antnewman · antnewman · May 9, 2026
@@ -1,14 +1,14 @@
 # Claude Code Instructions — PDA Platform
 
 ## Project Overview
-The PDA Platform is a Python monorepo providing AI-powered project delivery assurance for UK government. It exposes 122 MCP tools across 18 modules via a unified server, deployable locally or via Render.
+The PDA Platform is a Python monorepo providing AI-powered project delivery assurance for UK government. It exposes 124 MCP tools across 18 modules via a unified server, deployable locally or via Render.
 
 ## Repo Structure
 ```
 packages/
   pda-platform/          Meta-package (pip install pda-platform)
   pm-data-tools/         Core library: parsers, validators, AssuranceStore (SQLite)
-  pm-mcp-servers/        18 MCP modules, 122 tools
+  pm-mcp-servers/        18 MCP modules, 124 tools
   agent-task-planning/   AI reliability: confidence extraction, outlier mining
 docs/                    Practitioner guides and technical references
 .github/workflows/       publish.yml — PyPI publish on version tag
@@ -68,7 +68,7 @@ Use semantic versioning:
 
 ## MCP Tool Count — Keep in Sync
 
-The tool count (currently **122**) appears in multiple places. When adding new tools, update ALL of these:
+The tool count (currently **124**) appears in multiple places. When adding new tools, update ALL of these:
 
 | File | What to update |
 |---|---|

@@ -12,7 +12,7 @@
 
 ---
 
-PDA Platform gives Claude and other MCP-compatible AI assistants **122 structured tools** to perform rigorous, evidence-based project delivery assurance aligned to IPA Gate Reviews (Gates 0–5), HM Treasury Green Book appraisal, and GMPP reporting standards. It is the only open-source MCP platform specifically designed for Senior Responsible Owners, Project Managers, Independent Assurance Reviewers, and Portfolio Managers working in UK government major project delivery.
+PDA Platform gives Claude and other MCP-compatible AI assistants **124 structured tools** to perform rigorous, evidence-based project delivery assurance aligned to IPA Gate Reviews (Gates 0–5), HM Treasury Green Book appraisal, and GMPP reporting standards. It is the only open-source MCP platform specifically designed for Senior Responsible Owners, Project Managers, Independent Assurance Reviewers, and Portfolio Managers working in UK government major project delivery.
 
 Think of it as a **semantic layer for project data**: it ingests schedules from Primavera, MSP, Jira, and 5 other formats; normalises them into a consistent delivery model; pre-loads IPA benchmark statistics from five years of Annual Reports; and exposes everything through a structured API that AI can reason about directly — without parsing raw XML or making things up.
 
@@ -22,6 +22,19 @@ Connect Claude to the platform and ask it to conduct a full gate readiness asses
 
 ---
 
+## What it does, in 90 seconds
+
+<!-- LOOM_EMBED_PLACEHOLDER -->
+*A 90-second walkthrough of the platform on a real assumption register lives here. (Loom link to be added.)*
+
+In the video you'll see the platform:
+
+- **Generate the questions a senior gate reviewer would ask** — from your project data, with no setup
+- **Score and validate every assumption in your business case** against live external signals from ONS and the World Bank
+- **Produce a board-ready exception report** grounded in evidence rather than narrative
+
+---
+
 ## Who is this for?
 
 | Role | What the platform does for you |
@@ -67,9 +80,23 @@ See the full **[Connection Guide](docs/connection-guide.md)** for Claude.ai, Cla
 
 ---
 
-## Start here: scan for red flags
+## Start here: two ways in, no setup needed for the first
 
-The fastest way to get value from the platform is a single tool call:
+**Try this first** — works against zero project data, returns immediately:
+
+```
+generate_premortem_questions
+```
+
+Returns the questions an experienced IPA reviewer would actually ask at a gate review:
+
+> *"What is the programme team not telling senior stakeholders — and why?"*
+> *"Which assumption, if wrong, would most severely undermine the current delivery confidence rating?"*
+> *"Is the risk register a genuine picture of what could go wrong — or a curated list of manageable risks?"*
+
+Tailor the output to a specific gate stage (`gate_stage: 3`) or risk profile to focus the questions further.
+
+**Once you've loaded a project**, the next call is:
 
 ```
 scan_for_red_flags — project_id: "PROJ-001"
@@ -120,7 +147,17 @@ Once connected, Claude can answer questions like:
 
 ---
 
-## 17 MCP Modules
+## Built on PDA Platform
+
+Products built on the PDA Platform engine:
+
+- **[Evidence Engine](https://evidence-engine.netlify.app)** — a forecast-driven early-warning system for project assumption drift. Built by Laura White as the MPA Challenge 5 hackathon entry. Wraps PDA Platform with a forecasting layer, cascade propagation across the assumption dependency graph, and a confidence decomposition that turns assurance evidence into a live early-warning signal. Calls `generate_narrative` over MCP SSE; falls back gracefully if the engine is unreachable.
+
+If you've built something on PDA Platform, open an issue or PR — we'll add it here.
+
+---
+
+## 18 MCP Modules
 
 | Module | Tools | What it does |
 |--------|------:|-------------|
@@ -142,7 +179,7 @@ Once connected, Claude can answer questions like:
 | pm-lessons | 5 | AI extraction of lessons from gate reviews/PIRs, cross-project keyword search, systemic pattern analysis |
 | pm-reporting | 6 | IPA-format gate review summaries, SRO dashboards, board exception reports, portfolio summaries, PIR templates, UDS export |
 | pm-assumptions | 8 | Assumption drift detection, confidence scoring, live external signal integration (ONS, World Bank), AI executive reports, cascade analysis, UDS dashboard |
-| **Total** | **122** | One unified endpoint · One connection |
+| **Total** | **124** | One unified endpoint · One connection |
 
 ---
 

@@ -4,7 +4,7 @@
 
 ## What it is
 
-PDA Platform (Project Data Analytics) is a semantic layer for project delivery data. It ingests schedules from Primavera, MSP, Jira, and 5 other formats; normalises them into a consistent delivery model; pre-loads IPA benchmark statistics from five years of Annual Reports; and exposes everything through 122 structured MCP tools that AI can reason about directly.
+PDA Platform (Project Data Analytics) is a semantic layer for project delivery data. It ingests schedules from Primavera, MSP, Jira, and 5 other formats; normalises them into a consistent delivery model; pre-loads IPA benchmark statistics from five years of Annual Reports; and exposes everything through 124 structured MCP tools that AI can reason about directly.
 
 The platform is aligned to UK government delivery standards: IPA Gate Reviews (Gates 0-5), HM Treasury Green Book, GMPP reporting, and Cabinet Office project controls.
 

@@ -2,7 +2,7 @@
   "$schema": "https://static.modelcontextprotocol.io/schemas/2025-09-16/server.schema.json",
   "name": "io.github.antnewman/pda-platform",
   "title": "PDA Platform",
-  "description": "122 MCP tools for UK government IPA Gate Review assurance: risk, benefits, Monte Carlo simulation, assumption drift detection, narrative divergence, EV, gate readiness, and portfolio analysis.",
+  "description": "124 MCP tools for UK government IPA Gate Review assurance: risk, benefits, Monte Carlo simulation, assumption drift detection, narrative divergence, EV, gate readiness, and portfolio analysis.",
   "version": "1.0.0",
   "status": "active",
   "websiteUrl": "https://github.com/antnewman/pda-platform",