# Delphai > DELPHAI - multi-agent certification-readiness council on Microsoft Foundry (Agents League Battle #2). Real Foundry IQ on Azure AI Search, 11 hosted agents, GO/NEGOTIATE/NO-GO. - **Type:** MCP server - **Install:** `agentstack add mcp-jlynch160-delphai` - **Verified:** Yes β€” security-reviewed for prompt injection and unsafe behavior - **Seller:** [jlynch160](https://agentstack.voostack.com/s/jlynch160) - **Installs:** 0 - **Category:** [Cloud & Infrastructure](https://agentstack.voostack.com/c/cloud-infrastructure) - **Latest version:** 0.1.0 - **License:** MIT - **Upstream author:** [jlynch160](https://github.com/jlynch160) - **Source:** https://github.com/jlynch160/delphai - **Website:** https://delphai.wittyocean-5b0d21f3.westus3.azurecontainerapps.io/ ## Install ```sh agentstack add mcp-jlynch160-delphai ``` Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent. ## About # πŸ›οΈ DELPHAI ### The certification-readiness council that checks the math β€” then refuses to burn out your team *Named for **Delphi**, the oracle the ancient world consulted before it committed.* [](LICENSE) [](#) [](#) [](#) [](#) [](#observability-evaluation--proof) [](https://delphai.politedune-a7af3b6c.westus3.azurecontainerapps.io/) [](https://youtu.be/1CFzoLB6_fU) **Microsoft Agents League Β· Battle #2 β€” Reasoning Agents with Microsoft Foundry** [**β–Ά Live demo**](https://delphai.politedune-a7af3b6c.westus3.azurecontainerapps.io/) Β· [**Watch the video**](https://youtu.be/1CFzoLB6_fU) Β· [Architecture](#-architecture) Β· [The agents](#-the-council--11-reasoning-agents) Β· [How it works](#-how-it-works) Β· [Getting started](#-getting-started) --- ## πŸ“‹ Table of contents - [Overview](#-overview) - [The problem](#-the-problem) - [The solution](#-the-solution) - [What you'll see in the demo](#-what-youll-see-in-the-demo) - [What makes it different](#-what-makes-it-different) - [Key features](#-key-features) - [Architecture](#-architecture) - [The council β€” 11 reasoning agents](#-the-council--11-reasoning-agents) - [Microsoft IQ integration](#-microsoft-iq-integration) - [Tools & grounding](#-tools--grounding) - [How it works](#-how-it-works) - [A run, end to end](#-a-run-end-to-end) - [Tech stack](#-tech-stack) - [Getting started](#-getting-started) - [Project structure](#-project-structure) - [Submission package](#-submission-package) - [The synthetic dataset](#-the-synthetic-dataset) - [The math](#-the-math) - [Observability, evaluation & proof](#-observability-evaluation--proof) - [Responsible AI & security](#-responsible-ai--security) - [Battle #2 rubric mapping](#-battle-2-rubric-mapping) - [FAQ](#-faq) - [Roadmap](#-roadmap) - [Acknowledgments](#-acknowledgments) - [License](#-license) --- ## πŸ”­ Overview **DELPHAI** is a multi-agent enterprise-learning system that manages internal team certification programmes β€” and reasons about whether an organisation (or a single learner) will *actually be ready* by a deadline. It curates grounded study content, generates cited practice questions, gates readiness per exam domain, and runs an **11-agent debate** (optimist βš” skeptic βš” wellbeing-advocate) over an **exact, computed forecast** before an executive commits β€” delivering a **GO / NEGOTIATE / NO-GO** verdict, with the spine to say *no*. It runs live on **Microsoft Foundry** (gpt-4.1) across **three execution surfaces** β€” Foundry Agent Service hosted agents, the Microsoft Agent Framework, and a custom streaming orchestrator β€” grounded by **Foundry IQ on Azure AI Search**, the **Microsoft Learn MCP** server, and an agent-run **code interpreter**. > [!NOTE] > **Synthetic data only.** Every identifier (`L-2001`, `KB-SEC-001`, the fictional "Aegis" security > team) is fabricated β€” no real people, PII, customer data, or credentials. Wellbeing signals are used > only to build humane plans, **never to rank people**. --- ## ❗ The problem Engineering leaders constantly make a high-stakes promise: *"Yes, my team will be certified in time."* Then it goes wrong in predictable ways: - **The estimate is a vibe, not a calculation.** "We'll be fine" is a gut call on individual averages, not the actual probability that *K of N* engineers pass by a date. - **A single average hides a fatal gap.** A 78% average can mask a learner who is at 40% on the heaviest exam domain β€” and that's the one that fails them. - **Capacity is ignored.** People already drowning in meetings get handed a study plan that assumes free evenings, and the date is "hit" by quietly burning them out. - **AI tools only ever say *yes*.** Most copilots are relentlessly optimistic. None of them will tell a director the honest answer is *no* β€” or refuse an order to overwork the team. --- ## πŸ’‘ The solution DELPHAI turns that promise into a **defensible, grounded decision**: 1. **One question in** β€” *"Can we certify 5 of 8 engineers for SC-200 in 5 weeks?"* 2. **A council reasons over computed numbers** β€” an exact **Poisson-binomial** team forecast, a **per-domain coverage gate**, and **Work IQ-aligned capacity**, all deterministic and cited. 3. **The agents argue** β€” an optimist Planner, a Red-Team Skeptic (who can **veto**), and a Wellbeing advocate (who can **refuse**) debate the *same* numbers with *different* assumptions. 4. **One verdict out** β€” **GO / NEGOTIATE / NO-GO**, with the concrete conditions to get to *yes*, a per-person week-by-week plan, booked `.ics` study blocks, and a manager email brief. The model **narrates** the reasoning; it does **not invent** the numbers. --- ## 🎬 What you'll see in the demo > **[β–Ά Watch the 3:25 demo video](https://youtu.be/1CFzoLB6_fU)** β€” the quickest path through the reasoning council, architecture map, math-tool overrule, and responsible-AI refusal. > **[β–Ά Open the live app](https://delphai.politedune-a7af3b6c.westus3.azurecontainerapps.io/)** β€” no login. Suggested path: **Build my path β†’ Whole team β†’ SC-200 β†’ 5 of 8 β†’ 5 weeks β†’ Convene.** 1. **The ask.** A readiness console takes one question β€” cert Β· headcount Β· deadline β€” then a live ingestion screen brings Foundry IQ, IQ-aligned context/semantic layers, and the Microsoft Learn MCP "online". 2. **The debate, streamed.** Eleven advisors reason token-by-token on gpt-4.1, each turn showing real latency + token telemetry. The optimist lands ~93–96%. 3. **The tool-overrule** *(the moment to watch).* The Skeptic runs a code-interpreter tool, computes the exact Poisson-binomial β€” **77%** β€” and strikes through the optimist's guess. 4. **The refusal.** Order the team past its limits and the Wellbeing agent **declines the instruction** outright β€” on the record. 5. **The verdict.** A **GO / NEGOTIATE / NO-GO** gauge resolves, the tug-of-war needles settle, and a per-person plan appears with `.ics` study blocks and a manager email brief. 6. **The Live Architecture Map.** An interactive map traces a request end-to-end through every component; click any node to inspect it. --- ## ⭐ What makes it different Most submissions wrap five chatbots around a study plan. DELPHAI makes **reasoning the product**: - **πŸ—£οΈ The agents disagree on purpose.** Optimist vs. skeptic vs. wellbeing argue the *same* forecast with *different* assumptions. The verdict is **reconciled, not asserted**. - **πŸ›‘ It runs a tool to overrule itself.** The Skeptic invokes a **code-interpreter tool** to compute the exact Poisson-binomial β€” **77%** β€” and overrules the optimist's vibes-y **93–96%**. The math wins, *live, on the agent itself*. - **πŸ›‘οΈ It refuses to break people.** Order it to push the team past its limits and the Wellbeing agent **declines the instruction outright**. It safeguards people, never ranks them. - **πŸ”’ Every number is computed, not hallucinated** β€” pass-probability, capacity, skill-adjacency ramps, and the team forecast are all traceable to a cited source. --- ## ✨ Key features | | Feature | |---|---| | 🧠 | **11-agent reasoning council** that debates, cross-examines, reconciles, and negotiates | | βš–οΈ | **GO / NEGOTIATE / NO-GO** verdict from an exact **Poisson-binomial** team forecast | | πŸ”Ž | **Foundry IQ grounding** on real **Azure AI Search** (hybrid vector + keyword, verified citations) | | πŸ”Œ | Live **Microsoft Learn MCP** retrieval (`microsoft_docs_search`) for official guidance | | βš™οΈ | Agent-run **code interpreter** for the math (not a guess) | | πŸ‘€πŸ‘₯ | **Two modes, one engine** β€” learner self-service (+ AI tutor) and manager team council | | πŸ“… | The agent **acts** β€” per-person plans exported as `.ics` blocks + a manager email brief | | πŸ›‘οΈ | **Responsible-AI refusal** + a calibration scorecard of past-call accuracy | | πŸ“Š | Live **telemetry** (latency + tokens per turn), **8/8 evaluations**, captured proof artifacts | | πŸ–₯️ | Self-contained, zero-dependency dashboard with an interactive **Live Architecture Map** | --- ## πŸ—οΈ Architecture ```mermaid flowchart TB U["Manager / Learner β€” DELPHAI UI (quorum.html)"] U -->|"cert Β· headcount Β· deadline"| SRV["Flask Β· server.py/api/council/stream (SSE) Β· /api/ask Β· /api/learn"] SRV --> ORC["Orchestrator Β· src/orchestrator.py"] ENG["Reasoning engine Β· src/readiness.pypass-probability Β· coverage gate Β· Poisson-binomial forecast"] --> ORC subgraph COUNCIL["11-agent reasoning council (debate Β· reconcile Β· negotiate)"] direction LR A1["Conductor"] A2["Planner / Skeptic (veto)"] A3["Assessment / Coach / Study-Plan"] A4["Wellbeing (refuses overwork) / Engagement"] A5["Curator / Verifier / Historian"] end ORC --> COUNCIL COUNCIL -->|narration| MODEL["Microsoft Foundry β€” gpt-4.1 (quorum-gpt41)"] COUNCIL -. grounding .-> FIQ["Foundry IQ β€” Azure AI Searchhybrid vector + keyword"] COUNCIL -. work signals .-> WIQ["Work IQ-aligned contextcalendars Β· capacity"] COUNCIL -. ontology .-> FAB["Fabric IQ-aligned ontologyroleΒ·certΒ·skillΒ·threshold"] COUNCIL -. live tools .-> MCP["Microsoft Learn MCP Β· code interpreter"] AF["Microsoft Agent Framework Β· foundry_agents.py"] --> MODEL HOST["Foundry Agent Service Β· 11 hosted agents (asst_…)"] --> MODEL ORC --> OUT["Verdict: GO / NEGOTIATE / NO-GO+ per-employee plans Β· booked .ics Β· manager email"] OUT --> U ``` **Three execution surfaces, one council** (all on Foundry gpt-4.1): | Surface | What it is | Reproduce | |---|---|---| | **Foundry Agent Service** | All 11 advisors registered as *persistent hosted agents* (managed runtime + Entra agent identity, `asst_…` IDs) on project `delphai`; a live tool-using debate on one shared thread | `python foundry_hosted_agent.py` Β· `python foundry_council_live.py` | | **Microsoft Agent Framework** | The council as real `agent_framework` agents (`OpenAIChatClient.as_agent`, Responses API) | `python foundry_agents.py` | | **Custom orchestrator** | The streaming demo UI (SSE token-by-token + per-turn telemetry) behind the public URL | `python server.py` | All three are captured with agent IDs + transcripts in [`AGENT_FRAMEWORK_PROOF.md`](AGENT_FRAMEWORK_PROOF.md). --- ## 🧠 The council β€” 11 reasoning agents Each advisor is a real `agent_framework` agent in [`foundry_agents.py`](foundry_agents.py) **and** a streaming persona in `src/agents/` β€” same responsibilities, multiple execution surfaces. Plus **Kai**, the learner-mode AI tutor. | Agent (persona) | Responsibility | Reasoning pattern | |---|---|---| | **Conductor** β€” Dana Whitfield | Chairs the council, reconciles the debate, makes the GO / NEGOTIATE / NO-GO call, negotiates terms | Planner / Moderator | | **Planner** β€” Ben Russo | Optimistic, best-case team forecast and the confident case it can land | Planner–Executor | | **Skeptic** β€” Vera Lindqvist | Red-teams the forecast, weights first-pass/retest risk, **can VETO** β€” the "AI that says no" | Critic / Verifier | | **Assessment** β€” Nadia Okonkwo | Judges readiness *per exam domain*; writes grounded, cited practice questions; enforces the β‰₯75% coverage gate | Grounded generation + critic | | **Coach** β€” Sam Ellison | Enablement uplift β€” what behind learners should study and how much it moves the odds | Remediation / self-reflection | | **Study-Plan** β€” Leo Nakamura | Converts content into a capacity-aware weekly schedule; moves the *deadline*, not the willpower | Constraint planner | | **Wellbeing** β€” Maya Devlin | Protects people from overwork; **can VETO** and **refuses an unethical overwork order** | Responsible-AI guardrail | | **Engagement** β€” Ruth Adler | Times reminders to each person's work-context rhythm; eases off the overloaded | Context-aware nudging | | **Curator** β€” Theo Park | Maps roleβ†’cert and grounds every claim in an approved source (Foundry IQ / Microsoft Learn) | Grounded retrieval | | **Verifier** β€” Omar Said | Fact-checks every claim; flags stale/unsupported data; integrity score | Anti-hallucination | | **Manager-Insights** β€” Iris Vaughn | Aggregate, no-PII team-readiness summary + calibration scorecard | Memory / calibration | | **AI Tutor** β€” Kai Ferreira *(learner mode)* | Answers learner questions live; generates scenario-based practice calibrated to readiness | Socratic tutor | > Wellbeing signals (meeting load, after-hours, life context) are used **only to accommodate people, > never to rank them**. The refusal is real and reproducible β€” see [`AGENT_FRAMEWORK_PROOF.md`](AGENT_FRAMEWORK_PROOF.md). --- ## 🧬 Microsoft IQ integration DELPHAI implements a live **Foundry IQ** grounding layer and includes Work IQ / Fabric IQ-aligned synthetic layers for the rest of the enterprise-learning reasoning model (the rubric requires at least one Microsoft IQ layer): | Layer | Role in DELPHAI | Implementation | |---|---|---| | **Foundry IQ** | Grounded retrieval + citations for the Curator, Assessment, and Verifier agents | The synthetic KB is indexed into a real **Azure AI Search** service (`delphai-search`, index `delphai-knowledge`) and retrieved with **hybrid vector + keyword** search (`text-embedding-3-small`, RRF fusion). `src/connectors/foundry_iq.py` Β· `scripts/build_foundry_iq.py` | | **Work IQ-aligned context layer** | Capacity & engagement reasoning β€” study windows around work rhythm | `data/work_signals.json` (meeting load, focus hours, after-hours) via a swappable connector with a Microsoft Graph calendar stub (`WORK_IQ_SOURCE=graph`). `src/connectors/graph_calendar.py` | | **Fabric IQ-aligned semantic layer** | Role Β· cert Β· skill Β· exam-domain Β· threshold relationships | `data/certifications.json` is a certification ontology (skills, exam-domain weights, prerequisites, first-pass rates) powering skill-adjacency ramps and the coverage gate. | > **The connector swap seam:** agents consume a *derived signal shape*, never the source, so connectors > swap with zero agent changes (synthetic ⇄ live Graph; local corpus ⇄ Azure AI Search). --- ## πŸ”§ Tools & grounding | Tool | Use | Where | |---|---|---| | **Azure AI Search** (Foundry IQ) | Hybrid vector + keyword retrieval over the approved corpus, with verified citations | `src/connectors/foundry_iq.py` | | **Microsoft Learn MCP** | Live official exam content via `microsoft_docs_search` (JSON-RPC over Streamable HTTP) | `src/connectors/learn_mcp.py`, `/api/learn` | | **Code interpreter** | The Skeptic computes the exact Poisson-binomial team probability β€” live, on the agent | `foundry_council_live.py` | Every grounding path **degrades gracefully** β€” if a service is unconfigured or down, the system falls back to a deterministic local path, so the demo never breaks. --- ## βš™οΈ How it works ### Two ways in β€” same engine, same agents **πŸ‘€ One learner (baseline flow).** Pick a cert β†’ a curated, grounded path β†’ a capacity-aware study plan β†’ engagement nudges β†’ a **scenario-based, calibrated assessment** β†’ pass β‡’ advance, fail β‡’ loop back. Two sub-modes: **self-service** ("this is me", with the **Kai** AI tutor) and **manager/L&D** (pick a person and build their path). **πŸ‘₯ Whole team (manager council).** *"Can we certify N engineers for cert X by the deadline?"* β†’ the 11-agent council debates the computed forecast β†’ **GO / NEGOTIATE / NO-GO** + per-person plans, booked `.ics` blocks, and a manager email brief. ### The reasoning flow ``` Conductor.intake β†’ Curator (grounding + citations) β†’ Assessment (cited question + per-domain coverage gate) β†’ Planner ──┐ optimist team probability β†’ Skeptic ─── attacks it with realistic assumptions (may VETO) β†’ Coach ─── uplift plan: weak domains, sources, projected score gain β†’ Wellbeing β”€β”€β”˜ human-sustainability check (may VETO / refuse) β†’ Verifier (grounding integrity score) β†’ Conductor.reconcile (skeptic-weighted verdict + GO / NEGOTIATE / NO-GO) β†’ Conductor.negotiate (counter-offers: extend deadline / add / swap candidate) β†’ Historian (trust scorecard) β†’ Briefing (executive report) ``` --- ## 🎯 A run, end to end *Scenario: a manager asks DELPHAI to certify **5 of 8** engine … ## Source & license This open-source MCP server is cataloged on AgentStack and links to its original source β€” we do not rehost the code. - **Author:** [jlynch160](https://github.com/jlynch160) - **Source:** [jlynch160/delphai](https://github.com/jlynch160/delphai) - **License:** MIT - **Homepage:** https://delphai.wittyocean-5b0d21f3.westus3.azurecontainerapps.io/ Install and usage instructions live in the source repository linked above. ## Pricing - **Free** β€” Free ## Security capabilities Automated source analysis of v0.1.0 β€” what this tool can access: - **Network access:** no - **Filesystem access:** no - **Shell / process execution:** no - **Environment & secrets:** yes - **Dynamic code execution:** no *"Yes" means the capability is present in the source β€” more access means more to trust, not that it is unsafe.* ## Versions - **0.1.0** β€” security scan: passed β€” Imported from the upstream source. ## Links - Listing page: https://agentstack.voostack.com/l/mcp-jlynch160-delphai - Seller: https://agentstack.voostack.com/s/jlynch160 - Browse the marketplace: https://agentstack.voostack.com/browse --- Listed on AgentStack β€” the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.