Files
deepseek-harness/examples/plan-acp-agent/tests/snapshots/plan-mode/input.json
kingwl 99650a201b feat(mode): the access cap — plan mode composes with the sandbox instead of banning bash
A ModeDefinition may declare access: the widest sandbox access shell
commands run under while the mode holds, on the SANDBOX_MODES ladder.
The bash seam gains the resolution point to hang it on: BashExecutor.
resolveMode(session) folds override ?? default and dispatches the new
bash/resolve-mode waterfall; dsh-tool-bash consults it at both the
stamping site and the escalation baseline; dsh-mode's clamp listener
takes the ladder minimum per call. Two independent log folds compose at
read time — the mode never writes the sandbox knob, so the two switch
in any order and the knob re-emerges intact on exit.

The built-in plan definition ships access: read-only with the bash trio
allowlisted CONDITIONALLY: both policy layers admit bash/bash_output/
bash_kill only while a confining executor is mounted (an unconfinable
shell cannot honor the cap), and a bash call carrying sandbox_permissions
under a cap is denied at the gate — no widening mid-mode; the widened
step belongs in the plan.

examples/plan-acp-agent swaps bash-local for sandbox-local +
bash-sandbox (workspace-write default, clamped read-only inside plan)
plus the approval seam; the re-recorded plan-mode arc runs a real cat
inside plan under the clamped sandbox, and modes-advertise now pins the
sandbox-mode and approval config options. RFC amended to the landed
shape (access cap section, orthogonality FAQ, deferred item resolved
into effects self-declaration).
2026-07-12 22:51:09 +08:00

29 lines
1001 B
JSON

{
"steps": [
{
"op": "initialize"
},
{
"op": "newSession"
},
{
"op": "setMode",
"modeId": "plan"
},
{
"op": "prompt",
"text": "The typo is on line 2 of notes.txt itself; the workspace contains no other file. Inspect it with the bash tool: run exactly `cat notes.txt` (relative path, no other command; plan mode keeps the sandbox read-only, and a plain cat succeeds). Then present a one-step plan titled '# Fix the greeting typo' via exit_plan_mode: the single step is editing line 2 of notes.txt to say hello world. After the review approves, apply exactly that one edit to notes.txt with the edit tool and stop — no other commands, do not look for any other file."
},
{
"op": "prompt",
"text": "Read notes.txt (relative path notes.txt) and reply with the single word DONE if line 2 now says hello world."
}
],
"elicitationAnswers": [
{
"action": "accept",
"choice": "Approve"
}
]
}