Skip to main content
The MCP server provides 2 named prompts that guide multi-step operations. Prompts generate structured instructions that combine tool calls and resource access.

​
analyze_impact

Description: Pre-change workflow — understand an artifact’s current state, assess blast radius, verify chain integrity, and recommend cascade strategy.
​
artifact_id
string
required
The artifact ID to analyze before changing

​
Workflow steps

This prompt generates instructions to:
  1. Get full context: Call sdd_context({ artifact_id }) to review the artifact definition, coverage status, and any gaps
  2. Assess blast radius: Call sdd_impact({ artifact_id, direction: "downstream", maxDepth: 3 }) to understand:
    • Depth 1 = WILL_BREAK (direct dependents)
    • Depth 2 = LIKELY_AFFECTED (indirect dependents)
    • Depth 3 = MAY_NEED_REVIEW (transitive dependents)
  3. Verify chain integrity: Call sdd_trace({ artifact_id }) to check for breaks in the REQ→UC→WF→API→BDD→TASK→COMMIT→CODE→TEST chain
  4. Synthesize recommendation including:
    • Risk level (LOW/MEDIUM/HIGH)
    • Affected pipeline stages
    • Recommended cascade strategy (auto/manual/dry-run)
    • Specific artifacts to review

​
Example usage

When invoked with: 
{
  "artifact_id": "REQ-AUTH-001"
}
The prompt generates a user message with complete instructions for analyzing the impact of changing REQ-AUTH-001.

​
Use cases

  • Before modifying a requirement that’s already implemented
  • Before deprecating an API endpoint
  • Before changing a use case with downstream dependencies
  • As part of a change impact assessment process

​
generate_status_report

Description: Pipeline health report — current state, coverage gaps, priority actions, and overall assessment. Parameters: None

​
Workflow steps

This prompt generates instructions to:
  1. Read pipeline status: Access resource sdd://pipeline/status for current stage, staleness, and recommended next action
  2. Analyze coverage: Call sdd_coverage({}) to get coverage by domain and layer, plus uncovered requirements
  3. Read coverage gaps: Access resource sdd://coverage/gaps for priority gaps ordered by severity
  4. Read graph stats: Access resource sdd://graph/stats for total artifacts, relationships, and orphans
  5. Synthesize the report including:
    • Pipeline health assessment (Healthy/Warning/Critical)
    • Coverage summary per domain
    • Top 5 priority actions
    • Stale stages needing re-run

​
Use cases

  • Daily standup check-ins
  • Sprint planning preparation
  • Traceability audit reports
  • Stakeholder status updates
  • Identifying technical debt hotspots

​
Using prompts in conversations

In MCP-enabled conversations, you can invoke prompts by name: 
"Run the analyze_impact prompt for REQ-AUTH-001"
"Generate a status report using the generate_status_report prompt"
The MCP client will fetch the prompt template, populate any arguments, and inject the generated instructions into the conversation.

​
Custom workflows

You can also manually combine tools and resources to create custom workflows: Example: Find and fix orphaned artifacts
  1. Access sdd://graph/stats to see orphan count
  2. Call sdd_query({ query: "*", limit: 100 }) to get all artifacts
  3. Filter for artifacts with no upstream/downstream relationships
  4. For each orphan, call sdd_context({ artifact_id }) to understand why it’s isolated
  5. Decide whether to delete, merge, or link the orphan
Example: Coverage improvement sprint
  1. Call sdd_coverage({ domain: "Security & Auth" })
  2. Access sdd://coverage/gaps for specific gap details
  3. For each gap, call sdd_trace({ artifact_id }) to see the chain status
  4. Prioritize gaps by:
    • Business domain importance
    • Severity (CRITICAL > HIGH > MEDIUM > LOW)
    • Effort to close (1 missing link > 4 missing links)
  5. Create tasks to close the top 5 gaps

Build docs developers (and LLMs) love

Get started for freeTalk to us