Retrospective Analysis Skill

Overview

A project retrospective automation tool that helps teams identify process friction points, analyze failure causes, and generate actionable improvement candidates.

Key Features

• Structured Retrospectives: Track what went well, what was slow, what failed, and gate friction.
• Automatic Analysis: Compute issue counts, friction points, and severity (low/medium/high/critical).
• Improvement Generation: Derive prioritized improvement candidates from frictions and failures.
• Report Generation: Produce recommendations and action items from a retrospective.
• JSON Persistence: Save/load retrospectives and export reports to disk.
• Thread Safety: Concurrent adds and retrospective creation are safe.

Use Cases

• Post-change retrospective record keeping
• Process friction identification across gate checkpoints
• Continuous improvement candidate backlog generation
• Team retrospective reporting and archival

Installation

clawhub install retrospective-analysis

Usage

Basic Usage

from src import RetrospectiveAnalyzer
from src.models import ProjectInfo, GateFriction

analyzer = RetrospectiveAnalyzer()

project = ProjectInfo(name="Auth Refactor", team="Platform", duration="3 weeks", change_id="CH-42")
retro = analyzer.start_retrospective(project)

analyzer.add_what_went_well(retro.id, "fast design review")
analyzer.add_what_was_slow(retro.id, "manual test gate")
analyzer.add_what_failed(retro.id, "deploy rollback")
analyzer.add_gate_friction(
    retro.id,
    GateFriction(gate="testing-gate", issue="flaky tests", impact="blocked merge", suggested_change="stabilize suite"),
)

analysis = analyzer.analyze(retro.id)
print(analysis.severity, analysis.total_issues)

report = analyzer.generate_report(retro.id)
for item in report.action_items:
    print(item)

Persistence

analyzer.save_to_file(retro.id, "retro.json")
loaded = analyzer.load_from_file("retro.json")
analyzer.export_report(retro.id, "report.json")

API Reference

RetrospectiveAnalyzer

• start_retrospective(project_info) -> Retrospective — start a new retrospective
• add_what_went_well(retro_id, item) -> None — record something that went well
• add_what_was_slow(retro_id, item) -> None — record something slow/redundant
• add_what_failed(retro_id, item) -> None — record a failure/rework cause
• add_gate_friction(retro_id, friction) -> None — record gate friction
• analyze(retro_id) -> AnalysisResult — analyze the retrospective
• generate_report(retro_id) -> RetrospectiveReport — generate a report
• get_improvement_candidates(retro_id) -> List[ImprovementCandidate] — get candidates
• archive(retro_id) -> None — archive the retrospective
• save_to_file(retro_id, path) / load_from_file(path) / export_report(retro_id, path) — persistence

Models

• RetroStatus: ACTIVE / ANALYZED / REPORTED / ARCHIVED
• ProjectInfo: name, team, duration, change_id
• GateFriction: gate, issue, impact, suggested_change
• ImprovementCandidate: target, recommendation, reason, priority
• AnalysisResult: total_issues, friction_points, improvement_candidates, summary, severity
• Retrospective: id, project_info, status, what_went_well, what_was_slow, what_failed, gate_frictions, created_at
• RetrospectiveReport: retro_id, project_info, analysis, recommendations, action_items

Severity Logic

Quality Metrics

• 81 comprehensive tests (unit + extended unit + integration + e2e)
• JSON serialization round-trips for all models
• Concurrent operation safety
• MIT License

Changelog

v1.0.0 (2026-07-03)

• Initial release
• Retrospective lifecycle management
• Friction identification and failure analysis
• Improvement candidate generation with priority
• Report generation and JSON persistence