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

Merge remote-tracking branch 'origin/feat/plan-mode' into codex/pr239-plan-mode-review-fixes

Browse Source 
# Conflicts:
#	.agents/notes/implemented/feature/2026-07-07-plan-mode.md
#	.agents/notes/implemented/simplification/2026-07-04-fold-stdio-ui-helper.md
#	docs/capability-seams.md
#	docs/config-catalog.md
#	docs/event-producer-consumer.md
#	examples/acp-agent/tests/plan-mode.e2e.ts
#	examples/acp-agent/tests/snapshots/plan-mode/tool-schemas.expected.json
#	examples/plan-acp-agent/README.md
#	examples/plan-acp-agent/cordis.yml
#	examples/plan-acp-agent/tests/acp.snapshot.ts
#	scripts/gen-doc-graphs.ts
This commit is contained in:
Tianyi Cui
2026-07-21 01:21:33 +08:00
parent 6aa71aca94 81c239bc60
commit 4466d582c4
294 changed files with 14357 additions and 656 deletions
Download Patch File Download Diff File Expand all files Collapse all files
examples/acp-agent/README.md
+8 -2
View File
@@ -7,7 +7,7 @@ pnpm run demo:acp # needs DEEPSEEK_API_KEY (repo-root .env or env)
pnpm run demo:code-mode acp # the same server in Code Mode: one wire tool, run_code
```
The leaf config loads the ACP app, DeepSeek adapter, sandboxed bash, the sandboxed filesystem stack, approval and permission services, model-facing tools, and repeat guard. The app bundles the agent spine, JSONL persistence, and bridge, creates agents on `session/new`, and keeps stdout logger-free. [`fs.cordis.yml`](fs.cordis.yml) adds local tool-result spill storage for its dedicated scenarios; [`code-mode.cordis.yml`](code-mode.cordis.yml) adds `run_code` and its generated TypeScript SDK. See [Code Mode](../../packages/core/tools/README.md#code-mode).
The leaf config loads the ACP app, DeepSeek adapter, session modes, sandboxed bash, the sandboxed filesystem stack, approval and permission services, model-facing tools, and repeat guard. The app bundles the agent spine, JSONL persistence, and bridge, creates agents on `session/new`, and keeps stdout logger-free. [`fs.cordis.yml`](fs.cordis.yml) adds local tool-result spill storage for its dedicated scenarios; [`code-mode.cordis.yml`](code-mode.cordis.yml) adds `run_code` and its generated TypeScript SDK. See [Code Mode](../../packages/core/tools/README.md#code-mode).
## stdout is the protocol
@@ -31,9 +31,15 @@ Add to your Zed `settings.json` under `agent_servers`:
The editor sets each session's `cwd` to the project it opens, and bash uses that directory as its workdir. The current sandbox write boundary is nevertheless fixed when the server starts (`workspaceRoot: process.cwd()`), so launch the server from the workspace it should be allowed to modify; making that root session-scoped is deferred in the [sandbox Agent Note](../../.agents/notes/implemented/feature/2026-07-06-sandbox.md). The filesystem tools now ride the same sandbox policy through [`@deepseek-ai/dsh-fs-sandbox`](../../packages/fs/fs-sandbox/), so `read`/`write`/`edit` are available under every mode and confined to the same `workspaceRoot`.
## Plan mode
The same `demo:acp` server composes [`@deepseek-ai/dsh-mode`](../../packages/mode/mode/), so a capable client advertises `default` and `plan` in its mode picker. This composition owns the complete plan instructions in [`cordis.yml`](cordis.yml): remain in the selected mode, inspect before asking, avoid mutations, resolve discoverable repository facts, and submit a decision-complete plan through `exit_plan_mode`. Those are the instrumental behaviors shared by the local Codex and Claude Code references; product-specific plan files, phase machinery, and protocol tags stay out of the plugin contract.
Plan mode adds only that configured guidance section. Every tool, including `exit_plan_mode`, keeps the same schema in `default` and `plan`; the exit tool describes itself as plan-only and rejects if called outside plan mode. Stable native schemas and Code Mode SDK bindings avoid tool-catalog churn at the transition. `ask_user_question` carries blocking user-owned choices through ACP elicitation, while `exit_plan_mode` renders the exact logged plan for approval and returns keep-planning feedback to the model. The mode picker and permission select remain independent: switching modes never changes sandbox or approval state, and deployments that need a hard read-only planning floor configure that policy separately. The [plan-mode Agent Note](../../.agents/notes/implemented/feature/2026-07-07-plan-mode.md) owns the state and review contract.
## Snapshot tests (record-once / replay-deterministic)
This example hosts the ACP snapshot suite. It replays through `dsh-llm-replay`, which reconstructs model streams from `assistant/chunk` events in each scenario's session JSONL. Recording runs the real ACP agent and harvests its logs; refresh keeps the committed transcript as mock input and rewrites current replay outputs. `replay.override.json` covers throw and hang cases that chunks cannot express, and an optional `workspace/` seeds files. The [snapshot Agent Note](../../.agents/notes/implemented/testing/2026-06-19-acp-snapshot-tests.md) owns the ACP harness design.
This example hosts the ACP snapshot suite, including the picker advertisement and both plan-review branches. It replays through `dsh-llm-replay`, which reconstructs model streams from `assistant/chunk` events in each scenario's session JSONL. Recording runs the real ACP agent and harvests its logs; refresh keeps the committed transcript as mock input and rewrites current replay outputs. `replay.override.json` covers throw and hang cases that chunks cannot express, and an optional `workspace/` seeds files. The [snapshot Agent Note](../../.agents/notes/implemented/testing/2026-06-19-acp-snapshot-tests.md) owns the ACP harness design.
## Permissions and sandboxing