convez.ai
Skill
jkoehler-redhat

strategy-review

jkoehler-redhat/strategy-review

Ver no GitHub
InicialPluginScore · de 0 a 100

Claude Code skill for generating data-backed AICP strategy review documents for leadership meetings

Instalação

$npx convez-ai jkoehler-redhat/strategy-review

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
0
Forks
0
Acompanhando
0
Saúde
42%
Último commit
há 5 meses
Versão
—
Issues abertas
0
Contribuidores
1

README

A tradução para o português deste README ainda está na fila.

Strategy Review Skill

A Claude Code skill that generates data-backed strategy review documents for AI Core Platform (AICP) leadership meetings. It pulls live data from two sources, categorizes work by ownership and sub-team, and produces a formatted Word document ready for director-level review.

What It Does

Generates a 7-section strategy review document:

  1. What We Own — Roadmap features split by ownership (AICP-owned vs cross-team support), plus engineering portfolio breakdown by sub-team
  2. What We've Delivered — Resolved issues grouped by theme (xKS, Gateway, Observability, Security, etc.) over a configurable lookback period
  3. What's In Flight — Active issues by sub-team with key items, priority, and assignee
  4. Customer Signal — Customer-labeled issues from RHOAIENG and RHAISTRAT, with honest reporting when direct customer labels are zero
  5. Where We See Opportunity — Synthesized analysis of unassigned work, ownership gaps, and strategic alignment
  6. What We Need — Capacity asks, blocker escalations, and process needs for leadership
  7. Next Steps — Action/owner/timeline table derived from blockers and open items

Data Sources

1. AICP Feature Priorities Spreadsheet (Google Sheet)

The Roadmap tab is the source of truth for what AICP owns vs supports:

  • Ownership = "AICP" — Features this team owns and drives
  • Ownership = "Support" — Features another team owns that AICP reviews, consults on, or does supporting engineering for

Exported via gws drive files export and parsed with Python stdlib (zipfile + xml.etree.ElementTree — no pip install needed).

2. Jira REST API v3

Seven queries, all scoped to component in ("AI Core Platform", "AI Core Platform Security"):

#QueryPurpose
1All open issuesPortfolio total
2Status in (In Progress, In Review, In Testing, ...)In-flight by sub-team
3Resolved in lookback periodDelivered, grouped by theme
4Customer labels on RHOAIENGDirect customer signal
5Customer/field labels on RHAISTRATStrategic customer signal
6Blocker/Critical priority, openBlockers snapshot
7In-flight without aicp-team-* labelUnassigned gap

Uses cursor-based pagination (nextPageToken/isLast). Does not use Jira MCP tools — all queries go through the REST API directly.

How It Works

Step 1: Interactive Prompts

The skill asks three questions before collecting data:

  • Lookback period — How far back to look for delivered work (default: 90 days)
  • Output format — Word document, markdown, or both (default: Word)
  • Additional context — Optional strategic notes, meeting outcomes, or leadership asks

Step 2: Data Collection

  1. Exports the Google Sheet as .xlsx and parses the Roadmap tab
  2. Extracts Jira credentials from ~/.claude.json (mcpServers config)
  3. Runs all 7 Jira queries via REST API, paginating to get complete results
  4. Categorizes issues by sub-team label (aicp-team-forge, aicp-team-compass, aicp-team-heimdall)
  5. Groups delivered issues by theme using keyword matching

Step 3: Document Generation

Generates a Word document using python-docx with:

  • Red Hat red (#CC0000) table headers with white text
  • Alternating row shading (#F2F2F2)
  • Thin black borders on all table cells
  • Section headers in Red Hat red, body text in 11pt Calibri

Step 4: Output

Saves to docs/strategy/YYYY-MM-DD_strategy_review.docx (and/or .md), then asks the user for revisions.

Usage

/strategy-review

Or ask Claude Code for a "strategy review", "leadership review", or "portfolio review".

Prerequisites

  • Claude Code with access to this skill
  • Google Workspace CLI (gws) — for exporting the AICP Feature Priorities spreadsheet
  • Jira API token — configured in ~/.claude.json under mcpServers.mcp-atlassian
  • python-docx — for Word document generation

Sub-Teams

TeamLabelFocus Area
Forgeaicp-team-forgexKS Expansion, GitOps & CI/CD
Compassaicp-team-compassE2E Stability, QE & Observability
Heimdallaicp-team-heimdallSecurity, Gateway & Authentication
Ver o original no GitHub

Bom para

Relacionados