Agenda Intel MD

v1.1.0 Feature

This release adds 3 notable features for engineering teams evaluating rollout.

Published 1mo 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

Affected surfaces

breaking_upgrade

Summary

AI summary

A2A wire contract migrated to protocol v1.0, Cloudflare Worker parity with Python features added, and per-task cost accounting introduced.

Full changelog

A2A wire contract upgraded to A2A protocol v1.0, deployed Worker brought to parity with the Python service, and per-task cost accounting added. Schema freeze (ADR 0003) and the request/memo contract are untouched — minor bump.

Highlights

feat(a2a): migrate the A2A wire contract to A2A protocol v1.0 (ADR 0017). Member-discriminated parts with mediaType, TASK_STATE_* status, supportedInterfaces[] on the Python card. Inbound parsing unchanged (v0.3 + v1.0 both accepted). 52 Python + 62 Worker tests green.
feat(worker): Cloudflare Worker JS parity for Middle Corridor jurisdiction flags, two-layer exposure decomposition, and the OFAC-grounded vessel due-diligence checklist — features that shipped in Python but weren't live on the A2A surface.
feat(worker): per-task cost accounting + budget thresholds in /stats (only paid live-retrieval upstreams billed; conservative on degraded/disabled).
feat(middle_corridor_deal_risk): re-export / circumvention-watch jurisdiction flag (separate from high-risk list, no double-flag) + exposure_layers decomposition (domestic-legal vs foreign-sanctions) + vessel DSP verification checklist. All additive, non-breaking; presence-flagging routed to human review, never a determination.
fix(mcp): constrain category tool args to the packaged enum (computed, never drifts) + correct a stale energy-markets → energy slug.