Refactor pilot architecture and remove dashboard #15

2026-09-01T00:21:54Z

masi commented

2026-09-01 00:21:54 +00:00

Summary

split transport, model, and trusted configuration concerns into focused modules
preserve ai_review as the compatibility facade for existing webhook/CI callers
remove the obsolete dashboard server, data layer, Kubernetes manifest, docs, and tests
document the current pilot architecture and Langfuse as the analytics surface

Verification

python3 -m compileall -q pilot tests
242 focused tests passed
full suite previously passed with 476 tests

## Summary - split transport, model, and trusted configuration concerns into focused modules - preserve `ai_review` as the compatibility facade for existing webhook/CI callers - remove the obsolete dashboard server, data layer, Kubernetes manifest, docs, and tests - document the current pilot architecture and Langfuse as the analytics surface ## Verification - `python3 -m compileall -q pilot tests` - 242 focused tests passed - full suite previously passed with 476 tests

masi added 4 commits 2026-09-01 00:21:54 +00:00

test: trigger review for eval judges d746b1fdc2

pilot(eval): switch rule target from observation to trace 3109bd7c2d

The standard /api/public/ingestion path feeds only the trace-upsert
queue; evalService.createEvalJobs only dispatches targetObject in
{TRACE, DATASET}. Observation rules fire exclusively from the OTel
pipeline, which this pilot does not use. The trace body already carries
review input/output via langfuse_trace, so a trace rule sees the same
material an observation rule would.

test: trigger review after DB rule insert 5b8be61e2f

refactor: split pilot architecture b4041f6892

Remove the obsolete dashboard now that Langfuse is the analytics surface.\nIntroduce focused transport, model, and configuration modules while preserving the ai_review facade, and document the current runtime architecture.

pragent-bot bot reviewed 2026-09-01 00:27:05 +00:00

pragent-bot bot left a comment

🤖 AI Review · pragent pilot · headroom/MiniMax-M2.7 · b4041f68 · Merge confidence: 5/5 🟢

PR-Agent pilot on this repo. Comments are LLM-generated; treat as suggestions, not mandates.

Summary of Changes

Adds gitea_client.py (HTTP transport adapter) and model_client.py (Anthropic-compatible model adapter) extracted from ai_review.py
Adds review_config.py (opt-in policy with injected transport for testability) and review_ports.py (Protocol interfaces, currently unused)
Deletes dashboard.py, dashboard_data.py, k8s/pragent-dashboard.yaml, and four dashboard test files — obsolete after removing the dashboard subsystem
Delegates parse_text_blocks, _http, and call_model in ai_review.py to the new modules; updates webhook_server to use review_config.repo_enabled

Key Risks & Concerns

review_ports.py defines Forge/Reviewer/Telemetry protocols with no implementing callers — either premature scaffolding or forgotten interface contract
webhook_server.py retains base64 and urllib.parse imports from the old is_repo_enabled inline implementation — now dead code after the refactor
No dedicated test file for review_config.py despite it owning the security-sensitive opt-in gate; test coverage relies on test_webhook_server which patches at a distance

🔋 AI Usage & Run Details

Model / Engine: headroom/MiniMax-M2.7 · opencode · 24 steps · 307.5s
Total Tokens: 71,646 (71.6K) in / 10,108 (10.1K) out (0 reasoning, cache 1,498,191 (1.5M) read / 132,512 (132.5K) write, 1,712,457 (1.7M) total)
Actual: $0.00 (headroom/MiniMax-M2.7 — free tier)
Scope: Whole-repo checkout at head sha (agent can read any file + run linters, not just the diff) — input tokens include files read beyond the diff. Per-comment output is attributed (one model pass produces all findings; output split by each finding's body weight).

🤖 **AI Review** · pragent pilot · headroom/MiniMax-M2.7 · `b4041f68` · Merge confidence: 5/5 🟢 > PR-Agent pilot on this repo. Comments are LLM-generated; treat as suggestions, not mandates. ### Summary of Changes - Adds gitea_client.py (HTTP transport adapter) and model_client.py (Anthropic-compatible model adapter) extracted from ai_review.py - Adds review_config.py (opt-in policy with injected transport for testability) and review_ports.py (Protocol interfaces, currently unused) - Deletes dashboard.py, dashboard_data.py, k8s/pragent-dashboard.yaml, and four dashboard test files — obsolete after removing the dashboard subsystem - Delegates parse_text_blocks, _http, and call_model in ai_review.py to the new modules; updates webhook_server to use review_config.repo_enabled ### Key Risks & Concerns - review_ports.py defines Forge/Reviewer/Telemetry protocols with no implementing callers — either premature scaffolding or forgotten interface contract - webhook_server.py retains base64 and urllib.parse imports from the old is_repo_enabled inline implementation — now dead code after the refactor - No dedicated test file for review_config.py despite it owning the security-sensitive opt-in gate; test coverage relies on test_webhook_server which patches at a distance <details> <summary>🔋 AI Usage & Run Details</summary> - **Model / Engine**: `headroom/MiniMax-M2.7` · opencode · 24 steps · 307.5s - **Total Tokens**: 71,646 (71.6K) in / 10,108 (10.1K) out (0 reasoning, cache 1,498,191 (1.5M) read / 132,512 (132.5K) write, 1,712,457 (1.7M) total) - **Actual**: $0.00 (headroom/MiniMax-M2.7 — free tier) - **Scope**: Whole-repo checkout at head sha (agent can read any file + run linters, not just the diff) — input tokens include files read beyond the diff. Per-comment output is *attributed* (one model pass produces all findings; output split by each finding's body weight). </details>

masi merged commit 3a110ab52c into main

2026-09-01 00:28:07 +00:00

Sign in to join this conversation.

2 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: gitea_admin/pragent#15