Feature - Connect Copilot Studio and Azure repository with Agent forge

Description
Review and improve the FAF Implementation Recommender Agent so it can recommend Copilot-first accelerators and evaluate Copilot Studio-oriented GitHub repositories with the right metadata and discovery signals.

The current Implementation Recommender already evaluates accelerator repositories, prefers AGENTS.md as the discovery source, extracts metadata, and scores each accelerator across general dimensions. We need to extend this flow so Copilot-first scenarios are handled intentionally rather than being evaluated only through the existing Foundry/pro-code-oriented accelerator comparison model.

Current State
Relevant information is currently captured mostly in:

.github/agents/FAF-impl-recommender.agent.md
Workflow 1 extracts solution-metadata.yaml from the solution spec.
Workflow 1 reads the available accelerators catalogue and evaluates GitHub repositories.
Workflow 1 already retrieves each repository's AGENTS.md first and falls back to README/root structure if AGENTS.md is unavailable.
.github/agents/references/implementation-recommender/solution-metadata-template.yaml
Defines the technical metadata used for accelerator comparison.
Current metadata covers use case, agentic patterns, data scenario, API integration, UI scenario, AI stack, hosting, team profile, infrastructure, and implementation strategy.
.github/agents/references/implementation-recommender/accelerators-scoring-card.md
Defines the scoring dimensions and thresholds for Agentic Customization, Further Assessment Required, or Agentic Development.
.github/agents/references/implementation-recommender/accelerators-catalogue.md
Lists the current general-purpose GitHub accelerators.
Requested Change
Review and improve:

Solution metadata for accelerator comparison

Evaluate whether solution-metadata-template.yaml needs additional fields to represent Copilot-first and hybrid solutioning requirements.
Consider adding or refining metadata for:
Selected solutioning platform: Copilot Studio first, Foundry first, or hybrid.
Copilot Studio channel needs: Teams, M365 Copilot, web, custom website, other channels.
Knowledge source patterns: SharePoint, Microsoft Graph, Dataverse, public websites, uploaded files, connector-backed sources, custom APIs.
Action/tool patterns: Copilot Studio actions, connectors, OpenAPI plugins, Power Automate flows, MCP/API-backed tools.
ALM and governance expectations: environments, solution packaging, DLP, tenant policies, ownership model.
Customer delivery model: low-code makers, business owners, IT makers, pro-code developers, mixed team.
Hybrid extension needs where Copilot Studio is the primary user/channel layer but Foundry or custom Azure services provide retrieval, advanced tools, APIs, or orchestration.
Copilot accelerator repository evaluation guidance

Define how the Implementation Recommender should evaluate a GitHub repository as a Copilot accelerator.
Keep AGENTS.md as the preferred discovery source, but define Copilot-specific expectations for what the file should describe so scoring can be reliable.
The Copilot accelerator AGENTS.md should help the agent evaluate dimensions such as:
Supported Copilot Studio scenario patterns.
Supported knowledge/source patterns.
Supported actions, connectors, plugins, OpenAPI, Power Automate, or MCP/API integrations.
Environment setup, import/deployment steps, ALM, and governance assumptions.
Required licenses, tenant prerequisites, and configuration constraints.
Extensibility points for hybrid Foundry/custom Azure services.
Known limitations and customization boundaries.
If a Copilot accelerator does not include AGENTS.md, treat that as a discovery and confidence gap rather than relying only on README content.
Copilot accelerators catalogue

Add a new markdown reference file containing approved Copilot accelerator GitHub URLs, for example:
.github/agents/references/implementation-recommender/copilot-accelerators-catalogue.md
Instruct the Implementation Recommender Agent to read this Copilot accelerator catalogue when the solution metadata or solution spec indicates a Copilot-first scenario.
For Copilot-first scenarios, the agent should prioritize evaluating Copilot accelerators from the new catalogue before or alongside the general accelerator catalogue.
If a Copilot accelerator is selected, the recommendation should clearly explain why Copilot customization is preferred over Foundry-first development or general agentic development.
Copilot Studio-first scoring card sample

Add a new scoring card table sample specific to Copilot Studio-first solutions in accelerators-scoring-card.md or a related Implementation Recommender reference.
The Copilot Studio-first scoring sample should use Copilot-relevant dimensions instead of reusing Foundry/pro-code dimensions.
Do not include Foundry-oriented primary columns such as hosting/runtime choice or programming language/framework fit as first-class scoring dimensions for Copilot Studio-first accelerators.
Suggested Copilot Studio-first scoring dimensions include:
Copilot scenario fit: simple conversational Q&A, knowledge mining, simple/medium RAG, workflow assistant.
Channel fit: Teams, M365 Copilot, web, custom website, other Copilot-supported channels.
Knowledge source fit: SharePoint, Graph, Dataverse, files, public websites, connector-backed sources, custom APIs.
Actions/tools fit: Copilot Studio actions, connectors, OpenAPI plugins, Power Automate flows, MCP/API-backed tools.
Governance and ALM fit: environments, solution packaging, DLP, tenant policy, maker/developer handoff.
Maker operating model fit: business makers, low-code makers, IT makers, mixed teams.
Hybrid extensibility fit: ability to connect Copilot Studio to Foundry/custom Azure services for advanced retrieval, APIs, or orchestration.
Customization effort and limitations: expected configuration/customization effort and known gaps.
Include a table example with Copilot accelerator rows and an Overall Match calculation based on these Copilot-specific dimensions.
Implementation Recommender workflow instructions

Update Workflow 1 so Copilot-first or hybrid scenarios use the Copilot accelerator catalogue and Copilot-specific scoring considerations.
Ensure the agent records the selected Copilot accelerator URL and rationale in solution-metadata.yaml.
Confirm whether the existing scoring card needs new dimensions or weighting so Copilot-first suitability is not diluted by general pro-code accelerator dimensions.
Expected Outcome
The Implementation Recommender Agent should be able to recognize Copilot-first scenarios, evaluate Copilot Studio accelerator repositories using AGENTS.md-grounded discovery, and recommend the approved Copilot accelerator when it is the best fit.

The recommendation should clearly distinguish between:

Copilot Studio accelerator customization.
General agentic accelerator customization.
Foundry-first agentic development.
Hybrid solutioning where Copilot Studio is the primary experience and Foundry/custom Azure services provide advanced capabilities.

Start with .github/agents/FAF-impl-recommender.agent.md Workflow 1, then read the solution-metadata-template.yaml, accelerators-scoring-card.md, and accelerators-catalogue.md references. Done means Copilot-first and hybrid scenarios have a catalogue, metadata, scoring guidance, and workflow instructions that record the selected accelerator URL and rationale while distinguishing Copilot, general agentic, Foundry-first, and hybrid recommendations.

Written by the indexing model from the issue text.