Agenda Intel MD

v1.2.0 Feature

This release adds 2 notable features for engineering teams evaluating rollout.

Published 16d MCP Developer Tools

View tool

✓ No known CVEs patched

Read the diff → Tool health → What is this tool? →

✓ No known CVEs patched in this version

Topics

ai-agents claim-grounding cli deterministic evidence-packet evidence-validation

+6 more

human-review json-schema llm-evaluation mcp-server python source-grounding

Summary

AI summary

Added deterministic lexical claim-to-corpus checks and bounded Claim Verdicts over caller‑supplied evidence.

Changes in this release

Type	Severity	Summary	CVE
Feature
Feature	Medium	Adds `grounded_check` CLI command for lexical claim-to-corpus consistency checks. Adds `grounded_check` CLI command for lexical claim-to-corpus consistency checks. Source: llm_adapter@2026-07-15 Confidence: high	—
Feature	Medium	Adds `verify_claims` CLI command for bounded Claim Verdicts over supplied evidence. Adds `verify_claims` CLI command for bounded Claim Verdicts over supplied evidence. Source: llm_adapter@2026-07-15 Confidence: high	—
Feature	Medium	Adds MCP manifest entries, examples, contract tests, and references ADR 0023. Adds MCP manifest entries, examples, contract tests, and references ADR 0023. Source: llm_adapter@2026-07-15 Confidence: high	—
Feature	Medium	Adds freshness, authority, source‑group, conflict, jurisdiction, and identifier checks to verification logic. Adds freshness, authority, source‑group, conflict, jurisdiction, and identifier checks to verification logic. Source: llm_adapter@2026-07-15 Confidence: high	—
Feature	Medium	Adds new verdicts: `verified`, `contradicted`, `partially_supported`, `unresolved`, and `not_verifiable`. Adds new verdicts: `verified`, `contradicted`, `partially_supported`, `unresolved`, and `not_verifiable`. Source: llm_adapter@2026-07-15 Confidence: high	—
Feature	Medium	Adds request and response schemas for grounded_check and verify_claims capabilities. Adds request and response schemas for grounded_check and verify_claims capabilities. Source: llm_adapter@2026-07-15 Confidence: low	—

Full changelog

v1.2.0 — Grounding and bounded factual Claim Verdict

v1.2.0 separates two verification questions that must not share one score:
whether a claim is represented in a supplied corpus, and whether a supplied
evidence set meets an explicit factual-verification threshold as of a declared
date.

What changed

Added grounded_check and the grounded-check CLI command for deterministic
lexical claim-to-corpus consistency checks.
Added verify_claims and the verify-claims CLI command for bounded Claim
Verdicts over caller-supplied evidence records.
Added request and response schemas for both capabilities, MCP manifest
entries, examples, contract tests, and ADR 0023.
Added checks for source freshness, authority, independent source groups,
conflicts, jurisdiction, and exact subject identifiers.
Added verified, contradicted, partially_supported, unresolved, and
not_verifiable verdicts without changing existing evidence-audit or
grounding vocabulary.

Boundaries

No source discovery or live retrieval. No fuzzy entity resolution. A
verified verdict means the supplied evidence meets the caller's declared
threshold as of the declared date. It is not absolute truth or a legal,
compliance, sanctions, financial, investment, insurance, or trading
determination. Human review remains required before any commercial action.

Install

pip install agenda-intelligence-md==1.2.0

View diff on GitHub

Weekly OSS security release digest.

The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.

No spam, unsubscribe anytime.

Share this release

Share on X Share on Bluesky

Track Agenda Intel MD

Get notified when new releases ship.

About Agenda Intel MD

All releases →

Related context

Related tools

Earlier breaking changes

v0.8.0 MCP tool count increased from 11 to 16, adding five new tools.