Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
46 commits
Select commit Hold shift + click to select a range
c68197c
docs: backfill Deep Agents Code configuration
mdrxy Aug 26, 2026
f8cb006
Apply suggestion from @mdrxy
mdrxy Aug 26, 2026
86f8ae9
Apply suggestion from @mdrxy
mdrxy Aug 26, 2026
acb8f70
Apply suggestion from @mdrxy
mdrxy Aug 26, 2026
53f104d
Apply suggestion from @mdrxy
mdrxy Aug 26, 2026
399aed5
Apply suggestion from @mdrxy
mdrxy Aug 26, 2026
ce8551b
Apply suggestion from @mdrxy
mdrxy Aug 26, 2026
1797369
Apply suggestion from @mdrxy
mdrxy Aug 26, 2026
eefa0c2
Apply suggestion from @mdrxy
mdrxy Aug 26, 2026
a533e45
Apply suggestion from @mdrxy
mdrxy Aug 26, 2026
e4ab890
Apply suggestion from @mdrxy
mdrxy Aug 26, 2026
12e478f
Apply suggestion from @mdrxy
mdrxy Aug 26, 2026
0a7b22d
Apply suggestion from @mdrxy
mdrxy Aug 26, 2026
38481f2
Apply suggestion from @mdrxy
mdrxy Aug 26, 2026
4a99a44
Apply suggestion from @mdrxy
mdrxy Aug 26, 2026
24705ac
Apply suggestion from @mdrxy
mdrxy Aug 26, 2026
25ec453
Apply suggestion from @mdrxy
mdrxy Aug 26, 2026
62ba6cb
feat: document Deep Agents Code managed configuration
mdrxy Aug 21, 2026
3ddd615
Apply suggestion from @mdrxy
mdrxy Aug 24, 2026
cf6185e
docs: align managed config with current dcode
mdrxy Aug 24, 2026
3c2ffff
Apply suggestion from @mdrxy
mdrxy Aug 24, 2026
813ffee
docs: document remote managed configuration
mdrxy Aug 26, 2026
8f2318e
docs: make managed configuration guidance clearer
mdrxy Aug 26, 2026
d4b6e07
Apply suggestion from @mdrxy
mdrxy Aug 27, 2026
dbacb5f
Apply suggestion from @mdrxy
mdrxy Aug 27, 2026
33e6f1a
docs: group model restrictions with managed policy
mdrxy Aug 27, 2026
a2d66cc
docs: address configuration review feedback
mdrxy Aug 27, 2026
e8305d0
docs: update Deep Agents Code approval workflows
mdrxy Aug 26, 2026
f598d90
Apply suggestion from @mdrxy
mdrxy Aug 26, 2026
2334ec9
Apply suggestion from @mdrxy
mdrxy Aug 26, 2026
d5b8121
Apply suggestion from @mdrxy
mdrxy Aug 26, 2026
06a7726
Apply suggestion from @mdrxy
mdrxy Aug 26, 2026
77ba3ae
Apply suggestion from @mdrxy
mdrxy Aug 26, 2026
0940f1d
Apply suggestion from @mdrxy
mdrxy Aug 27, 2026
a4b5197
Apply suggestion from @mdrxy
mdrxy Aug 27, 2026
1a2eac8
Apply suggestion from @mdrxy
mdrxy Aug 27, 2026
60da870
docs: address remaining configuration feedback
mdrxy Aug 27, 2026
85040a7
docs: refresh runtime and retry configuration
mdrxy Aug 27, 2026
6144558
Merge branch 'mdrxy/docs/dcode-configuration-backfill' into mdrxy/doc…
mdrxy Aug 27, 2026
b2c4c16
docs: align approval workflows with current behavior
mdrxy Aug 27, 2026
f75f3a6
docs: address configuration review
mdrxy Aug 27, 2026
b704605
docs: address configuration review
mdrxy Aug 27, 2026
402f0bc
Merge branch 'mdrxy/docs/dcode-configuration-backfill' into mdrxy/doc…
mdrxy Aug 27, 2026
33676ec
docs: address approval workflow review
mdrxy Aug 27, 2026
f2a8081
docs: fix goal configuration reference
mdrxy Aug 27, 2026
57467f5
Merge main into Deep Agents Code approval docs
mdrxy Aug 27, 2026
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
42 changes: 9 additions & 33 deletions src/oss/deepagents/code/approval-modes.mdx
Original file line number Diff line number Diff line change
Expand Up @@ -22,26 +22,16 @@ Read-only tools such as `ls`, `read_file`, `glob`, and `grep` always run without
| **Auto** | Approves routine actions automatically; asks the model to review anything uncertain; falls back to you after repeated denials or failures |
| **YOLO** | Runs gated actions with no review at all |

Toggle between Manual and Auto at any time during a session with `Shift+Tab` or `Ctrl+T`. YOLO cannot be entered through the keyboard toggle.

<Warning>
Auto is an authorization heuristic for a local coding agent. It is **not** sandbox containment, an operating-system boundary, or a guarantee that model-generated actions are safe.
</Warning>

## Enable Auto

Auto is an experimental beta. To use it, set the opt-in flag and then choose Auto for your session.
Auto is eligible only in interactive, unsandboxed sessions.

<Steps>

<Step title="Set the experimental opt-in" icon="key">
Add the environment variable to your shell or `~/.deepagents/.env`:

```bash
export DEEPAGENTS_CODE_EXPERIMENTAL=1
```
</Step>

<Step title="Launch with Auto" icon="terminal">
```bash
dcode -y
Expand All @@ -54,12 +44,10 @@ Auto is an experimental beta. To use it, set the opt-in flag and then choose Aut
mode = "auto"
```

You can also toggle Auto on and off mid-session with `Shift+Tab` or `Ctrl+T`.
You can also cycle to Auto mid-session with `Shift+Tab`.
</Step>
</Steps>

If Auto is requested without the experimental opt-in, or in a sandboxed session, it falls back to Manual with a warning.

## Enable YOLO

YOLO runs gated actions without any review. Use it only when you accept that the agent can take any action without asking.
Expand All @@ -82,14 +70,14 @@ YOLO runs gated actions without any review. Use it only when you accept that the
</Step>
</Steps>

A session launched in YOLO moves to Manual when you press `Shift+Tab` or `Ctrl+T`. You cannot switch back to YOLO with the keyboard toggle.
Press `Shift+Tab` to cycle YOLO Manual → Auto → YOLO. Set [`startup.yolo_switcher = false`](/oss/deepagents/code/config-file#startup-approval-mode) to omit YOLO from the cycle.

## How Auto works

Auto keeps the same gated-action rules as Manual but changes how those actions are reviewed. It uses two stages:

1. **Routine actions run automatically.** A write to a source file like `src/parser.py` or a read-only Git command like `git status` proceeds without a prompt. Sensitive targets like `.github/workflows/ci.yml` or mutating commands like `git commit` go to the next stage.
2. **The model reviews the rest.** For anything not clearly routine, the active model checks whether the action matches what you asked for. Only your literal prompt can authorize an action. If the model denies a call, the agent gets an error result and can revise its plan.
2. **The model reviews the rest.** For anything not clearly routine, the active model checks whether the action matches your requested outcome. Ordinary steps reasonably necessary for that outcome can proceed even when you did not name each implementation detail. High-risk effects, such as sending local content to an unconfigured destination, still require explicit authorization. If the model denies a call, the agent gets an error result and can revise its plan.

After repeated denials or classifier failures, Auto stops and shows you the normal approval prompt for the next batch, then continues in Auto mode.

Expand Down Expand Up @@ -202,27 +190,15 @@ The decision plan is bound to the thread, mode, batch, and exact gated calls. Mi

## Where Auto and YOLO are available

Auto and YOLO are interactive-mode features. They are not available in non-interactive mode (`-n` or piped stdin) or in ACP server mode. Headless runs use fail-closed MCP routing and `--shell-allow-list` for shell access.

Auto also falls back to Manual when:

- `DEEPAGENTS_CODE_EXPERIMENTAL=1` is not set.
- A remote `--sandbox` is active (Auto is for unsandboxed local sessions only).
Auto and YOLO are interactive-mode features. Auto is eligible only in an interactive, unsandboxed session; a remote `--sandbox` forces it to Manual. YOLO is interactive-only as well and requires the risk acknowledgement regardless of sandbox. Headless runs use fail-closed MCP routing and `--shell-allow-list` for shell access.

## Reference
In non-interactive mode (`-n` or piped stdin), `-y`/`--auto-approve` and `--yolo` are ignored. Headless runs use fail-closed MCP routing and `--shell-allow-list` for shell access.

### Flag and config precedence
### Remember the last mode across sessions

`--yolo` takes priority over `-y`/`--auto-approve`, which takes priority over `[startup].mode`.
When no flag or configured mode applies, Deep Agents Code restores the last selected Manual or Auto mode. YOLO must be explicitly selected. See [Startup approval mode](/oss/deepagents/code/config-file#startup-approval-mode).

| Source | Value | Selects |
|---|---|---|
| `--yolo` | flag | YOLO (interactive only, after acknowledgement) |
| `-y`, `--auto-approve` | flag | Auto (requires `DEEPAGENTS_CODE_EXPERIMENTAL=1`) |
| `[startup].mode` | `"manual"` | Manual |
| `[startup].mode` | `"auto"` | Auto |
| `[startup].mode` | `"yolo"` | YOLO |
| `Shift+Tab`, `Ctrl+T` | toggle | Manual and Auto (never enters YOLO) |
For flag and configuration precedence, see [Startup approval mode](/oss/deepagents/code/config-file#startup-approval-mode).

## See also

Expand Down
10 changes: 9 additions & 1 deletion src/oss/deepagents/code/goals-and-rubrics.mdx
Original file line number Diff line number Diff line change
Expand Up @@ -29,7 +29,15 @@ Use `/goal` when you know the outcome you want, but want Deep Agents Code to pro

Deep Agents Code drafts acceptance criteria for review before starting the task.

In the inline review you can accept the proposal, edit the criteria, request another revision, or cancel it. After you accept the criteria, the goal stays active across turns until it is paused, completed, blocked, or cleared.
How draft criteria are handled depends on your [approval mode](/oss/deepagents/code/approval-modes):

<Accordion title="Review behavior by approval mode">
- **Manual**: Always opens the inline review.
- **Auto**: Opens the review by default. Set `goals.auto_accept_criteria = true` in `~/.deepagents/config.toml` to apply generated criteria automatically.
- **YOLO**: Applies generated criteria without review.
</Accordion>

After the criteria are accepted, the goal stays active across turns until it is paused, completed, blocked, or cleared.

This approach lets you work toward a larger objective over multiple turns:

Expand Down
Loading