L’interface est localisée ; les présentations et documents des Skills sont traduits automatiquement par IA dans votre langue. Le code, les commandes, chemins et termes techniques restent en version originale, et le document de l’auteur fait foi.
Explorer les Skills
Développement · SKILL.md

ediscovery-review-summary

ediscovery-review-summary

Produces an internal U.S. litigation e-discovery status summary covering ESI collection, processing, search-term hits, review throughput, relevance and privilege rates, and completion forecasts. Triggers when the user requests an e-discovery status report, collection log summary, processing report, review statistics, coding report, search-term results, or p…

Independent GitHub snapshot · Automated classification · Original documentation governs

Documentation de la Skill

GitHub

name: ediscovery-review-summary language: en description: >- Produces an internal U.S. litigation e-discovery status summary covering ESI collection, processing, search-term hits, review throughput, relevance and privilege rates, and completion forecasts. Triggers when the user requests an e-discovery status report, collection log summary, processing report, review statistics, coding report, search-term results, or privilege review summary. tags:

  • litigation
  • summarization
  • summary

E-Discovery Collection & Review Summary

Internal status report tracking ESI from collection through attorney review, with completion projections.

Prerequisites

  • Custodian list with collection scope (systems, date ranges, locations)
  • Collection logs and processing reports from vendor
  • Review platform exports (coding, reviewer activity, search-term results)
  • Search-term list or Boolean queries
  • Review protocol (relevance, privilege, issue tags) and QC standards
  • Key deadlines the review must meet
  • Protective order or confidentiality requirements for reporting

Quick Start

  1. Gather source data (collection logs, processing reports, review exports, search-term results, privilege summaries).
  2. Normalize custodian names into a master mapping table.
  3. Run integrity checks — pipeline totals must satisfy: Collected ≥ Processed ≥ Loaded ≥ Reviewed.
  4. Compute core metrics tables (see below).
  5. Draft executive summary with risks and recommendations.
  6. Apply work-product legend.

Core Metrics

A. Collection & Processing by Custodian

Per custodian: Collected GB, Item Count, Processed %, Load Failures, File-Type Mix, Exceptions.

B. Search-Term Effectiveness

Per term/query: Hits, Hit Rate %, Sample Precision %, Overlap %, Recommended Action.

C. Review Status (Matter-Level)

Total Reviewable, Reviewed Count, % Reviewed, Relevant %, Privileged %, Non-Responsive %, Est. Completion Date.

D. Reviewer Productivity

Per reviewer/team: Docs Reviewed, Hours, Docs/Hour, Relevant %, Privileged %, QC Rate %.

E. Privilege Patterns

Per custodian: Privileged %, Dominant Doc Types, Red Flags.

Key Formulas

MetricFormula
Hit Rate %hits ÷ total reviewable corpus
Precision %relevant hits ÷ sampled hits
Review Velocitydocs reviewed ÷ time period
Completion ETAremaining docs ÷ current velocity
Privilege Rate %privileged ÷ reviewed

Executive Summary Template

  • Collection completeness — status and anomalies
  • Processing health — success rate and key errors
  • Search-term performance — top/weak terms and overlap
  • Review progress — reviewed %, relevance %, privilege %
  • Forecast — ETA with assumptions
  • Risks & actions — top three items

Integrity Checks

  • Custodian names align across all sources (master mapping table).
  • Pipeline totals are monotonically decreasing through each stage.
  • Every search term in the term list appears in the results export.
  • Coding records contain reviewer, date, relevance, and privilege fields.
  • Processing errors are reconciled against re-runs.

Findings & Recommendations

  • List missing sources and resulting limitations.
  • Flag custodians with outlier volumes or error rates.
  • Identify low-precision or high-overlap search terms.
  • Note reviewer inconsistencies and QC actions taken.
  • Recommend resource changes or search-term refinement.
  • Document assumptions and gaps with impact and mitigation.

Pitfalls

  • Report only aggregate metrics — never disclose privileged content.
  • If counts conflict, surface the discrepancy and state the suspected cause.
  • Use U.S. discovery terminology; confirm local rules before external distribution.
  • Mark uncertain legal citations with [VERIFY].
  • Keep narrative neutral and data-driven; avoid strategy conclusions.
  • Always include: "Attorney Work Product / Privileged & Confidential. Internal use only."

Skills enregistrées