Public-Sector Bid Writing and Compliance Self-Check Assistant

Problem

Public-sector bids often fail for concrete engineering reasons: long tender documents hide scoring rules, eligibility gates, and rejection clauses; response writing lacks a repeatable structure; and compliance review depends on manual experience, which can miss disqualifying issues. This skill decomposes bid support into parsing, extraction, outline generation, and validation instead of relying on one open-ended model pass.

Core Capabilities and Workflow

It works with PDF and DOCX tender files using local dependencies such as pdfplumber, python-docx, and openpyxl. The main flow is:
- Structured parsing: clean headers, footers, watermarks, and TOC noise; split content by heading levels; extract project basics, qualification thresholds, rejection clauses, scoring rules, and formatting attachments.
- Outline generation: produce commercial, technical, and quotation bid outlines with scoring-item mapping and missing-materials hints as Word output.
- Compliance checking: run structured comparison, keyword matching, semantic validation, and global cross-checks; classify fatal rejection risks, score risks, and optimization suggestions into an Excel report.
- Knowledge QA and fault tolerance: answer built-in bid FAQ entries, and use retry, restart, circuit breaker, and fallback behavior when subflows fail.

Boundaries

It is intended for public-sector open tenders, not scanned or image-only PDFs, private projects, or classified files. It does not write the technical solution, invent qualifications or track record, sign or seal documents, or replace final human review. Complex layouts may still cause missed fields, and every risk item should be manually verified before editing.