* fix: surface Codex input prompts across runtimes Codex structured input now reaches bounded Gateway questions in native and ACP runs, with exact turn ownership, explicit unsupported outcomes, and cancellation fencing. Consume the published ACPX elicitation support. * chore: align elicitation helpers with current guards Use protocol-specific helper names required by current main and update the reservation regression to the generalized input owner. * fix: formalize structured input SDK surface Expose one documented, frozen agent-harness structured-input contract with runtime and subpath coverage, replacing the accidental function-property API. * fix: satisfy elicitation architecture gates Register the real-process ACPX fixture as an executable test root and move shared structured-input types into the boundary leaf to keep Knip and Madge clean. * fix: remove structured input lint suppression Preserve the rejected control and invisible-character ranges with an explicit code-point check so the production suppression inventory stays closed.
6.7 KiB
summary, read_when, title
| summary | read_when | title | |||
|---|---|---|---|---|---|
| Permission modes for host exec, Codex Guardian approvals, and ACPX harness sessions |
|
Permission modes |
Permission modes decide how much authority an agent has before it runs host commands, writes files, or asks a backend harness for extra access.
Permission mode is separate from `tools.exec.host=auto`. `tools.exec.host` chooses where a command runs. `tools.exec.mode` chooses how host exec is approved.Recommended default
Use auto for coding agents that need useful host access without making every miss a human prompt:
openclaw config set tools.exec.mode auto
openclaw approvals get
openclaw gateway restart
Then verify the effective policy:
openclaw exec-policy show
OpenClaw host exec modes
tools.exec.mode is the normalized policy surface for host exec. Each mode resolves to an underlying security (allowlist strictness) and ask (prompt-on-miss) pair:
| Mode | security / ask | Behavior | Use when |
|---|---|---|---|
deny |
deny / off |
Block host exec entirely. | No host commands are allowed. |
allowlist |
allowlist / off |
Run only allowlisted commands; silently deny misses. | You have a known-safe command set. |
ask |
allowlist / on-miss |
Run allowlist matches; ask a human on misses. | A human should review every new command. |
auto |
allowlist / on-miss |
Run allowlist matches; send misses through auto-review before falling back to human approval. | Coding sessions need practical guarded access. |
full |
full / off |
Run host exec without prompts. | This trusted host/session should skip approval gates. |
ask and auto share the same allowlist/ask settings; auto additionally enables the native auto-reviewer, which decides misses itself and only defers to the configured human approval route when it cannot safely approve.
For the full host exec policy, local approvals file, allowlist schema, safe bins, and forwarding behavior, see Exec approvals.
Codex Guardian mapping
For native Codex app-server sessions, tools.exec.mode: "auto" drives Codex toward Guardian-reviewed approvals when the local Codex requirements allow it. Typical resulting values:
| Codex field | Typical value |
|---|---|
approvalPolicy |
on-request |
approvalsReviewer |
auto_review |
sandbox |
workspace-write |
auto mode forces this policy over any configured Codex sandbox/approval overrides, so it does not preserve legacy unsafe combinations such as approvalPolicy: "never" with sandbox: "danger-full-access". tools.exec.mode: "deny" and "allowlist" block Codex app-server local execution entirely. Use tools.exec.mode: "full" only when you intentionally want the no-approval posture.
For app-server setup, auth order, and native Codex runtime details, see Codex harness.
ACPX harness permissions
ACPX sessions have no interactive TTY for permission prompts. Supported ACP
form and URL requests can still reach the operator as Gateway questions during
a channel-delivered turn; those are separate from permission approval. ACPX
uses separate harness-level settings under plugins.entries.acpx.config:
| Setting | Values | Meaning |
|---|---|---|
permissionMode |
approve-reads |
Auto-approve reads only. |
permissionMode |
approve-all |
Auto-approve writes and shell commands. |
permissionMode |
deny-all |
Deny all permission prompts. |
nonInteractivePermissions |
fail |
Abort when a prompt would be required. |
nonInteractivePermissions |
deny |
Deny the prompt and continue when possible. |
Set ACPX permissions separately from OpenClaw exec approvals:
openclaw config set plugins.entries.acpx.config.permissionMode approve-all
openclaw config set plugins.entries.acpx.config.nonInteractivePermissions fail
openclaw gateway restart
Use approve-all as the ACPX break-glass equivalent of a no-prompt harness session. For setup details and failure modes, see ACP agents setup.
Choosing a mode
| Goal | Configure |
|---|---|
| Block host commands completely | tools.exec.mode: "deny" |
| Let known-safe commands run only | tools.exec.mode: "allowlist" |
| Ask a human for every new command shape | tools.exec.mode: "ask" |
| Use Codex/OpenClaw auto-review before humans | tools.exec.mode: "auto" |
| Skip host exec approvals entirely | tools.exec.mode: "full" plus matching host approvals file |
| Make non-interactive ACPX sessions write/exec | plugins.entries.acpx.config.permissionMode: "approve-all" |
If a command still prompts or fails after changing mode, inspect both layers:
openclaw approvals get
openclaw exec-policy show
Host exec uses the stricter result of OpenClaw config and the host-local approvals file. ACPX harness permissions do not loosen host exec approvals, and host exec approvals do not loosen ACPX harness prompts.