Documentation Index

Fetch the complete documentation index at: https://grantmaster.dev/llms.txt

Use this file to discover all available pages before exploring further.

​

Glossary of Terms

​

🏢 Platform & Tenancy

• Platform: The entire multiservice ecosystem (GrantMaster).
• Tenant (Organization): A distinct customer entity (e.g., an NGO). All data is strictly isolated at the Tenant level.
• SuperAdmin: A GrantMaster employee with global permissions to manage all tenants, billing, and system configurations via the SuperAdmin dashboard.
• Admin: A user within a Tenant organization who has permissions to manage users and settings for their specific organization.

​

📦 Extensions

• Extension: A discrete set of features (e.g., “Grant Discovery,” “Project Management”) that can be individually enabled/disabled.
• Catalog: The master list of all available Extensions managed by SuperAdmins.
• Installation: The specific instance of an Extension being active for a Tenant. This tracks trial periods, subscription status, and extension-specific configurations.
• Extensions Page: The user-facing interface where Admins can browse and “buy” (install) Extensions.

​

🎯 Grant Lifecycle

• Grant Listing: A raw opportunity found by a Scraper.
• Grant Discovery: The process/engine that matches Grant Listings to an organization’s Mission Profile.
• Pursuit: An active attempt by a Tenant to win a specific Grant. A Pursuit moves through stages (Drafting, Submitted, Won/Lost).
• Pipeline: The visual kanban-style board where Pursuits are tracked.
• Project: A successful (Won) Pursuit that has transitioned into the implementation phase.
• Effort: A discrete block of time or resource allocated by a staff member toward a Pursuit or Project.

​

🤖 Intelligence & Data

• Mission Profile: The structured data describing a Tenant’s goals, impact areas, and target demographics, used to train the AI matching engine.
• Scraper: An automated bot that monitors 3rd-party portals (e.g., Grants.gov) to ingest new Grant Listings.
• Normalization: The process of using AI to convert unstructured scraper data into the unified GrantMaster schema.
• Impact Taxonomy: A hierarchical classification system (e.g., Health > Rural Support) used to standardize grant matching across different regions.

​

⚙️ Technical Architecture

• EventBus: The internal communication system that handles decoupling between services (e.g., when a grant is won, the EventBus notifies the billing service).
• BaseService: The standard parent class for all backend services, providing built-in logging, error handling, and event emitting.
• Listener: A frontend or backend process that reacts to real-time database changes or EventBus messages.
• Impersonation: A SuperAdmin feature allowing them to view the platform as a specific user for troubleshooting purposes.
• tRPC: Type-safe RPC layer used alongside Firebase Cloud Functions for the API layer between frontend and backend.

​

🤖 AI & Agents

• Agent: An autonomous AI entity that executes multi-step tasks (e.g., compliance review, report generation) within bounded credit budgets and RBAC-scoped permissions. See src/features/agents/.
• Agent Run: A single execution of an Agent, tracking steps, credit consumption, token usage, and status via a state machine (queued → running → paused → completed/failed/cancelled).
• Agent Step: A discrete action within an Agent Run — a single tool invocation with tracked I/O, cost, and duration.
• Agent Escalation: A human-in-the-loop pattern where an Agent pauses execution (awaiting_human status) and surfaces a decision to a user with the required permission.
• Agent Tool: A registered capability that Agents can invoke (e.g., “query expenses”, “generate report”). Each tool declares required permissions and is executed within the triggering user’s RBAC scope.
• Genkit: Google’s AI orchestration framework used to integrate Gemini models for generation, extraction, and RAG workflows. See src/features/ai/services/geminiService.ts.
• RAG (Retrieval-Augmented Generation): A pattern where relevant documents are retrieved from Firestore/vector storage and injected into the LLM prompt for grounded answers. See ragService.ts.

​

💳 Billing & Credits

• Credit: The unit of metered consumption for AI agent operations. Credits are reserved before an agent run and consumed per step.
• Credit Reservation: A hold placed on an organization’s credit balance before an agent run begins, preventing concurrent overdraw via Firestore transactions.
• Entitlement: A feature flag or quota gated by the tenant’s subscription tier. Defined in src/config/entitlements.ts.
• Credit Pack: A one-time purchasable top-up of credits, separate from the recurring subscription.

​

📊 Impact & M&E

• M&E Indicator (Monitoring & Evaluation Indicator): A quantifiable metric (e.g., “Number of beneficiaries served”) used to track project outcomes against grant objectives. See src/features/impact/.
• Impact Framework: A structured set of M&E indicators, targets, and reporting timelines attached to a grant or project.
• Impact Taxonomy: See Intelligence & Data section — the hierarchical classification system for standardizing grant matching.

​

🌐 Portal & Integrations

• Portal Token: A time-limited, scope-restricted token granting external stakeholders (auditors, grantors) read access to specific organizational data. Stored in portalTokens collection.
• Grantor Portal: An external-facing web interface where grantors or partners can review submissions, progress reports, and compliance data without a full platform account.