Files
manual_slop/docs/PIPELINE_ANALYSIS_PHASE5_INIT.md
T

3.5 KiB

Phase 5 Architectural Audit: Intensive Pipeline Mapping

This document provides a tool-assisted, intensive technical trace of the major processing routes within Manual Slop. It identifies exact function call chains, data transformations, and subsystem boundaries.


1. AI Interaction Pipeline

Primary Route: Traces the flow from a user send request to final execution.

Call Graph (Depth 3)

-> ai_client.send (src\ai_client.py)
  -> _append_comms (src\ai_client.py)
  -> _get_combined_system_prompt (src\ai_client.py)
    -> generate_tooling_strategy (src\tool_bias.py)
  -> _send_anthropic (src\ai_client.py)
    -> _build_chunked_context_blocks (src\ai_client.py)
    -> _execute_tool_calls_concurrently (src\ai_client.py)
    -> run (src\aggregate.py)
  -> _send_gemini (src\ai_client.py)
    -> _gemini_tool_declaration (src\ai_client.py)
      -> get_tool_schemas (src\mcp_client.py)

2. Context Aggregation Pipeline

Primary Route: Traces the transformation of project files into AI-ready context.

Call Graph (Depth 3)

-> aggregate.run (src\aggregate.py)
  -> build_file_items (src\aggregate.py)
    -> get_monitor (src\performance_monitor.py)
    -> resolve_paths (src\aggregate.py)
  -> build_markdown_from_items (src\aggregate.py)
    -> _build_files_section_from_items (src\aggregate.py)
    -> build_beads_section (src\aggregate.py)
    -> build_discussion_section (src\aggregate.py)
    -> build_summary_markdown (src\summarize.py)
  -> find_next_increment (src\aggregate.py)

3. GUI Event & State Synchronization

Primary Route: Traces the background event loop and state management.

Call Graph (Depth 3)

-> app_controller._process_event_queue (src\app_controller.py)
  -> refresh_external_mcps (src\app_controller.py)
    -> add_server (src\mcp_client.py)
    -> stop_all (src\mcp_client.py)
  -> run (src\aggregate.py)
    -> build_file_items (src\aggregate.py)
    -> build_markdown_from_items (src\aggregate.py)

4. Simulation Lifecycle Pipeline

Primary Route: Traces the automated verification and scaffolding flow.

Call Graph (Depth 3)

-> simulation.sim_base.run_sim (simulation\sim_base.py)
  -> run (src\aggregate.py)
    -> build_file_items (src\aggregate.py)
    -> build_markdown_from_items (src\aggregate.py)
  -> teardown (simulation\sim_base.py)

5. Performance & Curation Insights

5.1 Redundancy Hotspots

  • Aggregation Boilerplate: Both AppController and run_sim call aggregate.run directly, but ai_client also triggers it during context refreshes. This indicates a potential for a shared, reactive context manager.
  • Provider Divergence: The call graph for _send_anthropic is significantly more complex than _send_gemini, suggesting inconsistent abstraction for context chunking and history management.

5.2 Threading Boundaries

  • Context Switches: The audit confirms that _process_event_queue acts as the primary synchronization gate, transitioning foreground UI requests into background worker tasks.
  • Lock Contention: Heavy reliance on threading.Lock within mcp_client (implied by configure calls) and app_controller suggests areas where lock-free data structures could improve frame latency.

5.3 Automated Tooling Success

  • The derive_code_path tool successfully navigated cross-subsystem boundaries (e.g., from simulation back into src.aggregate).
  • Future curation should prioritize tools over manual reviews to maintain the Acton/Muratori standards of technical discipline.