fix: state low-score issue contract in review prompts

[jimmybrancaccio]

Problem

Focused review batches are rejected when they score a dimension below 85 without an issue, but the generated task requirements do not state that import contract. In a real project run, the abstraction-fitness reviewer twice returned a coherent 84 assessment with issues: []; both runs exited successfully and were then marked parse_failed.

Fix

State the existing low-score issue requirement directly in the shared task requirements used by batch and external review prompts. The wording also tells reviewers not to invent defects: when no issue is warranted, the score must remain consistent with that judgment.

Validation

• python -m pytest desloppify/tests/commands/review/test_review_batch_core_direct.py -q — 22 passed
• python -m pytest desloppify/tests/ -q — 5,811 passed, 3 skipped; one unrelated pre-existing Nim tree-sitter grammar mismatch
• Re-ran the previously failing immutable project packet, batch 3, through the patched source; it imported successfully and produced a concrete abstraction-fitness issue with the below-85 score.

[Copilot]

Copilot review overview

🟡 Changes recommended

The new import block violates the configured Ruff import-sorting rule and should be merged with the existing block.

Get a fresh assessment by requesting another Copilot review.

Review effort: Lite
Findings: 1

Open (1)

• Merge duplicate feedback_contract imports · New

What changed in this PR

Clarifies the low-score issue contract in shared batch/external review prompts.

Changes:

• Adds the below-85 score requirement and anti-invention guidance.
• Adds regression assertions for the batch prompt.

File	Description
prompt_sections.py	Adds shared low-score task requirements.
test_review_batch_core_direct.py	Verifies the new prompt contract.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.