building-preaction-report

Assembles pre-action analysis reports from request, candidates, and evidence JSON files.

6|Updated May 11, 2026
One-click install
npx skills add https://github.com/divalto/divalto-ia-devkit --skill building-preaction-report-divalto
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: building-preaction-report
Source: https://github.com/divalto/divalto-ia-devkit/tree/main/plugins/divalto-devkit/skills/building-preaction-report
Command: npx skills add https://github.com/divalto/divalto-ia-devkit --skill building-preaction-report-divalto

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires jinja2, and includes scripts (resource) and references (resource) components.

What problem does it solve? Developers working on Divalto Harmony ERP code need a consolidated, traceable analysis report before modifying code, but manually merging request parsing, candidate search, and evidence verification results into a coherent markdown document is slow and error-prone. ## Core Features & Use Cases - Catalog-driven report assembly: Evaluates 9 content types by deterministic relevance scoring and includes only pertinent sections (examples, impact study, action sites, business constants, dossier settings). - Dual deliverables: Produces a human-readable markdown report plus a JSON metrics file with confidence level, Neo4j coverage, and X.12/X.13 traceability counts. - Quality enforcement: Applies hard bounds, absolute-path validation, anti-stub rules, and golden tests for Jinja2 fragments before any report ships. - Use Case: After running the three analysis phases on a bug ticket, feed request.json, candidates_x12.json, and evidence_x13.json to the script to get a final markdown report with confirmed code examples, impact callers, and action proposals for the developer. ## Quick Start Run the build_report.py script with the paths to your request, candidates, and evidence JSON files to generate the final pre-action markdown report and metrics JSON in the output directory.

Frequently Asked Questions about building-preaction-report

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate a pre-action analysis report from JSON files?

Run build_report.py with --request, --candidates, and --evidence arguments pointing to the three phase JSON files, plus an optional --output-dir. The script writes a markdown report and a metrics JSON named preaction-<slug>-<date> into the output directory.

What inputs does the pre-action report builder require?

It requires three JSON files: request.json from phase 1 parsing, candidates_x12.json from phase 2 candidate search, and evidence_x13.json from phase 3 verification. A preflight check rejects degraded inputs before any report is generated.

How are report sections selected for inclusion?

A catalog of 9 content types is scored deterministically from the input data, such as confirmed X.13 count, callers, detected literals, and request type. A section is included only when its score meets its threshold, and the selection is traced in metrics.json.

Does the report builder work without Neo4j?

Yes, it supports a degraded direct mode. When Neo4j is unavailable, the metrics report couverture_neo4j as absente and the report relies on grep-based evidence from the evidence JSON instead of graph relations.

Why does report generation fail validation?

The validator rejects reports containing relative file paths, stub placeholders like TODO, sentence fragments starting with relative pronouns, or unsupported Mermaid and Markdown constructs. Run the golden tests in scripts/tests before modifying any Jinja2 fragment.