convez.ai
PluginAutomaçãoIntermediário
CivicWork

Municipal Governance

CivicWork/municipal-governance

Automação de workflows para gestão municipal: pautas, orçamentos, ordenanças e políticas

Ver no GitHub
InicialPluginScore · de 0 a 100

Plugin Claude para automação de fluxos de trabalho em gestão pública municipal. Oferece análise de ordenanças, preparação de reuniões, pesquisa de políticas, revisão orçamentária e síntese de pautas, com integração a contextos intergovernamentais e monitoramento de políticas.

Instalação

$npx convez-ai CivicWork/municipal-governance

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
3
Forks
0
Acompanhando
0
Saúde
37%
Último commit
há 5 meses
Versão
—
Issues abertas
0
Contribuidores
2

Casos de uso

  • Análise rápida de ordenanças contra código municipal existente
  • Preparação de briefings a partir de pautas de reunião
  • Revisão de impactos fiscais e itens orçamentários
  • Monitoramento de atividades legislativas estaduais/federais com impacto local
  • Avaliação de contratos e propostas de fornecedores
  • Rastreamento de votes, ações e institucional memory pós-reunião

Veredito

Plugin especializado e bem estruturado para gestão municipal, com scaffolding de projeto, configuração automática via busca web e arquitetura modular robusta. Direcionado especificamente a equipes de governo local que buscam reduzir trabalho manual em análise normativa e acompanhamento de políticas.

README

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

Municipal Governance Plugin

AI-powered productivity tools for local government elected officials and staff.

Overview

This plugin automates common workflows for municipal elected officials and staff:

  • Ordinance Analysis: Review proposed ordinances against existing municipal code
  • Meeting Preparation: Generate briefings from agenda packets
  • Policy Research: Multi-source research with intergovernmental context
  • Budget Review: Analyze fiscal impacts and budget items
  • Constituent Communications: Draft responses following municipal communication standards
  • Agenda Synthesis: Quick-turnaround agenda packet summaries
  • Intergovernmental Scanning: Monitor state/federal policy impacts on local government
  • Vendor Evaluation: Analyze vendor contracts, decompose deliverables, assess lock-in, and produce build feasibility specs
  • Meeting Close-Out: Capture votes, follow-ups, and surprises after meetings for persistent institutional memory
  • PolicyAide Sync: Send local profile/context summaries to PolicyAide for deeper multi-agent research
  • Watcher Workflows: Monitor agendas, statehouse activity, vendor renewals, grants, and council follow-ups

Quick Start

  1. Install the plugin
  2. Customize — Click "Customize plugin settings" to configure your municipality. This creates a local municipal.local.md file.
  3. Create workspace — Run the Setup Project agent to scaffold a Cowork Project folder
  4. Try it — Upload an agenda packet and run /municipal-governance:agenda-synthesis

Installation

Cowork (Claude Desktop)

  1. Download or clone this repository
  2. In Cowork, go to Settings > Plugins > Install from folder
  3. Select the CivicWorkPlugin directory
  4. The plugin's commands will appear as /municipal-governance:*

Claude Code (CLI)

  1. Clone this repository
  2. Open Claude Code in the repository directory — plugin files are detected automatically

Setup (Cowork)

  1. Configure your municipality — Click "Customize plugin settings" on the Cowork home screen. Cowork auto-discovers most municipality info via web search (population, government type, council structure, code provider, meeting schedule) and asks you about what it can't find. This writes a local municipal.local.md file.

  2. Create a project workspace — Run the Setup Project agent from the Agents tab. It creates a folder on your computer with the right structure for governance work:

    ~/Municipal-Governance/{your-city}/
    ├── CLAUDE.md              ← Project instructions (auto-generated)
    ├── documents/
    │   ├── agendas/           ← Drop agenda packets here
    │   ├── budgets/           ← Budget documents, AFRs
    │   ├── contracts/         ← Vendor contracts and renewal notices
    │   ├── ordinances/        ← Proposed ordinances under review
    │   └── plans/             ← Comprehensive plan, strategic plan, CIP
    ├── meeting-notes/         ← Post-meeting summaries and action items
    ├── research/              ← Saved policy research
    └── watchers/              ← Local watcher memory files
        ├── agenda-watcher.md
        ├── statehouse-monitor.md
        ├── vendor-renewal-watcher.md
        ├── grant-radar.md
        └── council-follow-up-tracker.md
    

    Then point Cowork to it: "Work in a project" → "+" → "Use an existing folder" → select the created folder.

  3. (Optional) Configure connectors — See Connectors below to enable live municipal code lookup.

Setup (Claude Code CLI)

  1. Run the Setup Municipality agent — it walks you through the full configuration interactively, uses municipal.local.example.md as the starting template when needed, writes a local municipal.local.md, and offers to create a project workspace at the end.
  2. Or copy municipal.local.example.md to municipal.local.md and edit it manually.
  3. Optional connectors are configured locally by copying .mcp.example.json to .mcp.json and updating machine-specific paths.

Why a Project Workspace?

The plugin provides analytical tools. The project workspace provides persistent memory. With a Cowork Project, Claude remembers prior meeting decisions, research, and your preferences across sessions — building institutional knowledge over time. Without it, every conversation starts fresh.

Watcher Memory

The Setup Project agent seeds watchers/*.md from canonical starter templates in templates/watchers/. Each watcher memory file keeps the same simple shape: purpose, last checked, sources watched, current watchlist or open items, recent changes, next check, and notes or verification needed.

Watcher memory is meant to be factual, source-tagged, and public-record-adjacent. Keep campaign strategy, privileged legal advice, closed-session substance, confidential personnel details, private constituent casework, and sensitive negotiation strategy out of these files. Watcher workflows may draft notifications, outreach, or external updates, but they do not send, submit, sync, or post anything without explicit confirmation.

Commands

CommandDescription
/municipal-governance:analyze-ordinanceAnalyze proposed ordinances against municipal code
/municipal-governance:meeting-prepPrepare comprehensive briefings for council/committee meetings
/municipal-governance:policy-researchMulti-source policy research workflow
/municipal-governance:agenda-synthesisQuick synthesis of agenda packets into briefings
/municipal-governance:constituent-responseDraft constituent communications
/municipal-governance:budget-reviewAnalyze budget items and fiscal impact
/municipal-governance:intergovernmental-scanScan state/federal policy impacts
/municipal-governance:vendor-evaluateEvaluate vendor contracts, assess lock-in, and spec open-source alternatives
/municipal-governance:meeting-close-outRecord meeting outcomes, follow-ups, and briefing-book updates
/municipal-governance:sync-to-policyaideSync municipality, official profile, and standing document summaries to PolicyAide
/municipal-governance:skill-qaReview plugin skills for provenance, trust gates, stale references, and connector assumptions
/municipal-governance:agenda-watcherMonitor upcoming agendas, packets, addenda, and agenda changes
/municipal-governance:statehouse-monitorTrack state/federal bills, mandates, preemption risks, and agency updates
/municipal-governance:vendor-renewal-watcherWatch contract renewals, auto-renewal windows, escalation, and data-export deadlines
/municipal-governance:grant-radarMonitor grants, NOFOs, deadlines, eligibility, match, and strategic fit
/municipal-governance:council-follow-up-trackerTrack staff directives, deferred items, promised reports, and open meeting follow-ups

Configuration

The plugin uses a three-layer architecture:

LayerFileWhat it contains
Municipality config templatemunicipal.local.example.mdGeneric starter fields for a new municipality
Municipality configmunicipal.local.mdYour council structure, contacts, fiscal context, priorities
State referencestate-references/{STATE}.mdStatutory requirements, deadlines, thresholds, penalties
Skillsskills/*/SKILL.mdAnalytical frameworks (state-agnostic)

municipal.local.md is created when you customize the plugin. It includes: municipality name, state, government type, council structure, meeting schedule, code provider, budget context, fiscal thresholds, policy priorities, contacts, and committee structure.

Local deployment files are intentionally ignored by git and should not be included in public distribution zips: municipal.local.md, official.local.md, standing-documents.md, briefing-book.md, and .mcp.json. Share municipal.local.example.md and .mcp.example.json as portable templates instead.

State references provide jurisdiction-specific statutory content — statute citations, compliance deadlines, dollar thresholds, penalty structures, and institutional resources. Currently available: Illinois (state-references/IL.md). Additional states can be contributed by following the IL.md structure as a template.

The plugin works for any US municipality. State references add precision for compliance-critical areas; without one for your state, skills provide general framework guidance with a recommendation to verify state-specific requirements with legal counsel.

Skills

The plugin includes 11 domain expertise modules. You don't invoke these directly — Claude activates them automatically based on what you're working on.

SkillWhat it coversExample prompt
municipal-code-analysisCode interpretation, cross-referencing, MunicipalMCP tool reference"Does this amendment conflict with Chapter 19?"
parliamentary-procedureRobert's Rules, scripted chair language, quasi-judicial hearings"What do I say to open a public hearing?"
land-use-zoningZoning, variances, TIF districts, form-based codes, inclusionary zoning"Walk me through the variance criteria for this parcel"
public-financeBudgets, fund accounting, debt, pension/OPEB, CIP planning"What's the pension impact of this staffing proposal?"
intergovernmental-relationsHome rule, preemption decision tree, IGA evaluation, grant scoring"Should we pursue this grant opportunity?"
policy-evaluationBardach framework, logic models, decision matrices, stakeholder analysis"Evaluate these three approaches to short-term rentals"
open-meetings-foiaOMA, FOIA exemption decision tree, closed session framework, response templates"Can we discuss this in executive session?"
council-communicationStaff reports, ordinances vs resolutions, legal drafting, constituent triage"Should this be an ordinance or a resolution?"
ethics-conflictsConflicts of interest, recusal procedures, gift rules, financial disclosure"Do I need to recuse myself from this vote?"
vendor-assessmentVendor lock-in, build-vs-buy, technical decomposition, procurement analysis"Should we renew this contract or build our own?"
vendor-alternativesMunicipal software alternatives, replacement tiers, open-source landscape"What open-source options exist for agenda management?"

Connectors

The plugin can optionally connect to external data sources via MCP servers. Copy .mcp.example.json to .mcp.json for a local deployment, then update any machine-specific paths.

ConnectorDescriptionProvider
municipal-codeSearch and retrieve municipal code sections (7 tools)MunicipalMCP (Municode API)
document-managementRetrieve agenda packets, staff reports, FOIA records, and standing documents when availableBox MCP HTTP endpoint (https://mcp.box.com)

municipal-code requires a local MunicipalMCP installation and machine-specific paths in .mcp.json. document-management is shown in the example as an optional Box MCP endpoint and depends on account access/authentication. Without connectors, commands work with uploaded documents and clearly marked web or model-derived assumptions.

How It Works

  1. You run a command (e.g., /municipal-governance:meeting-prep) and upload a document
  2. Claude asks a few quick scoping questions — what items matter to you, how deep should the analysis go
  3. Claude loads your municipality's configuration from municipal.local.md and the applicable state reference from state-references/
  4. Relevant skills activate automatically based on the subject matter
  5. You get structured output with attention-priority flags (red/yellow/green), confidence and provenance indicators on key claims, and recommended next steps — matched to the depth you asked for
  6. For high-stakes items, the output includes an Analysis Boundaries section — a transparent disclosure of what the analysis can and cannot responsibly conclude as a single AI instance, with specific recommendations for verification (attorney review, staff confirmation, or escalation to PolicyAide's multi-agent deliberation framework)

Why Analysis Boundaries Matter

Most AI tools pretend they can handle everything. This plugin tells you when it's not enough. A single AI instance is great for research, organization, and synthesis — but policy decisions with real fiscal, legal, or community impact deserve adversarial testing, independent verification, and human judgment. The plugin knows its own limits and says so.

License

Apache 2.0 License — Built by CivicWork

Ver o original no GitHub

Bom para

Relacionados