tylergarlick/abraxas/ethos

Ethos Credibility MCP

The Ethos Credibility service is the Sovereign system's trust arbiter. it provides a deterministic weight-based system for evaluating the reliability of information sources and resolving contradictory claims.

Identity

Ethos is the Sovereign Truth-Weighting Engine. It ensures that the system does not treat all sources as equal, instead applying a tiered credibility model to prevent high-confidence fabrications from overriding grounded evidence.

Commands

/ethos_score

• Behavior: Evaluates a specific source against the Ethos Registry to determine its credibility tier.
• Input: source (string)
• Sensing: Cross-references the input source against known domains, organizations, and publication types in the registry.
• Output: A detailed report including the Tier (T1-T5), trust weight, and status description.

/ethos_resolve

• Behavior: Resolves conflicts between two contradicting sources by comparing their respective credibility weights.
• Input: source_a (string), source_b (string)
• Logic: Computes the weight delta between sources. The source with the higher weight is designated as the winner.
• Output: A resolution report showing the weights of both sources and the definitive winner.

Operational Logic

• The Ethos Registry: Credibility is governed by ethos-registry.json, which maps sources to specific tiers.
• Weighting System:
• T1 (Highest): Primary, peer-reviewed, or sovereign-verified sources.
• T5 (Lowest): Unverified, speculative, or known unreliable sources.
• Weight Delta: Significant deltas between sources trigger a "Definitive Resolution," while small deltas may require an Soter trigger for further verification.

Implementation Details

• Architecture: Python FastMCP server utilizing a logic layer that reads the Ethos Registry.
• Sovereign Integration: Used heavily during the "Conflict Resolution" phase of the reasoning pipeline to determine which claim to accept as the "Best Claim."

MCP tools

Available MCP tools: ethos_score, ethos_resolve.