112 lines
6.9 KiB
Markdown
112 lines
6.9 KiB
Markdown
# TASKS.md
|
|
<!-- Quick-read pointer to active and planned conductor tracks -->
|
|
<!-- Source of truth for task state is conductor/tracks/*/plan.md -->
|
|
|
|
## Active Tracks
|
|
- `feature_bleed_cleanup_20260302` — Dead code & conflicting design state cleanup (Phase 1-3)
|
|
|
|
## Completed This Session
|
|
- `context_token_viz_20260301` — Token budget panel (color bar, breakdown table, trim warning, cache status, auto-refresh). All phases verified. Commit: d577457.
|
|
|
|
## Planned: Next Track
|
|
|
|
### `mma_agent_focus_ux_20260302` (initialized — run after bleed cleanup)
|
|
**Priority:** High
|
|
**Depends on:** `feature_bleed_cleanup_20260302` Phase 1 (dead comms panel removed)
|
|
**Track dir:** `conductor/tracks/mma_agent_focus_ux_20260302/`
|
|
|
|
**Audit-confirmed gaps:**
|
|
- `ai_client._append_comms` emits entries with no `source_tier` key
|
|
- `ai_client` has no `current_tier` module variable — no way for tiers to self-identify
|
|
- `_tool_log` is `list[tuple[str,str,float]]` — no tier field, tuple must migrate to dict
|
|
- `run_worker_lifecycle` replaces `comms_log_callback` but never stamps `source_tier`
|
|
- `generate_tickets` (Tier 2) does NOT replace callback at all
|
|
- No Focus Agent selector widget in Operations Hub
|
|
|
|
**Scope:** Phase 1 (tier tagging) → Phase 2 (tool log dict migration) → Phase 3 (Focus Agent UI + filter). Per-tier token stats deferred to sub-track.
|
|
|
|
### `tech_debt_and_test_cleanup_20260302` (initialized)
|
|
**Priority:** High
|
|
**Depends on:** `feature_bleed_cleanup_20260302`
|
|
**Track dir:** `conductor/tracks/tech_debt_and_test_cleanup_20260302/`
|
|
|
|
**Audit-confirmed gaps:**
|
|
- 13 test files duplicate `app_instance` fixture instead of using `conftest.py`.
|
|
- Duplicate test files (`test_ast_parser_curated.py`).
|
|
- Multiple simulation tests silently pass with no assertions.
|
|
- `gui_2.py` initializes 9 state variables in `__init__` that are never read.
|
|
- `gui_2.py` has over 15 uncalled HTTP/background methods.
|
|
|
|
**Scope:** Phase 1 (Fixture deduplication) → Phase 2 (False-positive test fixing) → Phase 3 (Dead code excision in `gui_2.py`).
|
|
|
|
### `conductor_workflow_improvements_20260302` (initialized)
|
|
**Priority:** High
|
|
**Depends on:** None
|
|
**Track dir:** `conductor/tracks/conductor_workflow_improvements_20260302/`
|
|
|
|
**Audit-confirmed gaps:**
|
|
- Tier 2 skill lacks enforcement of AST pre-implementation scans to prevent duplicate state variables.
|
|
- Tier 2 skill lacks explicit rejection of non-TDD execution.
|
|
- Tier 3 skill does not strictly forbid implementing code without failing tests.
|
|
- `workflow.md` lacks explicit warnings against zero-assertion tests and redundant `__init__` state.
|
|
|
|
**Scope:** Phase 1 (Update MMA Skill prompts) → Phase 2 (Update `workflow.md`).
|
|
|
|
### `architecture_boundary_hardening_20260302` (initialized)
|
|
**Priority:** High
|
|
**Depends on:** None
|
|
**Track dir:** `conductor/tracks/architecture_boundary_hardening_20260302/`
|
|
|
|
**Audit-confirmed gaps:**
|
|
- `ai_client.py` loops execute `set_file_slice` and `py_update_definition` instantly without checking `pre_tool_callback`, bypassing GUI approval.
|
|
- New `mcp_client.py` tools are not exposed in the GUI or `manual_slop.toml` config for user control.
|
|
- `mma_exec.py` bypasses skeletonization for `mcp_client`, causing token bloat.
|
|
- `dag_engine.py` does not cascade `blocked` states, causing orchestrator infinite loops.
|
|
|
|
**Scope:** Phase 1 (Meta-tooling token fix) → Phase 2 (Complete MCP Tool Integration & Seal GUI HITL bypass) → Phase 3 (Fix DAG Engine cascading blocks).
|
|
|
|
### `testing_consolidation_20260302` (initialized)
|
|
**Priority:** Medium
|
|
**Depends on:** `tech_debt_and_test_cleanup_20260302`
|
|
**Track dir:** `conductor/tracks/testing_consolidation_20260302/`
|
|
|
|
**Audit-confirmed gaps:**
|
|
- `visual_mma_verification.py` manually runs `subprocess.Popen` instead of using the robust `live_gui` fixture.
|
|
- Duplicate architectural logic between tests and `simulation/` directories causing fragmentation.
|
|
|
|
**Scope:** Phase 1 (Migrate manual launchers to fixtures) → Phase 2 (Consolidate simulation scripts).
|
|
|
|
---
|
|
|
|
## Track Dependency Order (Execution Guide)
|
|
To ensure smooth execution, execute the tracks in the following order:
|
|
1. `feature_bleed_cleanup_20260302` (Base cleanup of GUI structure)
|
|
2. `mma_agent_focus_ux_20260302` (Depends on feature bleed cleanup Phase 1)
|
|
3. `architecture_boundary_hardening_20260302` (Fixes critical HITL & Token leaks; independent but foundational)
|
|
4. `tech_debt_and_test_cleanup_20260302` (Re-establishes testing foundation; run after feature tracks)
|
|
5. `testing_consolidation_20260302` (Refactors testing methodology; depends on tech debt cleanup)
|
|
6. `conductor_workflow_improvements_20260302` (Meta-level updates to skills/workflow docs; can be run anytime)
|
|
|
|
---
|
|
|
|
## Future Backlog (Post-Cleanup)
|
|
*To be evaluated in a future Tier 1 session after the immediate tech debt queue is cleared.*
|
|
|
|
### `gui_decoupling_controller`
|
|
**Context:** `gui_2.py` is over 3,500 lines and operates as a Monolithic God Object. It violates the "Data-Oriented & Immediate Mode" heuristics by owning complex business logic, orchestrator hooks (`_bg_create_track`), and markdown file building instead of acting as a pure view.
|
|
**Goal:** Create a headless `orchestrator_pm.py` or `app_controller.py` that handles the core lifecycle, allowing `gui_2.py` to be a lagless, immediate-mode projection of the state.
|
|
|
|
### `robust_json_parsing_tech_lead`
|
|
**Context:** In `conductor_tech_lead.py`, the `generate_tickets` function relies on a generic `try...except` block to parse the LLM's JSON ticket array. If the model hallucinates or outputs invalid JSON, it silently returns an empty array `[]`, causing the GUI to fail the track creation process without giving the model a chance to self-correct.
|
|
**Goal:** Implement a programmatic retry loop that catches `JSONDecodeError` and feeds the error back to the Tier 2 model for self-correction before failing the UI operation.
|
|
|
|
### `strict_static_analysis_and_typing`
|
|
**Context:** Running `uv run ruff check .` and `uv run mypy --explicit-package-bases .` revealed massive technical debt in type safety (512+ Mypy errors across 64 files, 200+ remaining Ruff violations). The `gui_2.py` and `api_hook_client.py` files specifically have severe "Any" bleeding and incorrect unions.
|
|
**Goal:** Resolve all static analysis errors. Enforce strict `mypy` compliance, remove implicit `Optional` types, and fix ambiguous variables (`l`). Integrate `ruff` and `mypy` into a CI pre-commit hook so Tier 3 workers are forced to write type-safe code going forward.
|
|
|
|
### `test_suite_performance_and_flakiness`
|
|
**Context:** Running `uv run pytest` takes over 5.0 minutes to execute and frequently hangs on integration tests (e.g. `test_spawn_interception.py`). Several simulation tests (`test_sim_ai_settings.py`, `test_extended_sims.py`) are also currently failing or timing out.
|
|
**Goal:** Audit the test suite for `time.sleep()` abuse. Replace hardcoded sleeps with `threading.Event()` hooks or robust polling. Isolate slow integration tests with `@pytest.mark.slow` and ensure the core unit test suite runs in under 10 seconds to maintain high-velocity TDD.
|
|
|
|
|