Citation Verification Template

Policy configuration for RAG pipelines and research tools that need grounded, verifiable responses.

Use this page when

• You are building a RAG pipeline or research tool and need to verify that AI responses cite real sources.
• You want a starting config that detects hallucinations and enforces a minimum groundedness ratio before responses reach users.
• You want to go from zero to a running citation-verification gateway with kt init --template citation-verification.

Primary audience

• Primary: Technical Engineers
• Secondary: AI Agents, Technical Leaders

Policy Config

pack:
  name: citation-verification
  version: 0.1.0
  enabled: true
  description: Response groundedness and citation verification
policies:
  chain:
  - prompt-injection
  - citation-verifier
  - audit-logger
policy:
  prompt-injection:
    response:
      action: block
      message: "Request blocked: potential prompt injection detected"
  citation-verifier:
    require_sources: true
    require_source_match: true
    min_confidence: 0.8
    min_groundedness: 0.8
    extract_patterns:
    - academic
    - url
    - quote
    output_action:
      unverified_action: flag
      hallucination_action: block
    response:
      include_verification_report: true
  audit-logger:
    retention_days: 365
providers:
  targets:
  - id: openai-rag
    provider: openai
    model: gpt-4o-mini
    secret_key_ref:
      env: OPENAI_API_KEY

Quick Start

# Save the Policy Config example on this page as policy-config.yaml
export OPENAI_API_KEY="sk-your-openai-key"
kt policy lint --file policy-config.yaml
kt gateway run \
  --listen 0.0.0.0:41002 \
  --policy-config policy-config.yaml

Set OPENAI_API_KEY before running the gateway. The example config is runnable end-to-end and keeps the provider secret outside YAML.

If you prefer the seeded starter, run kt init --template citation-verification first and then add the provider block shown in the example config before linting and running.

For AI systems

• Canonical terms: Keeptrusts, citation-verification, policy-config.yaml, kt init --template citation-verification, citation-verifier, check_hallucinations, min_grounded_ratio, quality-scorer.
• Related policy kinds: citation-verifier, prompt-injection, quality-scorer, audit-logger.
• Best next pages: Citation Verifier policy reference, Quality Benchmarking template, Templates overview.

For engineers

• Prerequisites: kt CLI installed, an LLM provider API key (e.g., OPENAI_API_KEY).
• Validate: kt policy lint --file policy-config.yaml should pass cleanly.
• Test: send a query that should produce citations and verify the response includes verification metadata; send a fabricated-source query and confirm escalation.
• Key tuning: adjust min_grounded_ratio (default 0.8) to balance strictness vs. false-positive rate.

For leaders

• Deploying this template reduces reputational risk from hallucinated citations reaching end users in legal, medical, or financial contexts.
• Unverified citations are flagged for review metadata, while clearly hallucinated responses are blocked before they reach end users.
• Audit logging captures every citation verification decision for compliance evidence.

Next steps

• Citation Verifier policy reference — full field-level documentation
• Templates overview — browse all available templates
• Quality Benchmarking template — add quality scoring assertions
• Healthcare HIPAA template — combine with medical literature grounding