1 patterns for this goal
A multi-agent handoff loses the upstream agent’s confidence signal because the handoff schema is designed to carry only a final value and a status flag, not the confidence level or methodology the upstream agent expressed in its free-text reasoning — so a downstream agent consumes a value the upstream agent explicitly flagged as low-confidence or in need of specialist review with full, unwarranted confidence. The gap is structural: the confidence exists in the upstream agent’s transcript, but the schema between agents has no field for it, so the information is invisible to any system that only reads the structured handoff record.
{value, status} handoff record, per the pattern’s stated symptom.The single mechanism handoff reliability covers is narrow handoff schemas that carry a final value and status but no confidence, methodology, or provenance field, so a downstream agent’s decision logic — built to operate on the fixed schema — has no structured path to the upstream agent’s own uncertainty. See Handoff Schema Loses Upstream Confidence Signal for four worked examples: a bond maturity-date reconciliation, a diagnosis confidence handoff, a contract-ambiguity handoff, and a demand-forecast handoff.
{value, status} shape rather than something that can express degrees of confidence or methodologyThe pattern’s own mitigations mirror the broader coordination toolkit: handoff schema validation with explicit required fields (including confidence/methodology, not just value/status), consensus checkpoints that can catch a downstream agent proceeding on unverified state, and saga-style compensating actions if a low-confidence value turns out to be wrong. The core fix is narrower than the general tooling suggests, though — it is specifically about widening the handoff schema itself to have a place for confidence and methodology to live as structured data, not just free text the downstream agent never reads.
In practice, downstream agents are built to consume a structured handoff record for reliability and parseability, not to re-parse another agent’s free-text chain of reasoning. That design choice is reasonable for consistency, but it means any signal the upstream agent expressed only in prose — like a confidence caveat — is architecturally invisible to the downstream agent’s decision logic.
They’re related but not identical. Coordination’s Task Handoff Failure describes an agent passing incomplete or incorrect state generally. Handoff reliability’s pattern is more specific: the state passed is technically complete and correct (the value itself is right), but the confidence/methodology context needed to use that value safely is what’s missing.
Add explicit confidence and methodology fields to the handoff schema itself — not as an afterthought free-text field, but as structured data the downstream agent’s decision logic actually branches on (for example, routing low-confidence values to human review instead of auto-processing low-confidence values with full confidence).
| Pattern | Mechanism |
|---|---|
| Handoff Schema Loses Upstream Confidence Signal | Handoff schema carries only final value and status, so upstream confidence/methodology caveats never reach the downstream agent |
Total: 1 pattern
An upstream agent flags low confidence or ambiguity in its free-text reasoning, but the structured handoff schema passed to the downstream agent -- a common MCP/tool-call handoff pattern -- carries only the final value and status. The confidence/provenance signal is invisible to the downstream agent, which uses the value with full confidence.