WIP: stage all uncommitted work — sprints 46-221, graduation headers, specialist fleet, test steps 909-1988

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
This commit is contained in:
Bill
2026-04-22 10:15:48 -06:00
parent 486940cbe4
commit 72ffee68fa
2179 changed files with 82979 additions and 1150 deletions

39
sprint135_plan.md Normal file
View File

@@ -0,0 +1,39 @@
# Sprint 135 Plan: Deterministic MCP Execution Telemetry and Replay Packs
## Context
To debug wiring regressions, execution traces must be deterministic and compact.
Sprint 135 standardizes MCP execution telemetry with replay-ready packets.
---
## Goals
1. Capture deterministic MCP execution telemetry per tool call
2. Normalize argument/result traces for replay and comparison
3. Build compact replay packs for cross-run debugging
4. Support low-context agent diagnosis of MCP regressions
---
## Steps
### Step 1589: MCP execution event schema (10 tests)
### Step 1590: Argument/result canonicalizer model (10 tests)
### Step 1591: Deterministic replay pack composer (10 tests)
### Step 1592: Replay divergence comparator model (10 tests)
### Step 1593: `whetstone_capture_mcp_execution_trace` MCP tool (8 tests)
### Step 1594: `whetstone_export_mcp_replay_pack` MCP tool (8 tests)
### Step 1595: `whetstone_compare_mcp_replays` MCP tool (8 tests)
### Step 1596: Trace quality scorer model (8 tests)
### Step 1597: MCP replay diagnostics artifact (8 tests)
### Step 1598: Sprint 135 integration summary + regression (8 tests)
---
## Architecture Gate
- Trace normalization must produce stable outputs
- Replay comparisons must return structured divergence packets
- Trace capture must preserve privacy and workspace scoping rules
- Max 600 lines per header