Coder/deepseek-harness
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

refactor(agent-loop): drop the steering/message session event

Browse Source 
Steer, inject, and followup now land as durable user/message events on the
session surface; the steering/message event type and its ConversationNode
kind are removed from the client projection. Update tests, docs, generated
catalogs, and agent notes to match, and align the steering e2e fixture and
prompt inventory assertions with the durable user/message landing.
This commit is contained in:
_Kerman
2026-08-03 16:04:26 +08:00
parent f4a2e0d10a
commit cb2f01f48c
86 changed files with 243 additions and 458 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
packages/context/session-reference/README.md
View File

@@ -12,9 +12,9 @@ English | [中文](README.zh.md)
## Snapshot semantics
Preparation calls `ctx.sessionQuery.readSurface()` once per distinct source and never rereads it after enqueue. It projects only direct-user `user/message`, direct-user `steering/message`, assistant text, and `user/message` checkpoints carrying the canonical `dsh-compact` source marker from the folded current surface. For a source prompt that already contains baked prefix context, projection reads only its model-hidden display content, preventing recursive snapshot propagation. Shadowed pre-compaction events, tools, reasoning, context, plugin-generated user messages other than marked compact checkpoints, and unfinished assistant chunks are excluded. A compacted source therefore contributes its latest checkpoint plus retained later conversation, not restored shadowed text.
Preparation calls `ctx.sessionQuery.readSurface()` once per distinct source and never rereads it after enqueue. It projects only direct-user `user/message`, assistant text, and `user/message` checkpoints carrying the canonical `dsh-compact` source marker from the folded current surface. For a source prompt that already contains baked prefix context, projection reads only its model-hidden display content, preventing recursive snapshot propagation. Shadowed pre-compaction events, tools, reasoning, context, plugin-generated user messages other than marked compact checkpoints, and unfinished assistant chunks are excluded. A compacted source therefore contributes its latest checkpoint plus retained later conversation, not restored shadowed text.
The context source is `{ kind: 'session-reference', version: 1, references }`; each reference records its source id and label, capture seq, compact presence, retained/omitted message counts, omitted UTF-8 bytes, and truncation state. When the agent is idle, the standard TUI installs a one-shot `agent/pre-step` wrapper that adds the snapshot only to an `enter` decision containing the claimed direct prompt. While the agent is running, it calls `inject()` immediately before `steer()`, placing both messages in the next-step inbox for the same later claim. The target log therefore records a sourced context `user/message` followed by the readable direct `user/message` or `steering/message`. Later source mutation, compaction, or deletion cannot change target replay.
The context source is `{ kind: 'session-reference', version: 1, references }`; each reference records its source id and label, capture seq, compact presence, retained/omitted message counts, omitted UTF-8 bytes, and truncation state. When the agent is idle, the standard TUI installs a one-shot `agent/pre-step` wrapper that adds the snapshot only to an `enter` decision containing the claimed direct prompt. While the agent is running, it calls `inject()` immediately before `steer()`, placing both messages in the next-step inbox for the same later claim. The target log therefore records a sourced context `user/message` followed by the readable direct `user/message`. Later source mutation, compaction, or deletion cannot change target replay.
## Configuration