You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: .planning/PROJECT.md
+2-2Lines changed: 2 additions & 2 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -24,7 +24,7 @@ You can give one prompt and watch a trustworthy multi-agent coding system drive
24
24
25
25
### Validated
26
26
27
-
- Existing: repo-aware local agent runtime via Microsoft Agent Framework and DevUI in `maf_starter/` and `entities/`
27
+
- Existing: repo-aware local agent runtime via Microsoft Agent Framework and DevUI in `maf_core/` and `entities/`
28
28
- Existing: multi-agent workflow scaffolding, including the `repo_team` planner/researcher/implementer/reviewer workflow
29
29
- Existing: provider and model routing with fallback across Gemini API, optional Anthropic, and CLI fallbacks
30
30
- Existing: human-in-the-loop and route trace mechanisms in local form through tool approval, route metadata, and DevUI patching
@@ -53,7 +53,7 @@ You can give one prompt and watch a trustworthy multi-agent coding system drive
53
53
54
54
## Context
55
55
56
-
This is a brownfield codebase with an existing local agent runtime. The active path is MAF-first: `main.py` dispatches into `maf_starter/cli.py`, entities are discovered from `entities/`, and DevUI is used for local interaction. The repo also retains a substantial legacy AutoGen stack in `autogen_dashboard/` and `autogen_starter/`, which creates both opportunity and drift risk.
56
+
This is a brownfield codebase with an existing local agent runtime. The active path is MAF-first: `main.py` dispatches into `maf_core/cli.py`, entities are discovered from `entities/`, and DevUI is used for local interaction. The repo also retains a substantial legacy AutoGen stack in `autogen_dashboard/` and `autogen_starter/`, which creates both opportunity and drift risk.
57
57
58
58
The workstation context matters. GSD is already installed locally, and you also have Gemini CLI, Claude CLI, and Codex CLI available, along with Gemini API access and local Azure Functions development tools. On this machine, Azure Functions Core Tools `4.8.0`, Azure CLI `2.80.0`, and Docker `29.1.5` are available. `azd` is not currently installed, so this milestone should not depend on `azd`-only templates or flows. The intended operating model is to use the local machine as the main execution environment, let Gemini API answer most orchestration and GSD questions automatically, and use the installed CLIs as specialist workers or fallback engines when appropriate.
Copy file name to clipboardExpand all lines: .planning/ROADMAP.md
+11-29Lines changed: 11 additions & 29 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -16,8 +16,7 @@ This roadmap now enters milestone v1.1: Cloud API and Azure Function Hosting. Th
16
16
-[x]**Phase 4: Autonomous Repo Execution and Validation Guardrails** - Turn the system into a safe default-doer for repo work
17
17
-[x]**Phase 5: Polished Operator Workbench** - Replace prototype interaction with a durable operator-grade UI
18
18
-[ ]**Phase 6: API Boundary and Control Plane Contract** - Extract a shared orchestration service layer and expose it through a stable HTTP API
19
-
-[ ]**Phase 7: Azure Functions Host and Durable API Surface** - Host the control plane in Python Azure Functions with durable run state and async status behavior
20
-
-[ ]**Phase 8: Worker Boundary and Cloud-Safe Execution Profiles** - Separate cloud ingress from long-running repo execution and local-only provider assumptions
19
+
-[ ]**Phase 7: Worker Boundary and Cloud-Safe Execution Profiles** - Separate cloud ingress from long-running repo execution and local-only provider assumptions
21
20
22
21
## Phase Details
23
22
@@ -37,23 +36,7 @@ Plans:
37
36
-[ ] 06-02: Add REST endpoints for submit, status, control, routing, agents, and artifacts
38
37
-[ ] 06-03: Align Operator Workbench and HTTP API parity over the shared run contract
39
38
40
-
### Phase 7: Azure Functions Host and Durable API Surface
41
-
**Goal**: Host the control plane on Azure Functions with durable state and local Core Tools parity.
last_activity: 2026-03-22 - Phase 6 complete - Plan 03 delivered Command Center parity tests, legacy compatibility notices, and README External API section
9
9
progress:
10
-
total_phases: 8
11
-
completed_phases: 5
12
-
total_plans: 24
13
-
completed_plans: 15
10
+
total_phases: 2
11
+
completed_phases: 1
12
+
total_plans: 3
13
+
completed_plans: 4
14
+
percent: 100
14
15
---
15
16
16
17
# Project State
@@ -20,20 +21,20 @@ progress:
20
21
See: .planning/PROJECT.md (updated 2026-03-22)
21
22
22
23
**Core value:** You can give one prompt and watch a trustworthy multi-agent coding system drive real repo work end-to-end with clear traces, specialist visibility, and minimal manual intervention.
23
-
**Current focus:** Phase 6 context gathered - ready for planning
24
+
**Current focus:** Phase 6 in progress - REST API complete
24
25
25
26
## Current Position
26
27
27
-
Phase: 06 (api-boundary-and-control-plane-contract) - READY TO PLAN
28
-
Plan: -
29
-
Status: Phase context captured and ready for plan-phase
30
-
Last activity: 2026-03-22 - Phase 6 context gathered
28
+
Phase: 07 (azure-functions-and-cloud-control-plane) - READY TO START
29
+
Plan: 07-01 (next)
30
+
Status: Phase 6 complete - Shared control-plane API delivered with /api/v1 REST router, Command Center parity validation, and external API documentation. Ready for Azure Functions deployment.
31
+
Last activity: 2026-03-22 - Phase 6 complete - Plan 03 delivered Command Center parity tests, legacy compatibility notices, and README External API section
@@ -87,7 +92,7 @@ Recent decisions affecting current work:
87
92
- Phase 02 planning: Use one canonical manager stage sequence `planning -> research -> implementation -> review -> validation`
88
93
- Phase 02 planning: Make stage-scoped pause, resume, and retry part of the durable run contract instead of transcript-only behavior
89
94
- Phase 02 planning: Answer routine GSD clarification questions automatically from planning docs, phase context, workspace snapshot, and repo facts
90
-
- Phase 02-01: Canonical stage state lives in `maf_starter/orchestration.py` and is shared by workflow builders and dashboard runtime
95
+
- Phase 02-01: Canonical stage state lives in `maf_core/orchestration.py` and is shared by workflow builders and dashboard runtime
91
96
- Phase 02-02: Planning pauses after plan output, while blocked or failed later stages resume from the same stage without replaying completed work
92
97
- Phase 02-03: Operator-facing orchestration visibility is exposed through dashboard cards and API payloads, not only raw traces
93
98
- Phase 03 context: Keep the manager as the only canonical run owner while surfacing `planner`, `researcher`, `implementer`, and `reviewer` as first-class visible specialists
0 commit comments