refactor(team): add fast-advance notification and knowledge transfer protocol

- team-worker: add fast_advance message bus log after spawning successor,
  closing coordinator state blind spot during fast-advance
- team-worker: add Knowledge Transfer section with upstream loading,
  downstream publishing, and context_accumulator conventions
- role-spec-template: add Knowledge Transfer Protocol with Transfer
  Channels table and shared-memory.json namespaced write convention
- monitor.md (v2+v5): add fast-advance reconciliation step reading
  fast_advance messages, add State Sync section for coordinator wake
- lifecycle-v5 SKILL.md: update cadence diagram with fast_advance log
This commit is contained in:
catlog22
2026-02-28 22:53:56 +08:00
parent 67b2129f3c
commit e42597b1bc
4 changed files with 101 additions and 2 deletions

View File

@@ -404,6 +404,48 @@ Write discoveries to corresponding wisdom files:
---
## Knowledge Transfer
### Upstream Context Loading (Phase 2)
When executing Phase 2 of a role-spec, the worker MUST load available cross-role context:
| Source | Path | Load Method |
|--------|------|-------------|
| Upstream artifacts | `<session>/artifacts/*.md` | Read files listed in task description or dependency chain |
| Shared memory | `<session>/shared-memory.json` | Read and parse JSON |
| Wisdom | `<session>/wisdom/*.md` | Read all wisdom files |
| Exploration cache | `<session>/explorations/cache-index.json` | Check before new explorations |
### Downstream Context Publishing (Phase 4)
After Phase 4 verification, the worker MUST publish its contributions:
1. **Artifact**: Write deliverable to `<session>/artifacts/<prefix>-<task-id>-<name>.md`
2. **shared-memory.json**: Read-merge-write under role namespace
```json
{ "<role>": { "key_findings": [...], "decisions": [...], "files_modified": [...] } }
```
3. **Wisdom**: Append new patterns to `learnings.md`, decisions to `decisions.md`, issues to `issues.md`
### Inner Loop Context Accumulator
For `inner_loop: true` roles, `context_accumulator` is maintained in-memory:
```
context_accumulator.append({
task: "<task-id>",
artifact: "<output-path>",
key_decisions: [...],
summary: "<brief>",
files_modified: [...]
})
```
Pass the full accumulator to each subsequent task's Phase 3 subagent as `## Prior Context`.
---
## Message Bus Protocol
Always use `mcp__ccw-tools__team_msg` for logging. Parameters: