AI Effective Config and Review Execution

This page is the architecture index for Lintro’s AI review execution seams. The normative decision record is ADR-0006 (epic #1972).

Ownership boundaries

Concern	Owner today	Target
Typed AIConfig from raw ai: mapping	AIConfig.resolve_from_mapping() → ResolvedAIConfig	Same resolver; #1923 extends it
Invocation transport / timeout / cost-cap overrides	CLI review: apply_cli_overrides on ResolvedAIConfig	Same resolver pipeline (#1970 / #1923 / #2024)
Monotonic cost-cap clamp	MCP adapter (resolve_budget_policy)	Shared domain prep; adapters keep policy
Diff review preparation	Duplicated in CLI + MCP	prepare_review / execute_review (Phase 3)
Review execution facade	run_review / run_review_async	Unchanged facade; internals split (Phase 4)
Provider client aclose() API	Not yet (#1885)	Provider-side only in #1885
Provider close call-site wiring	N/A until #1885	Phase 5 of #1972

Shared preparation (current duplicated steps)

Both the CLI (lintro review) and MCP (lintro_review) currently:

1. Resolve AI config. CLI review uses AIConfig.resolve_from_mapping plus apply_cli_overrides (keeps provenance). MCP uses resolve_ai_config, which unwraps the same env-aware parse.
2. Gate on ai.review being enabled (adapter-specific error shape).
3. Collect review context, classify changed files, select/format checklist items.
4. Optionally build a lint digest.
5. Resolve sensitivity policy and construct a provider.
6. Invoke run_review and translate failures.

Adapter-only policy that must stay out of the shared layer:

• CLI: Click errors, progress UI, JSON/terminal rendering, GitHub posting, exit 0/1/2.
• MCP: workspace locking, budget clamp, structured tool envelopes, no posting.
• Advisory-only CLI mode: master-switch semantics without the diff-review sub-toggle.

Orchestrator phase plan

run_review remains the stable facade. Phase 4 decomposes internals into runner/session, planning/chunks, prompts/passes, response pipeline, merge/filter, and metadata modules without changing prompts, findings, severity, or exit semantics. Every provider call continues through call_ai; prompt redaction remains mandatory.

Exit and error contracts

• Exit 0 — successful review, no P1 findings.
• Exit 1 — successful review with P1 findings.
• Exit 2 — no review produced (REVIEW_ERROR_EXIT_CODE).

CLI JSON failures and MCP review failures both build diagnosis fields through build_error_contract.

Characterization tests

Phase 1 locks the gaps listed in ADR-0006:

• tests/unit/test_core_ai_import_boundary.py — AC10 / #724 import edge.
• tests/unit/ai/review/test_architecture_characterization.py — CLI/MCP preparation, effective-config parity, metadata keys, error mapping, exit 0/1/2.
• tests/unit/ai/review/test_architecture_characterization_1972.py — gap coverage: config-resolution idempotence, shared run_review kwargs, error-contract body parity, MCP error mapping.