Belief State Runtime

belief-state-runtime.skill

Evaluates the trustworthiness of a claim based on evidence.

When to use

• User asks "is this true?" or "can I trust this?"
• Agent collects information from multiple sources and needs to assess reliability
• Detecting contradictions between evidence pieces
• Quantifying confidence in a conclusion

⚡ AUTO WORKFLOW (Default - Always Run This First)

IMPORTANT: When this skill is triggered, you MUST follow this workflow automatically:

from assess import get_assessment_prompt, assess_claim_with_response

# Step 1: SEARCH FOR EVIDENCE (MANDATORY)
# Use your search tool (online-search, multi-search-engine, etc.) to find
# real, verifiable evidence about the claim. Search for:
#   - Official data and statistics
#   - Academic research and studies
#   - Credible news reports
#   - Expert opinions
#
# Example search queries:
#   - "[claim topic] facts evidence research"
#   - "[claim topic] statistics studies 2024"
#   - "专家观点 [claim topic]"

# Step 2: Get the assessment prompt
prompt = get_assessment_prompt(claim="The claim", evidence="Search results + user's evidence")

# Step 3: YOU (the AI) answer the prompt with 6 boolean judgments
# Based on your search results and analysis:
#   - direct_support: Does the evidence support the claim?
#   - new_info: Does it provide new information?
#   - logical_consistent: Is it logically consistent?
#   - direct_refute: Does it explicitly refute the claim?
#   - limitation: Are there limitations or exceptions?
#   - error_outdated: Is the claim outdated or wrong?
#
# Your answer format:
#   {"direct_support": true/false, "new_info": true/false, "logical_consistent": true/false, "direct_refute": true/false, "limitation": true/false, "error_outdated": true/false}

# Step 4: Get final result
result = assess_claim_with_response(
    claim="The claim",
    evidence="Search results + user's evidence",
    llm_response='{"direct_support": true, ...}'  # YOUR judgment
)

# Step 5: Present the result to the user
# Include: state, confidence, summary, and your interpretation

Workflow Summary (Quick Reference)

How it works

1. Search for evidence (MANDATORY): Use search tools to find real, verifiable evidence about the claim. This includes official data, academic research, credible news, and expert opinions.

2. Rule layer (Python): Extracts source reliability, evidence density, temporal freshness from the evidence text using URL matching, keyword detection, and year extraction.

3. LLM layer (YOU): The AI agent answers 6 boolean questions about the evidence:

   • Does it directly support the claim?
   • Does it provide new information?
   • Is it logically consistent?
   • Does it explicitly refute the claim?
   • Does it point out limitations?
   • Does it reveal errors or outdated info?

4. Aggregation (Python): Combines rule signals and your judgments into a calibrated confidence score and state.

Output

{
  "state": "VERIFIED",
  "confidence": 0.83,
  "confidence_range": [0.68, 0.98],
  "features": {"direct_support": true, "new_info": true, ...},
  "summary": "Evidence strongly supports the claim"
}

States:

• VERIFIED (confidence >= 0.65): Agent can cite this information
• CONTESTED (0.25 < confidence < 0.65): Agent should note "there is disagreement"
• UNCERTAIN (confidence <= 0.25): Agent should say "need more information"

Incremental updates

When evidence arrives in stages, the engine updates beliefs incrementally:

prompt = get_assessment_prompt(claim, evidence="stage 1")
# AI answers...
result1 = assess_claim_with_response(claim, evidence="stage 1", llm_response=ai_answer)

prompt = get_assessment_prompt(claim, evidence="stage 1 + stage 2")
# AI answers...
result2 = assess_claim_with_response(claim, evidence="stage 1 + stage 2",
                                     llm_response=ai_answer,
                                     previous_confidence=result1["confidence"])

Parameters

Legacy API

For backward compatibility, assess_claim() with llm_func callback still works.

from assess import assess_claim
result = assess_claim(claim="...", evidence="...", llm_func=my_llm)