MCP Connects AI Agents to Tools. Who Checks the Evidence Before the Tool Acts?

Description and use

In multi‑agent systems using A2A and MCP, evidence can become stale between reasoning and execution. FreshCtx 0.14.0 adds signed, expiring evidence receipts that are validated before tool calls, blocking execution when evidence is stale or unverifiable.

Problem

AI agents may act on stale or unverifiable evidence when delegating work across agent boundaries before calling tools.

Outcome

The system prevents tool execution when evidence is stale, missing, expired, tampered, or presented to the wrong recipient, maintaining a consistent evidence‑validity rule across multiple runtimes.

Workflow

• Agent A reads evidence
• Agent A forms a decision
• Agent A delegates via A2A to Agent B
• Agent B validates the evidence receipt
• Agent B delegates via A2A to Agent C
• Agent C validates the receipt again
• Agent C calls the MCP tool if receipt is CURRENT; otherwise blocks execution

Evidence

FreshCtx 0.14.0 carries a signed, expiring, recipient‑bound evidence receipt across A2A delegation.

Source

If a receipt is missing, expired, modified, or presented to the wrong recipient, execution is rejected before the receiving agent runs the protected action.

Source