| name | woos-agent-decision |
| description | Resolve multi-agent review conflicts with explicit authority matrix, evidence weighting, and escalation criteria. |
| version | 1.1.0 |
| author | Hermes Profile |
| license | MIT |
Woos Agent Decision
Purpose
Provide deterministic conflict resolution when reviewers disagree on gate outcomes.
Decision Policy
- Domain authority first:
- Security issues:
woos-security-reviewer authoritative.
- Architecture issues:
woos-architect authoritative.
- Planning/sequencing issues:
woos-product-planner (with mode: planning) authoritative.
- Evidence-backed conclusion over opinion-only conclusion.
- Lower-risk path over speed-optimized path when evidence is similar.
Escalation Triggers
- Same blocking disagreement unresolved after
reconciliation_attempt_max.
- Conflict spans multiple domains with no clear authority winner.
- High-risk decision without sufficient evidence.
Counter Contract
review_round_max: 2
reconciliation_attempt_max: 1 (within each round)
- Counter precedence:
- if
reconciliation_attempt_max exceeded -> escalate immediately
- else if
review_round_max exceeded -> escalate
Output Contract
status: PASS | REQUEST_CHANGES | BLOCKED
resolution: agent_a | agent_b | hybrid | escalated_to_human
reason
evidence_used
requires_human_handoff: boolean