Assessment: L1 Eliminatory filters, Post-Processing & Results page

[vprashrex]

Target Issue: #187

Summary

Extends the assessment flow from a single LLM pass into a configurable pipeline:
Mapper → Eliminatory → Evaluation → Post Processing → Review. Adds pre-filtering to cut LLM cost on irrelevant rows, post-processing to shape results without manual spreadsheet work, and a dedicated full-page results view.

What users can do

• Eliminatory (L1) filters — optional pre-filters run before the LLM batch:

  • Topic Relevance — gates rows; decision=REJECT rows are excluded from Evaluation and flagged in the export. Ships with a default rubric, editable in an expandable editor.
  • Duplicate Detection — passthrough on rows that passed Topic Relevance; appears in the export, does not gate Evaluation.

• Post Processing — define rules applied at export time, available both as a pipeline step and as a panel on completed runs ("Apply & Preview"):

  • Computed columns via formulas with @column autocomplete + an operator toolbar (@Novelty_score + @Feasibility_score)
  • Filter (AND logic) across =, ≠, >, <, ≥, ≤, contains, is empty …
  • Sort by priority order (asc/desc)

• Results page — preview opens at /assessment/results/[runId] as a full-page spreadsheet view (replaces the modal), with editable-state persistence.

• At-a-glance signals

  • L1 stats on each run (passed / rejected counts)
  • Latest provider · model badge on saved config cards
  • New pipeline statuses (l1_processing, l2_processing, l1_failed)
  • Attachment columns support mixed type (image or PDF per row)

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 0587361e-ffd8-4a7a-bb0f-dc8af7f9d20f

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch feat/assessment-pipeline-l1

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[Ayush8923]

Approve with few comments.