@akasecurity/ai-tc-antigravity 0.9.11 → 0.9.13

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -811,10 +811,11 @@ what happened, show the one-liner so they can retry later, and continue the
811
811
  wizard normally. The plugin is already fully set up and works on its own; a
812
812
  failed CLI install changes nothing about that.
813
813
 
814
- **Close the wizard.** The first-run summary already confirmed the saved posture
815
- and pointed at the aka-health skill. Whichever way the CLI offer went —
816
- installed, declined, or a failed install you already reported — end with one
817
- warm close: "That's it — I'm watching out for Antigravity going forward."
814
+ **Close the wizard.** The first-run summary already confirmed the saved posture,
815
+ reported the health score, and pointed at the aka-dashboard and aka-scan
816
+ skills. Whichever way the CLI offer went — installed, declined, or a failed
817
+ install you already reported — end with one warm close: "That's it — I'm
818
+ watching out for Antigravity going forward."
818
819
 
819
820
  Before you finish, confirm every AKA_SHOW region on the path you took was
820
821
  relayed to the user. If you summarized one instead of pasting it, paste it now.
@@ -844,12 +845,21 @@ reaches the model. AKA records it (so it appears in findings and the dashboard)
844
845
  but cannot stop it. This is the single biggest difference from the Claude Code
845
846
  plugin, where a flagged prompt is blocked outright.
846
847
 
847
- **A redact policy blocks instead of masking.** Claude Code's PreToolUse can hand
848
- back rewritten tool arguments; Antigravity's cannot — its output is
849
- `{ decision, reason, permissionOverrides }` with no field for modified args. So
850
- when a policy says "redact", AKA denies the tool call and explains what to
851
- remove, rather than masking the value and letting the call run. This applies to
852
- file writes too, not just shell commands.
848
+ **A redact policy cannot mask, and what it does instead is a setting.** Claude
849
+ Code's PreToolUse can hand back rewritten tool arguments; Antigravity's cannot —
850
+ its output is `{ decision, reason, permissionOverrides }` with no field for
851
+ modified args. So a policy that says "redact" cannot be carried out here at all,
852
+ on file writes as much as on shell commands.
853
+
854
+ What happens in its place is the workspace's **redact fallback** (`redactFallback` in
855
+ `~/.aka/settings/settings.json`, or pinned for a machine by an administrator). It ships as **warn**, which on this host means the tool call
856
+ **runs with the value unmasked** — and since PreToolUse has no message channel
857
+ (below), the warning appears only in findings and the dashboard, not in the
858
+ session. Set it to **block** to get the older behaviour, where AKA denies the
859
+ call and explains what to remove. Either way the finding is recorded with the
860
+ action that actually applied, never as a "redact" that did not happen. Say this
861
+ plainly if a user asks why a flagged value went through: this host cannot mask,
862
+ and their fallback is set to let it run.
853
863
 
854
864
  **A warn is recorded but not shown inline.** PreToolUse's output has no
855
865
  message channel (its `reason` accompanies a deny), so a `warn` policy captures