ZiyaZhang
e8f5c0b51b
Minimal Electron shell over the DSH JSON-RPC runtime — a first-look at what a ChatGPT.app-style host on top of the DeepSeek Harness looks like, with the harness's normally-invisible internals (trace timeline, context surface, subagent tree, compaction, plugin registry, rubrics) brought forward as first-class UI surfaces so plugin authors and researchers can see what the agent is actually doing. Runs against three keyless-to-live profiles (stdio-echo works on master out of the box; daemon-echo / daemon-vibe-echo activate once the daemon-demo lands; stdio-deepseek and daemon-vibe hit the real DeepSeek API when you supply a key). HARNESS_DEV auto-resolves to the in-repo runtime when this shell ships under examples/desktop/, so a fresh clone launches without config; env DSH_DEV_ROOT overrides for custom layouts, and a sibling deepseek-harness-dev/ checkout is the original dev workflow. Cold-clone gate (P0 fixes for first-time-clone usability): - HARNESS_DEV: 3-candidate resolver (env → walk-up in-repo marker → sibling), unit-tested via mock fs so ordering is locked without needing either real layout on disk. - config yml leaves rewritten at assemble time so the sibling-clone paths (../../deepseek-harness-dev/examples/echo-agent/…) become the in-repo paths (../../echo-agent/…) in the released tree — source yml stays usable for local dev, released tree ships a working shape. - pnpm-workspace.yaml allowBuilds.electron = true (was placeholder). - missing-key card in stdio-deepseek offers a one-click switch to stdio-echo (the keyless profile that works on master) rather than daemon-echo (blocked on the not-yet-shipped daemon-demo). - assemble-oss-release.sh rewrites the source-side breadcrumb name 'dsh-desktop-demo' → 'dsh-desktop' for the released package.json. FOUC guard on the onboarding gate (41fc5df carried) keeps the first-launch splash from flashing before the runtime probe finishes. Test suite (1634 tests in source, 3990 in the runtime repo) covers resolver ordering, renderer classifiers, trace timeline shape, compaction diff rendering, rubric parity, and the missing-key onboarding paths.
75 lines
3.8 KiB
JSON
75 lines
3.8 KiB
JSON
[
|
|
{
|
|
"_mock": true,
|
|
"_mockReason": "#162 rec 32 (compact before/after diff tab): compact/summary event carrying 12 shadowed seqs (representing ~3.8k tokens of prior turn content), a ContentBlock[] summary block (~380 tokens), and full strategy metadata so the ratio bar can render (10.0x compression)."
|
|
},
|
|
{
|
|
"type": "assistant/message",
|
|
"time": 1721403000050,
|
|
"seq": 801,
|
|
"data": {
|
|
"turn": 12,
|
|
"step": 4,
|
|
"content": [
|
|
{ "type": "text", "text": "Refactor complete — will summarise the session so far." }
|
|
],
|
|
"usage": { "inputTokens": 4200, "outputTokens": 210 }
|
|
}
|
|
},
|
|
{
|
|
"type": "compact/start",
|
|
"time": 1721403000100,
|
|
"seq": 802,
|
|
"data": {
|
|
"reason": { "kind": "user" },
|
|
"model": "deepseek-chat",
|
|
"maxTokens": 800,
|
|
"shadowedRange": { "start": 121, "end": 132 },
|
|
"trigger": { "kind": "user" }
|
|
}
|
|
},
|
|
{
|
|
"type": "compact/summary",
|
|
"time": 1721403000450,
|
|
"seq": 803,
|
|
"data": {
|
|
"model": "deepseek-chat",
|
|
"maxTokens": 800,
|
|
"shadowedRange": { "start": 121, "end": 132 },
|
|
"shadowedSeqs": [121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132],
|
|
"shadowedTokenCount": 3812,
|
|
"summary": [
|
|
{ "type": "text", "text": "The session refactored src/lib/app.ts to split App into AppShell (thin) and AppCore (business logic), moved two utility functions to src/lib/util/*, and ran the unit suite (all 96 tests passed). Key decisions: (a) AppShell owns cli/env parsing only, (b) AppCore is constructor-injectable to make tests hermetic, (c) new util modules keep the previous public API by re-export. Follow-ups queued: rename AppShell.start → AppShell.run for symmetry with the interface, and cover the new AppCore.render path with an integration test." }
|
|
],
|
|
"trigger": { "kind": "user" }
|
|
}
|
|
},
|
|
{
|
|
"type": "compact/end",
|
|
"time": 1721403000500,
|
|
"seq": 804,
|
|
"data": { "model": "deepseek-chat", "shadowedRange": { "start": 121, "end": 132 } }
|
|
},
|
|
|
|
{
|
|
"_mock": true,
|
|
"_note": "The shadowedSeqs above are fully expanded so the demo can render one row per shadowed event without a live session/events fetch. In a real machine session, the compact-before column would lazy-load each seq via session/events {seq} (P0-1 expander). Below we ALSO include synthetic previews for each shadowed seq under _shadowedPreview so the demo can render text without wire access. Consumers of the diff renderer should call the wire path in production; this field is a demo-only convenience."
|
|
},
|
|
{
|
|
"_shadowedPreview": [
|
|
{ "seq": 121, "type": "user/message", "gist": "Refactor App: split shell and core, keep public API." },
|
|
{ "seq": 122, "type": "assistant/message", "gist": "Plan: 1) read src/lib/app.ts 2) draft AppShell/AppCore 3) move utils." },
|
|
{ "seq": 123, "type": "tool/call", "gist": "read(path=src/lib/app.ts)" },
|
|
{ "seq": 124, "type": "tool/result", "gist": "app.ts 240 lines, 3 exported functions, mixed cli+business logic." },
|
|
{ "seq": 125, "type": "assistant/message", "gist": "Splitting App into AppShell (cli parsing) and AppCore (business)." },
|
|
{ "seq": 126, "type": "tool/call", "gist": "edit(path=src/lib/app.ts, old=..., new=...)" },
|
|
{ "seq": 127, "type": "tool/result", "gist": "edit applied (1 replacement); AppShell/AppCore extracted." },
|
|
{ "seq": 128, "type": "tool/call", "gist": "write_file(path=src/lib/util/parse-args.ts)" },
|
|
{ "seq": 129, "type": "tool/result", "gist": "wrote 88 bytes" },
|
|
{ "seq": 130, "type": "tool/call", "gist": "run_bash(command=pnpm test)" },
|
|
{ "seq": 131, "type": "tool/result", "gist": "96 passed / 0 failed; 1.2s" },
|
|
{ "seq": 132, "type": "assistant/message", "gist": "Refactor complete — will summarise the session so far." }
|
|
]
|
|
}
|
|
]
|