Knowledge Base

Knowledge Base

The Knowledge Base is a cross-project repository of lessons learned, recurring patterns, and reusable insights extracted from your organization's completed projects. It prevents institutional knowledge from evaporating when a project closes.

Why the Knowledge Base Matters

Every SpecGraph project generates valuable information beyond just the final PRD:

• Which requirement trade-offs worked well and which caused problems in development.
• Technical decisions that turned out to be the right call (or the wrong one).
• Department concerns that were dismissed but proved valid in hindsight.
• Conflict resolutions that could inform similar decisions in future projects.
• Implementation challenges that tripped up the team.

Without deliberate capture, all of this context lives in people's heads and disappears when team members move on or projects are archived. The Knowledge Base makes this learning explicit and searchable.

How Knowledge Is Captured

AI Extraction

When a project reaches the Closed phase, you can trigger AI Knowledge Extraction. The AI reviews the entire project history — documents, PRD, wishes, conflicts, resolutions, implementation notes, and post-build feedback — and extracts key insights as structured knowledge entries.

Each extracted entry includes:

• A title summarizing the insight.
• A description explaining the context, what happened, and what was learned.
• A category (e.g., Architecture Decision, Process Lesson, Technical Pattern, Risk Identified).
• The source project it came from.
• Tags for searchability.

Manual Entry

Team members can also add knowledge entries manually at any time. This is useful for:

• Capturing a lesson in the moment, without waiting for project close.
• Adding organization-wide knowledge that isn't tied to a specific project.
• Documenting decisions made outside SpecGraph that the team wants to preserve.

Navigating the Knowledge Base

The Knowledge Base page (accessible from the main navigation) shows all knowledge entries across all projects in your organization. You can:

• Search entries by keyword.
• Filter by category, source project, or tag.
• Sort by date, relevance, or project.

Each entry links back to the source project, so you can always drill down into the full context if the summary isn't enough.

Using Knowledge in New Projects

The Knowledge Base is most valuable when starting a new project. Before creating a project and generating a PRD, browse relevant knowledge entries from past projects:

• Have similar projects encountered specific risks or blockers? You can add those as concerns or constraints in your new project's documents upfront.
• Did a past project find a particularly effective technical pattern? Reference it in your architecture notes.
• Did conflict resolution in a past project produce a reusable policy (e.g., "We always prioritize security requirements over performance preferences for customer-facing tools")? Document it in your organization's guardrails.

Post-Build Feedback

Closely related to the Knowledge Base is the Post-Build Feedback feature, available on projects in the Deployed or Closed phase. This lets team members submit structured feedback about how the final product compares to the spec:

• Did the development match what the PRD described?
• Were there significant deviations, and why?
• What would you do differently in the requirements process next time?

Post-build feedback is included in the AI knowledge extraction and forms some of the richest entries in the Knowledge Base.