# Case timeline builder
> Source: https://legalcode.md/skills/legalcode-case-timeline-builder | [HTML version](https://legalcode.md/skills/legalcode-case-timeline-builder)

Build defensible litigation chronologies from document corpora using a dual-path multi-agent pipeline.

Build defensible litigation chronologies from document corpora using a dual-path multi-agent pipeline. Process PDFs, Word documents, emails, images, and other evidence formats at scale — converting to markdown, extracting dates, events, entities, and relationships using Claude Haiku workers in parallel, validating sequences with Sonnet, and synthesizing final timelines with Opus. Supports two execution paths: (1) Claude Code CLI scripts (`claude -p --model haiku`) for maximum throughput on large corpora, and (2) in-skill sub-agent fallback via the Task tool when CLI is unavailable. Produces citation-linked master timelines, issue timelines, entity timelines, deadline calendars, and court-ready outputs (FRCP Rule 56 statements, proposed findings of fact) across jurisdictions. Handles 10K+ document corpora with automatic chunking, deduplication, privilege screening, protective order designations, and conflict resolution.

This skill builds comprehensive, defensible litigation chronologies from heterogeneous document corpora using a **dual-path multi-agent pipeline**. It processes raw documents at scale, converts them to markdown, extracts temporal and entity information using parallel Haiku workers, validates sequences, resolves contradictions, and produces multiple timeline views optimized for different litigation phases and use cases. The skill supports two execution architectures: 1. **CLI Path** (preferred for large corpora): Shell scripts that invoke `claude -p --model haiku` workers in parallel, orchestrated by `timeline-build.sh` and stage scripts in `scripts/corpus-processing/`. Maximum throughput, cost tracking, resumable. 2. **Sub-Agent Path** (fallback): When the `claude` CLI is not installed or not available, the skill falls back to launching Task tool sub-agents with `model: "haiku"` from within the skill execution context. Same extraction logic, same prompts, lower throughput. **Covers:** - **Document conversion**: PDF, DOCX, MSG, EML, PST, HTML, images, transcripts, spreadsheets to markdown - **Multi-agent extraction**: Parallel date/event/entity/relationship extraction using Haiku workers - **Coreference resolution**: Cross-document entity linking, alias tables, name normalization - **Timeline construction**: Master chronology, issue timelines, entity timelines, relationship maps - **Quality assurance**: Contradiction detection, sequence validation, gap analysis, confidence auditing - **Privilege screening**: Privilege log integration, redacted timeline entries, parallel privilege chronology - **Court-ready outputs**: FRCP ...

## Jurisdictions

Global



---
*Legalcode -- legalcode.md*