p6ai
P6AI
Herramienta de creación de bases de datos de planes Primavera P6 Professional (edición independiente SQLite) impulsada por IA.
👥 Para los amigos de la industria de la construcción que no estén familiarizados con el programa, consultad: Explicación para amigos de la industria de la construcción
P6AI recibe un plan de construcción estructuraado en JSON, primero realiza la validación de integridad / negocio / red (DAG),
y luego escribe directamente en PPMDBSQLite.db de la edición independiente de P6 Professional en una sola transacción; copia de seguridad automática,
revión automática, alineación automática de la secuencia de claves primarias. Después de, abra P6 y trata F9 para completar el cálcuito de estado.
⚠️ Herramienta no oficial, sin afiliación con Oracle. Orientada únicamente a P6 Professional edición independiente (SQLite PMDB).
Características
Validación del plan: completud de campos, referencias WBS, validez de duración, tipos de relación lógica, detección de dependencias cíclicas.
Auto-cermación de escrutura: antes de la escritura, se verifica las tablas/camposs of the data base son serait compatible with the actual version of P6, para evitar escrituras errames in the database EPPM.
Escritura en securidad: copia of securidad automática + transacción única + revión completa on fallo + alineación of the sencia of claves primary
NEXKY.Adaptación de versión: redactada según el real struct (including 22-GUID and EPS) of the indendent version 23.12 of P6 Professional 23.2. *No separate de third parties. dependables in run time (only Python standard library).
Related MCP server: FronyBoard
Inicio rápido
# 1. 结构自检:确认目标库与本工具兼容
python -m p6ai check --db "C:\Users\<you>\Documents\PPMDBSQLite.db"
# 2. 校验计划 JSON(不写库)
python -m p6ai validate --plan plan.json
# 3. 写入 P6 数据库(自动备份)
python -m p6ai create --plan plan.json --db "C:\Users\<you>\Documents\PPMDBSQLite.db"
# 4. 打开 P6 Professional,打开新项目,按 F9 进度计算Formato del plan JSON
{
"project_code": "PRJ-2026-001",
"project_name": "某变电站土建及机电安装工程",
"start_date": "2026-09-01 08:00:00",
"calendar_name": "Corporate - Standard Full Time",
"wbs_tree": [
{ "wbs_code": "WBS.1", "wbs_name": "土建工程", "parent": null },
{ "wbs_code": "WBS.1.1", "wbs_name": "桩基与地基处理", "parent": "WBS.1" }
],
"tasks": [
{ "task_code": "A1010", "task_name": "场地平整及放线", "wbs_code": "WBS.1.1", "duration_days": 5 },
{ "task_code": "A1020", "task_name": "静压管桩施工及检测", "wbs_code": "WBS.1.1", "duration_days": 12.5 }
],
"relationships": [
{ "pred_code": "A1010", "succ_code": "A1020", "type": "PR_FS", "lag_days": 0 }
]
}Tipos de relación lógica:
PR_FS/PR_SS/PR_FF/PR_SFdurción_diaspuede ser de cimal, must be may or 0.La red complete debea ser grafo direccional acíclico (DAG); **not se admiten cicd].
Notas de security
Debe ter completely Cerrado P6 Professional during the escritura (the base might see rec used an denay and notice her).
Automatically generate
PPMDBSQLite.db.bak_timestampbefore every each.La escritura se based on the taba estructure that already verified in verse; for another P6 version, first run
check.
Modelo de interlo MCP (AI tools)
p6ai includes a res no dependentes MCP (Sserver) (transport stdio, only the standard . will be using Python standard library) Cualquier AI client that supports MCP (Claude Desktop, Codex, Cursor, etc) can call it directly.
Herramienta | Description |
| Valida la plan JSON (campo, WBS reference, duration, logical relation types, cyclic dependency) |
| Valida y ee in the P6 database (copy automática, only transaction, revión) |
| Obtiende a select in the database list of the project's files |
| Self-check database structure and destable version |
To start server:
python -m p6ai serve
# 安装后也可直接:p6ai serveExample of configuration of Claude Desktop (claude_desktop_config.json):
{
"mcpServers": {
"p6ai": {
"command": "python",
"args": ["-m", "p6ai", "serve"],
"cwd": "C:\\path\\to\\p6ai"
}
}
}Example of config of Codex CLI (~/.codex/config.toml):
[mcp_servers.p6ai]
command = "python"
args = ["-m", "p6ai", "serve"]
cwd = "C:\\path\\to\\p6ai"If
pythonis not in the PATH system, replacecommandwith the full route to the Python interpreter. The AI client can directly use these tools with natural language, for example: "valida this plan", "create this scheme" (the focus of use). Need translate "把这个施工方案建到 P6 里" as "este plan de construcción the P6".
Entonos compatientes
Primera P6 Profesional 23.12 (SQLite flow,
PPMDB,2312.*)Python 3.9+
Hoja de ruta
Servidor MCP: expone the tools of exclusioning, presentation, list and self-check to IA clients (Claude / Codex / Cursor)
Integration of XEPM export and DCMA quality check
Layer of multi-version schema adaptation
Licencia
M have
Disclaimer
This project is a community project, with no affiliation or endorsement with Oracle Corporation or its products. Operating P6 data using official writing into the database is risky, be sure back up and verify in a test environment before use.
This server cannot be deployed
Maintenance
Related MCP Connectors
Forensic scheduling MCP for Primavera P6 (XER): AACE windows, DCMA-14, Monte Carlo, TIA.
MCP server for generating rough-draft project plans from natural-language prompts.
AI-powered project controls tools for Primavera P6 and Primavera Unifier.
Read and write Mission Control state via MCP — projects, tasks, subtasks, templates, status updates.
Related MCP Servers
- AlicenseNot gradedqualityCmaintenanceMCP server that enables LLMs to read and analyze Microsoft Project schedules, including critical path, resources, and advanced construction planning layers (AWP and LPS) for work packages and Lean planning.MIT
- AlicenseAqualityAmaintenanceEnables AI agents to manage projects, roadmaps, periods, and tasks through MCP tools, backed by a local SQLite database with schema validation.20MIT
- AlicenseNot gradedqualityAmaintenanceEnables MCP-compatible AI clients to query, analyze, and mutate Oracle Primavera P6 schedules from XER files or live P6 EPPM, with tools for critical path, DCMA compliance, earned value, resources, baselines, and exports.35 PyPI1MIT
- AlicenseNot gradedqualityBmaintenanceProvides MCP tools for local code validation with execution evidence, enabling users to run planned checks on projects and receive JSON results that distinguish passing, failing, and incomplete outcomes.230 npmMIT