41 lines
1.4 KiB
Markdown
41 lines
1.4 KiB
Markdown
# Sprint 119 Plan: Deterministic Cross-Node Failure Triage and Prioritization
|
|
|
|
## Context
|
|
|
|
Sprint 118 standardizes distributed failure evidence. Sprint 119 adds
|
|
root-cause-first triage and deterministic prioritization so low-context agents can
|
|
select the highest-leverage fix path without manual interpretation.
|
|
|
|
---
|
|
|
|
## Goals
|
|
|
|
1. Cluster distributed failures into root-cause-first groups
|
|
2. Rank triage actions using deterministic impact and confidence scoring
|
|
3. Add MCP tools to compute and retrieve triage queues
|
|
4. Preserve machine-readable escalation and stop reasons
|
|
|
|
---
|
|
|
|
## Steps
|
|
|
|
### Step 1429: Cross-node failure cluster model (12 tests)
|
|
### Step 1430: Root-cause ranking policy model (10 tests)
|
|
### Step 1431: Blast-radius estimator for distributed failures (10 tests)
|
|
### Step 1432: Deterministic triage queue planner model (10 tests)
|
|
### Step 1433: `whetstone_cluster_distributed_failures` MCP tool (8 tests)
|
|
### Step 1434: `whetstone_rank_failure_triage_actions` MCP tool (8 tests)
|
|
### Step 1435: `whetstone_get_distributed_triage_queue` MCP tool (8 tests)
|
|
### Step 1436: Escalation reason classifier model (8 tests)
|
|
### Step 1437: Triage packet bundle model (8 tests)
|
|
### Step 1438: Sprint 119 integration summary + regression (8 tests)
|
|
|
|
---
|
|
|
|
## Architecture Gate
|
|
|
|
- Ranking rules must be deterministic and auditable
|
|
- Triage queue priorities must be reproducible
|
|
- Escalation reasons must be explicit and structured
|
|
- Max 600 lines per header
|