbit inference engine (autogpt_platform/backend/CLAUDE.md)

Use poetry run prefix for all Python commands, including testing, linting, formatting, and migrations

Files:

• autogpt_platform/backend/backend/util/settings.py

autogpt_platform/backend/backend/**/*.py

📄 CodeRabbit inference engine (autogpt_platform/backend/CLAUDE.md)

Use Prisma ORM for database operations in PostgreSQL with pgvector for embeddings

Files:

• autogpt_platform/backend/backend/util/settings.py

autogpt_platform/**/*.py

📄 CodeRabbit inference engine (AGENTS.md)

Format Python code with poetry run format

Files:

• autogpt_platform/backend/backend/util/settings.py

🧠 Learnings (7)

📚 Learning: 2026-02-04T16:49:56.189Z

Learnt from: CR
Repo: Significant-Gravitas/AutoGPT PR: 0
File: autogpt_platform/CLAUDE.md:0-0
Timestamp: 2026-02-04T16:49:56.189Z
Learning: Applies to autogpt_platform/backend/.env* : Backend configuration: Use `backend/.env.default` for default values (tracked in git) and `backend/.env` for user-specific overrides (gitignored)

Applied to files:

• autogpt_platform/backend/.env.default

📚 Learning: 2026-02-04T16:50:20.508Z

Learnt from: CR
Repo: Significant-Gravitas/AutoGPT PR: 0
File: autogpt_platform/backend/CLAUDE.md:0-0
Timestamp: 2026-02-04T16:50:20.508Z
Learning: Applies to autogpt_platform/backend/**/{.env.default,.env} : Use `.env.default` for default environment configuration and `.env` for user overrides in the backend

Applied to files:

• autogpt_platform/backend/.env.default

📚 Learning: 2026-02-04T16:50:33.615Z

Learnt from: CR
Repo: Significant-Gravitas/AutoGPT PR: 0
File: autogpt_platform/frontend/CLAUDE.md:0-0
Timestamp: 2026-02-04T16:50:33.615Z
Learning: Applies to autogpt_platform/frontend/**/.env* : Configure environment with `.env.default` (defaults) → `.env` (user overrides)

Applied to files:

• autogpt_platform/backend/.env.default

📚 Learning: 2026-02-04T16:49:56.189Z

Learnt from: CR
Repo: Significant-Gravitas/AutoGPT PR: 0
File: autogpt_platform/CLAUDE.md:0-0
Timestamp: 2026-02-04T16:49:56.189Z
Learning: Applies to autogpt_platform/**/docker-compose*.yml : All Docker services use hardcoded defaults in docker-compose files without `${VARIABLE}` substitutions; use `env_file` directive to load variables into containers at runtime

Applied to files:

• autogpt_platform/backend/.env.default

📚 Learning: 2026-02-04T16:49:56.189Z

Learnt from: CR
Repo: Significant-Gravitas/AutoGPT PR: 0
File: autogpt_platform/CLAUDE.md:0-0
Timestamp: 2026-02-04T16:49:56.189Z
Learning: Applies to autogpt_platform/**/.env* : Platform configuration: Use `.env.default` for Supabase/shared defaults (tracked in git) and `.env` for user-specific overrides (gitignored)

Applied to files:

• autogpt_platform/backend/.env.default

📚 Learning: 2026-02-04T16:49:56.189Z

Learnt from: CR
Repo: Significant-Gravitas/AutoGPT PR: 0
File: autogpt_platform/CLAUDE.md:0-0
Timestamp: 2026-02-04T16:49:56.189Z
Learning: Applies to autogpt_platform/frontend/.env* : Frontend configuration: Use `frontend/.env.default` for default values (tracked in git) and `frontend/.env` for user-specific overrides (gitignored)

Applied to files:

• autogpt_platform/backend/.env.default

📚 Learning: 2026-02-04T16:49:56.189Z

Learnt from: CR
Repo: Significant-Gravitas/AutoGPT PR: 0
File: autogpt_platform/CLAUDE.md:0-0
Timestamp: 2026-02-04T16:49:56.189Z
Learning: Applies to autogpt_platform/**/*agent*graph*.json : Agent Graphs are workflow definitions stored as JSON and executed by the backend

Applied to files:

• autogpt_platform/backend/.env.default

⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (8)

• GitHub Check: types
• GitHub Check: Greptile Review
• GitHub Check: Seer Code Review
• GitHub Check: test (3.11)
• GitHub Check: test (3.13)
• GitHub Check: test (3.12)
• GitHub Check: Analyze (python)
• GitHub Check: Check PR Status

🔇 Additional comments (3)

autogpt_platform/backend/backend/util/settings.py (2)

370-373: Port 8009 is consistent with the existing internal service port allocation.

The old default (8000) conflicted with Kong/Supabase (SUPABASE_URL=http://localhost:8000, PLATFORM_BASE_URL=http://localhost:8000). Port 8009 follows naturally after copilot_executor_port: 8008 with no conflicts.

---

366-369: Breaking change: Default agentgenerator_host value change from empty string to "localhost" requires Agent Generator service.

The empty-string guard in service.py line 121 (bool(settings.config.agentgenerator_host)) still exists. When is_external_service_configured() returns False (which occurs only if both agentgenerator_host is empty AND dummy mode is disabled), any agent generation operation raises AgentGeneratorNotConfiguredError (see core.py lines 101-105).

Changing the default from "" to "localhost" means all deployments using the default will now attempt to connect to an external service on localhost:8009. Any deployment without Agent Generator running will fail with connection errors on agent generation requests, rather than gracefully disabling the feature.

Confirm that either:

1. Agent Generator service is always deployed and running on localhost:8009, or
2. Users explicitly configure AGENTGENERATOR_HOST to empty string (or set AGENTGENERATOR_USE_DUMMY=true) to disable agent generation features.

autogpt_platform/backend/.env.default (1)

193-199: LGTM — additions follow the existing section pattern and are consistent with settings.py defaults.

The blank-line separator, section comment block, and both AGENTGENERATOR_HOST/AGENTGENERATOR_PORT entries mirror how REDIS_HOST/REDIS_PORT and other service pairs are structured in the file.

---

Walkthrough

Updates Agent Generator Service configuration by changing the default host from empty string to "localhost" and the default port from 8000 to 8009 across both the environment configuration file and the settings module.

Changes

Cohort / File(s)	Summary
Agent Generator Service Configuration autogpt_platform/backend/.env.default, autogpt_platform/backend/backend/util/settings.py	Updates default host and port for Agent Generator Service: host changed from "" to "localhost", port changed from 8000 to 8009, and host description simplified.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~5 minutes

Poem

🐰 Configuration hops along the way,
New defaults set for localhost's day,
Port 8009 now takes its place,
Agent Generator runs with grace!

🚥 Pre-merge checks | ✅ 3

✅ Passed checks (3 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title accurately summarizes the main change: standardizing microservice host/port configuration for the Agent Generator service.
Description check	✅ Passed	The description is comprehensive and directly relates to the changeset, detailing the specific configuration updates, rationale, and impact.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

• 📝 Generate docstrings (stacked PR)
• 📝 Generate docstrings (commit on current branch)

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment
• Commit unit tests in branch fix/microservice-config-consistency

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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

[github-actions]

🔍 PR Overlap Detection

This check compares your PR against all other open PRs targeting the same branch to detect potential merge conflicts early.

🔴 Merge Conflicts Detected

The following PRs have been tested and will have merge conflicts if merged after this PR. Consider coordinating with the authors.

• feat(copilot): add dummy agent generator for testing #12071 (Otto-AGPT · updated 8d ago)

🟢 Low Risk — File Overlap Only

These PRs touch the same files but different sections (click to expand)

• feat(copilot): Replace bubblewrap with e2b sandbox for unified filesystem #12162 (Otto-AGPT · updated 17h ago)

  • Shared files: autogpt_platform/backend/backend/util/settings.py

• feat(backend): pre-populate CoPilotUnderstanding from Tally form on signup #12119 (Swiftyos · updated 2d ago)

  • autogpt_platform/backend/.env.default: L192-197

Summary: 1 conflict(s), 0 medium risk, 2 low risk (out of 3 PRs with file overlap)

---

Auto-generated on push. Ignores: openapi.json, lock files.

[greptile-apps]

Greptile Summary

This PR fixes a configuration bug in the Agent Generator service defaults and standardizes the microservice configuration pattern.

Changes:

• Changed agentgenerator_host default from "" (empty string) to "localhost" to align with other client services (rabbitmq_host, redis_host, clamav_service_host)
• Changed agentgenerator_port default from 8000 to 8009 to avoid conflict with Kong gateway and follow the sequential port allocation pattern
• Added AGENTGENERATOR_HOST=localhost and AGENTGENERATOR_PORT=8009 to .env.default with documentation

Impact:

• Fixes the service_not_configured error that occurred when only AGENTGENERATOR_PORT was set (because bool("") returns False in is_external_service_configured())
• Eliminates port conflict with Kong gateway which uses port 8000
• Maintains consistency with the established pattern: client services use "localhost" as default host, server services use "0.0.0.0"
• Port allocation follows the existing sequence: 8001-8003, 8005-8008, and now 8009

Confidence Score: 5/5

• This PR is safe to merge with minimal risk - it's a straightforward configuration fix that follows established patterns
• The changes are minimal, well-justified, and follow existing patterns in the codebase. The fix addresses a real bug (empty string causing service_not_configured errors), eliminates a port conflict with Kong gateway, and standardizes configuration. All existing tests pass (24/24), and the changes are backward-compatible since they only affect default values that users can override via environment variables.
• No files require special attention

_{Last reviewed commit: 11cfd87}

[autogpt-reviewer]

📋 PR Review Squad — PR #12189: standardize microservice host/port configuration

Specialist Verdicts

Reviewer	Verdict	Key Finding
🛡️ Security	✅ APPROVE	No security concerns. Port fix avoids accidental cross-service communication.
🏗️ Architecture	⚠️ REQUEST CHANGES	Default "localhost" breaks is_external_service_configured() feature gate.
⚡ Performance	✅ APPROVE	No startup delays. Connection-refused on localhost is instantaneous.
🧪 Testing	✅ APPROVE	All 24 tests mock settings — immune to default changes. Tests adequate.
📖 Quality	⚠️ REQUEST CHANGES	Same feature gate blocker. Port 8009 gap (8008 skipped) undocumented.
📦 Product	⚠️ REQUEST CHANGES	Silent breaking change for ~95% of self-hosted deployments.
📬 Discussion	ℹ️ INFO	No human reviews yet. Merge conflict with #12071 on settings.py.
🔎 QA Validator	⚠️ REQUEST CHANGES	Same feature gate blocker. No docker-compose entry for agent generator.

Blocker (Must Fix Before Merge)

1. Default host "" → "localhost" breaks is_external_service_configured() feature gate — The function at service.py:93-96 uses bool(settings.config.agentgenerator_host) as a feature gate. The old empty-string default was intentionally falsy → service disabled by default → clean "not configured" responses. The new "localhost" default makes the gate always return True, so every deployment without the Agent Generator microservice (most/all current deployments) will get connection-refused errors instead of clean "Agent Generator not configured" messages.

   Recommended fix (least invasive): Keep agentgenerator_host default as "" in settings.py. Only set localhost in .env.default (either active or commented out). This preserves opt-in behavior while documenting the expected config. The port change to 8009 can ship as-is.

   Alternatively, update is_external_service_configured() to use an explicit agentgenerator_enabled: bool field.

What's Good ✅

• Port 8000 → 8009 — Correctly avoids collision with Supabase/Kong (port 8000). Follows the sequential 800x internal service pattern.
• .env.default documentation — Clear section header and consistent format.
• CI all green — All Python tests pass (3.11/3.12/3.13), types pass, CLA signed.

Should Fix

2. Port 8009 gap — Port 8008 is assigned to copilot_executor, 8009 to agentgenerator. Port 8004 (agent_server) may be skipped or deprecated. A comment in settings.py would help.
3. Merge conflict with #12071 — Both PRs modify agent generator settings in settings.py. Coordinate merge order.

Risk Assessment

• Merge risk: LOW (if host default is reverted to "")
• Rollback difficulty: EASY

---

🎯 Final Verdict: REQUEST CHANGES

The port standardization (8000→8009) is a good fix. However, changing agentgenerator_host default from "" to "localhost" silently flips a feature gate that affects all deployments. Keep the code default as empty string and only set localhost in .env.default for documentation/opt-in.

@Pwuts @Bentlybro — tiny PR, one blocker: the host default change breaks the feature gate in is_external_service_configured(). Port fix is fine to ship once the host default is reverted.

[autogpt-reviewer]

📋 Final Review Summary

PR: #12189 — fix(backend): standardize microservice host/port configuration

Author: majdyz | HEAD: 11cfd8756c | CI: ALL GREEN | CLA: ✅ Signed
Changes: 1 commit, 2 files, 32 lines

---

What Changed

1. agentgenerator_host default: "" → "localhost" — aligns with rabbitmq_host, redis_host, clamav_service_host
2. agentgenerator_port default: 8000 → 8009 — avoids Kong gateway conflict (port 8000), follows sequential allocation (8001-8008 taken)
3. .env.default — Added AGENTGENERATOR_HOST and AGENTGENERATOR_PORT with documentation comment

Specialist Verdicts

Reviewer	Verdict	Key Finding
🛡️ Security	✅	localhost binding is correct for client service pattern
🏗️ Architecture	✅	Consistent with established config patterns
⚡ Performance	✅	No performance implications — config only
🧪 Testing	✅	24 existing tests pass
📖 Quality	✅	Clean, well-documented change
📦 Product	✅	Fixes "service_not_configured" error
📬 Discussion	✅	No prior review threads
🔎 QA Validator	✅	Config-only, no runtime behavior change

Risk Assessment

• Merge risk: LOW — config defaults only
• Rollback difficulty: TRIVIAL — single commit revert
• Breaking change: Minimal — only affects users without explicit env vars (and the old defaults were broken anyway)

---

🎯 Final Verdict: APPROVE

Recommendation

Straightforward config fix that resolves a real bug (port 8000 conflict with Kong gateway) and aligns with established patterns. The empty-string host default was causing "service_not_configured" errors when only the port was set. Clean, well-documented, CI green.

For Reviewers

@Pwuts @Bentlybro Ready to merge. Config-only, no blockers.

---

Automated review by PR Review Squad (8 specialists) • HEAD: 11cfd87

[autogpt-reviewer]

⚠️ Review Correction — REQUEST CHANGES

Previous APPROVE Withdrawn

My earlier APPROVE was premature. The security specialist identified a blocker that I missed, which was also flagged in a prior review:

🔴 BLOCKER — agentgenerator_host default "" → "localhost" breaks feature gate

File: settings.py:366-367

The is_external_service_configured() gate uses bool(settings.config.agentgenerator_host) to decide whether to route to an external Agent Generator. Changing the default from "" to "localhost" makes this gate always return True, even for deployments that never configured an agent generator.

Impact:

• Every deployment will attempt HTTP calls to localhost:8009 (which won't be listening in standard deployments)
• Converts a clean AgentGeneratorNotConfiguredError into opaque connection timeouts
• SSRF-adjacent risk in shared-host environments

Fix: Keep agentgenerator_host default as "" in settings.py. Set localhost only in .env.default as a suggested dev value. This preserves the feature gate while documenting the correct value.

✅ Port change is good

8000 → 8009 correctly avoids Kong gateway conflict. Keep this.

---

🎯 Revised Verdict: REQUEST_CHANGES

The port fix is correct but the host default change breaks the feature gate. One-line fix needed.

---

Automated review by PR Review Squad • Correction after specialist analysis

[autogpt-reviewer]

📋 Final Review Summary

PR: #12189 — "fix(backend): standardize microservice host/port configuration"

Specialist Verdicts

Reviewer	Verdict	Key Finding
🛡️ Security	✅ Pass	No secrets, no auth changes
🏗️ Architecture	⚠️ Concern	Feature gate broken by default change
⚡ Performance	✅ Pass	Config-only change, no runtime impact
🧪 Testing	⚠️ Concern	No test verifying feature gate behavior
📖 Quality	✅ Pass	Good .env.default documentation added
📦 Product	⚠️ Concern	Silent behavior change for all deployments
📬 Discussion	⚠️ Flagged	Original REQUEST CHANGES blocker unaddressed; no author response
🔎 QA Validator	✅ Pass	Backend-only config, no UI impact

Blockers (Must Fix Before Merge)

1. Feature gate broken by default change (settings.py:362)
   • agentgenerator_host default changed from "" → "localhost"
   • is_external_service_configured() returns bool(settings.config.agentgenerator_host) — this now always returns True
   • Every deployment will attempt to connect to an external Agent Generator service, causing connection errors when none is running
   • Used in core.py:102 and service.py:478 to gate external vs built-in behavior
   • Fix: Keep the code default as "" in settings.py. Set localhost only in .env.default (which already does that correctly in this PR)

Good Parts ✅

• Port fix 8000→8009 is correct and uncontroversial
• .env.default additions with documentation are well-done
• The .env.default approach is the right pattern — explicit opt-in per deployment

Risk Assessment

• Merge risk: HIGH — silently breaks feature gate for all deployments without Agent Generator
• Rollback difficulty: EASY — 2-file revert

---

🎯 Final Verdict: REQUEST_CHANGES

Recommendation

The port fix and .env.default additions are good, but changing the code default in settings.py from "" to "localhost" breaks the is_external_service_configured() feature gate. This is the same blocker flagged in the first review and by CodeRabbit — it remains unaddressed. The fix is simple: revert agentgenerator_host default to "" in settings.py while keeping localhost in .env.default.

For Nick

@ntindle Small 2-file PR with a good port fix (8000→8009) but the host default change breaks the Agent Generator feature gate. One-line fix needed — keep code default as "", env default as localhost. Author hasn't responded to any feedback yet.

[autogpt-reviewer]

📋 PR Review Squad — Re-review #3 (2026-02-24)

Status: No changes since last review

HEAD: 11cfd8756c (unchanged since Feb 21) | CI: All green | Author response: None

---

🔴 BLOCKER — Still Outstanding

agentgenerator_host default "" → "localhost" breaks is_external_service_configured() feature gate

• service.py:93-96 uses bool(settings.config.agentgenerator_host) as a feature gate
• "localhost" is truthy → gate always returns True → every deployment attempts to connect to Agent Generator
• Deployments without Agent Generator get connection errors instead of clean "not configured" responses

Fix (one line): Revert agentgenerator_host default to "" in settings.py. Keep localhost in .env.default (which is correct as-is).

✅ Good Parts (unchanged)

• Port 8000 → 8009 avoids Kong conflict — correct
• .env.default documentation — well-structured

📊 Review History

Date	Verdict	Notes
Feb 21	REQUEST_CHANGES	Feature gate blocker identified by 4/8 specialists
Feb 22	REQUEST_CHANGES	Same blocker, no author response
Feb 24	REQUEST_CHANGES	No changes, no response — 3 days stale

---

🎯 Verdict: REQUEST_CHANGES

Same one-line fix needed. PR is going stale with no author engagement.

@majdyz — friendly ping: the fix is a single line change (revert the host default to "" in settings.py). The rest of the PR is good to go once that's addressed.

[autogpt-reviewer]

📋 PR Review Squad — Re-review #4 (2026-02-24)

PR: #12189 — "fix(backend): standardize microservice host/port configuration"

Author: majdyz | HEAD: 11cfd8756ce2 (unchanged since Feb 21) | CI: All green | CLA: ✅

---

Specialist Verdicts (8/8 reported)

Reviewer	Verdict	Key Finding
🛡️ Security	❌ REQUEST CHANGES	Feature gate bypass enables blind SSRF — unauthenticated HTTP POSTs to localhost:8009 with user data
🏗️ Architecture	❌ REQUEST CHANGES	Breaks opt-in contract — 19 call sites depend on the empty-string feature gate
⚡ Performance	⚠️ CONCERN	Instant fail (~0ms) becomes 600s connection timeout for deployments without Agent Generator
🧪 Testing	❌ REQUEST CHANGES	No test covers feature gate with actual defaults — change is invisible to CI
📖 Quality	❌ REQUEST CHANGES	(empty to use built-in) description removed, erasing important usage context
📦 Product	❌ REQUEST CHANGES	Silent breaking change for all deployments without Agent Generator
📬 Discussion	ℹ️ INFO	No author response in 3+ days across 4 review cycles. No human maintainer reviews. Merge conflict with #12071
🔎 QA Validator	⚠️ CONCERN	UX regression: users see opaque connection errors instead of clean "Agent Generator not configured" messages

---

🔴 BLOCKER — agentgenerator_host default "" → "localhost" breaks feature gate

Impact: Every deployment without Agent Generator (majority of deployments)

is_external_service_configured() in service.py:93-96 uses bool(settings.config.agentgenerator_host) as a feature gate:

• Old default "" → bool("") = False → feature disabled → clean AgentGeneratorNotConfiguredError
• New default "localhost" → bool("localhost") = True → feature always enabled → connection errors

Consequences:

1. Copilot agent tools (create_agent, edit_agent, customize_agent) attempt HTTP calls to non-existent localhost:8009
2. Users see opaque connection timeouts (up to 600s) instead of helpful "not configured" messages
3. Unauthenticated HTTP POSTs with user data sent to whatever occupies port 8009
4. Feature gate that 19 call sites depend on is silently bypassed

Fix (one line): Revert agentgenerator_host default to "" in settings.py. Keep localhost in .env.default for dev documentation.

agentgenerator_host: str = Field(
    default="",  # ← keep empty for feature gate
    description="The host for the Agent Generator service (empty to use built-in)",
)

✅ What's Good

• Port 8000 → 8009 — Correctly avoids Kong gateway conflict, follows sequential allocation
• .env.default documentation — Well-structured, consistent format
• CI all green — Tests pass on Python 3.11/3.12/3.13

⚠️ Should Fix

• Restore description text (empty to use built-in) on the agentgenerator_host field
• Consider commenting out values in .env.default to prevent accidental opt-in when copying

📊 Review History

Date	Verdict	Notes
Feb 21	REQUEST_CHANGES	Feature gate blocker first identified
Feb 22	REQUEST_CHANGES	Same blocker (after corrected premature APPROVE)
Feb 22	REQUEST_CHANGES	Full re-review, same finding
Feb 24 03:00	REQUEST_CHANGES	Re-review #3 — no changes, no response
Feb 24 03:07	REQUEST_CHANGES	Re-review #4 — 8 specialists unanimous, same blocker

Risk Assessment

• Merge risk: HIGH as-is (one-line fix makes it LOW)
• Rollback difficulty: TRIVIAL

---

🎯 Final Verdict: REQUEST_CHANGES

The port fix (8000→8009) and .env.default documentation are good. The single blocker remains: changing agentgenerator_host default from "" to "localhost" in settings.py breaks the is_external_service_configured() feature gate. This is a one-line fix.

@majdyz — 4th review, same one-line fix needed. Happy to help if you'd like guidance.

---

Automated review by PR Review Squad (8 specialists) • HEAD: 11cfd87