data2dsl

Ticket 065: Robust deserialization of nested dict structures in skill normalize_raw (F07)

• ID: ticket-065
• Owner: unresolved:human
• Status: DONE
• Workflow state: DONE
• Created: 2026-08-28

Goal and scope

Fix audit finding F07: When raw data is passed into data2dsl_skill._normalize_raw from JSON/MCP payloads, nested collections (such as tickets in Planfile, pages in Curllm/Diagit, services in Deta) arrive as plain Python dicts rather than typed dataclasses. Add robust deserialization in _normalize_raw so dicts are converted into their respective dataclasses (PlanfileTicketEvidence, CurllmPageEvidence, DetaServiceEvidence, DiagitPageEvidence) before calling adapter normalize.

SESSION_EXECUTION_AUTHORIZATION recorded from user prompt.

Acceptance criteria

• AC-01: Scope is approved (SESSION_EXECUTION_AUTHORIZATION recorded).
• AC-02: _normalize_raw converts raw nested dicts to typed evidence dataclasses for Planfile, Curllm, Deta, and GitHub.
• AC-03: MCP tool execution and direct skill calls with nested JSON payloads succeed without AttributeError.
• AC-04: Unit tests in tests/test_skill_raw_deserialization.py verify deserialization of dict payloads.
• AC-05: Full pytest suite passes (115/115) and governance-check.bat reports GOV-PASS.

Participants

• Human participant: unresolved; no user-* file was created by this script.
• Agent participant: ai-antigravity.md