Technical Health Assessment

This assessment is based on repository metadata, static code/manifests audit, README evidence, and open issue titles. It should be refreshed from GitHub before each release or go-live review.

Current status by domain

Domain	Status	Evidence / rationale
Core contracts	Healthy	alphaswarm_core has no open issues and remains dependency-light.
Hosted platform	At-risk before go-live	alphaswarm_platform#18 is P0 for deployment prerequisites/CD runbook; many Helm/Terraform/K8s surfaces increase release complexity.
Monolith/data isolation	Critical risk	alphaswarm carries RLS strict-mode, missing policy, context writer, migration, and test debt.
Identity	Needs integrated rollout proof	alphaswarm_auth appears strong architecturally, but tenant-router/RLS/Entra flows require cross-repo validation.
Hosted UI	At-risk before go-live	alphaswarm_ui#8 P0: replace mocks with backend-backed data and E2E coverage.
IDE/MCP	At-risk before go-live	alphaswarm_ide#4 and alphaswarm_mcp#3 are P0 production hardening/go-live issues.
Agent runtime	Moderate risk	alphaswarm_agents has robustness and PII-redaction debt.
AI/ML governance	Scaffolded/healthy but migration-dependent	alphaswarm_mlops has no open issues, but README states source migrations proceed by gated rollout.
Learning/Scholar	High risk	alphaswarm_learning has P0 production data-path TODO and test collection failures.
Observability	Roadmap incomplete	alphaswarm_observe, KB federation, and research plane issues require more spans/read APIs/replay features.
Docs	Needs reconciliation	alphaswarm_docs#19 calls out stale readiness docs vs current source-of-truth runbooks.

Open issue themes

P0 / critical

• alphaswarm#32 — validate RLS strict-mode rollout and null-workspace backfill before go-live.
• alphaswarm#31 — route broker venue order submissions through the OrderIntent airlock.
• alphaswarm#137 — context-less writers fail closed under tenancy RLS enforcement.
• alphaswarm#138 — 16 tenant-scoped tables with workspace_id lack RLS policy backstop.
• alphaswarm#139 — apply agent-ledger RLS DDL to every target DB before enforcement.
• alphaswarm_platform#18 — complete minimum go-live deployment prerequisites and CD runbook.
• alphaswarm_ui#8 — replace go-live mock surfaces with backend-backed data and E2E coverage.
• alphaswarm_ide#4 — complete IDE go-live checklist for image, identity, DNS/TLS, MCP, monitoring.
• alphaswarm_mcp#3 — harden MCP deployment for production readiness.
• alphaswarm_learning#1 — replace Scholar graph/LLM TODO stubs with production data paths.
• alphaswarm_client#4 — finish and validate hosted demo-mode deployment for classic client.

P1 / should-fix

• alphaswarm#143 — Alembic cannot upgrade head from empty.
• alphaswarm#145 — bootstrap does not seed default tenancy rows.
• alphaswarm#142 — Alpha Vantage loader tests fail due to attribute drift.
• alphaswarm#141 — destructive Postgres RLS test with hardcoded port.
• alphaswarm#144 — frozen screening dashboard schema not vendored.
• alphaswarm_config#29 — credentials subpackage imports monolith.
• alphaswarm_docs#19 — stale readiness docs need source-of-truth reconciliation.
• alphaswarm_observe#2 — evals/lineage/read API, registry/annotation/alert/replay roadmap.
• alphaswarm_api#1 — remove generated and IDE artifacts.

P2 / cleanup and quality

• Dead code in strategies/momentum.py.
• Health collector under-counting under RLS.
• Config validation benign fail-open in screening datasets.
• Agent registry default directory import-time resolution.
• Empty tool_call_id runtime robustness.
• RAG ingest parser provenance/hash drift.
• Additional retrieval/federation/research spans.

Prioritized follow-up plan

Priority	Workstream	Owners / repos	Acceptance criteria
P0	Tenant isolation go-live gate	alphaswarm, alphaswarm_platform, alphaswarm_auth	All RLS DDL applied, missing policies resolved, strict-mode validation passed, default seed rows proven, rollback documented.
P0	Hosted deployment readiness	alphaswarm_platform, alphaswarm_ops_console, alphaswarm_docs	CD runbook complete, Helm/Terraform smoke tests documented, rollback path tested, ops console confirm gates verified.
P0	Customer-facing UI readiness	alphaswarm_ui, alphaswarm_controller, alphaswarm_api	Mock surfaces replaced, BFF backed by services, E2E coverage against staging-like environment.
P0	MCP/IDE production path	alphaswarm_mcp, alphaswarm_ide, alphaswarm_auth	Identity, DNS/TLS, MCP connectivity, monitoring, and deployment image checklist complete.
P1	Docs reconciliation	alphaswarm_docs, all service owners	This audit lands in docs repo; stale readiness pages replaced or linked to canonical runbooks.
P1	AI/ML observability coverage	alphaswarm_observe, alphaswarm_kb_federation, alphaswarm_research, alphaswarm_agents	AGENT/EVAL/RETRIEVAL/federation spans emitted and visible in read API dashboards.
P1	Package-boundary cleanup	alphaswarm_config, alphaswarm_api, alphaswarm_agents, alphaswarm_learning	Monolith imports removed/guarded, generated artifacts removed, tests pass.
P2	Health debt burn-down	Individual repo owners	Cosmetic/quality issues closed; no new P0/P1 regressions.

Business-team follow-ups

• Decide release/go-live criteria for hosted platform, UI, IDE, and MCP based on the P0 list above.
• Assign explicit owners for tenant isolation/RLS, deployment readiness, UI go-live, MCP/IDE, and docs reconciliation.
• Confirm customer-facing identity story: Entra-only hosted app, staff admin flow, worker/device auth, and support process.
• Define customer documentation boundaries: public marketing/docs vs internal operations/security runbooks.
• Maintain a roadmap view linking MLOps governance maturity to product claims around model risk, evaluations, and agent reliability.

Development-team follow-ups

• Add or update tests for RLS migrations and tenant-router auth modes before strict enforcement.
• Add service catalog automation to alphaswarm_docs from GitHub metadata and repo manifests.
• Ensure each deployable repo documents image name, chart path, health endpoints, env vars, and rollback.
• Require issue labels for priority/domain to make future health reports machine-sortable.
• Add CI checks that prevent generated artifacts, IDE files, raw secrets, and stale doc links from entering service repos.