This directory documents how APME works at runtime — the processing pipeline as a numbered sequence of stages, from CLI invocation through to final output.
architecture/ vs design/: Architecture docs describe the runtime pipeline — what happens, in what order, and how data flows between services. Design docs (in
../design/) explain why a subsystem was built the way it was — alternatives considered, trade-offs, and rationale.
The former monolithic ARCHITECTURE.md and DATA_FLOW.md have been
archived; their content is now covered by docs 00–17
below.
flowchart TD
subgraph cli [CLI Layer]
A[apme check / remediate / format]
end
subgraph primary [Primary Orchestrator :50051]
B[Session Management]
C[YAML Formatting]
D[Parse and Graph Construction]
E[Collection Resolution and Venv]
F[Validator Fan-out]
G[Tier 1 Deterministic Remediation]
H[AI-Assisted Remediation]
I[Post-AI Deterministic Pass]
J[Human Approval Flow]
K[Result Assembly]
end
subgraph validators [Validators]
F1[Native :50055]
F2[OPA :50054]
F3[Ansible :50053]
F4[Gitleaks :50056]
end
subgraph external [External Services]
GP[Galaxy Proxy :8765]
AB[Abbenay AI :50057]
end
subgraph persistence [Persistence Layer]
GW[Gateway :50060/:8080]
UI[UI :8081]
end
A -->|gRPC FixSession| B
B --> C
C --> D
D --> E
E -->|PEP 503| GP
D --> F
F --> F1 & F2 & F3 & F4
F --> G
G -->|rescan dirty| F1 & F2 & F3 & F4
G --> H
H -->|propose_node_fix| AB
H --> I
I --> J
J --> K
K -->|emit_fix_completed| GW
GW --> UI
K -->|SessionResult| A
| # | Document | Stage |
|---|---|---|
| 00 | 00-overview.md | Pipeline overview, check vs remediate vs format |
| 01 | 01-initialization-and-ingestion.md | CLI initialization and file ingestion |
| 02 | 02-session-management.md | Session creation and file staging |
| 03 | 03-formatting.md | YAML formatting and idempotency check |
| 04 | 04-parse-and-graph.md | Parse, graph construction, and hierarchy |
| 05 | 05-collection-resolution.md | Collection resolution and venv management |
| 06 | 06-validator-fanout.md | Validator fan-out and violation detection |
| 07 | 07-tier1-remediation.md | Tier 1 deterministic remediation |
| 08 | 08-ai-remediation.md | Tier 2 AI-assisted remediation |
| 09 | 09-post-ai-deterministic.md | Post-AI deterministic pass |
| 10 | 10-human-approval.md | Human approval flow |
| 11 | 11-result-assembly.md | Result assembly and reporting |
| 12 | 12-output-and-presentation.md | CLI output and presentation |
| 13 | 13-gateway-and-persistence.md | Gateway persistence and REST API |
| 14 | 14-ui-integration.md | UI and WebSocket integration |
| 15 | 15-concurrency-model.md | Concurrency model and executor discipline |
| 16 | 16-diagnostics.md | Diagnostics instrumentation and timing |
| 17 | 17-scaling-and-deployment.md | Scaling strategy and deployment topology |
- DESIGN_REMEDIATION.md — remediation engine design
- DESIGN_VALIDATORS.md — validator design
- DESIGN_AI_ESCALATION.md — AI escalation design
- ADR index: .sdlc/adrs/README.md
Start with 00-overview.md for the big picture, then read sequentially or jump to a specific stage. Each document references its neighbors with "Previous" and "Next" links.