diff --git a/docs/post-fill-exposure.md b/docs/post-fill-exposure.md index 36c52b8..fc8af81 100644 --- a/docs/post-fill-exposure.md +++ b/docs/post-fill-exposure.md @@ -109,6 +109,7 @@ names the evidence, or says unverified. | D4 | Agent enables the Network domain and captures the login POST body | **host** — the POST goes to the bound origin by design; watching it needs raw CDP, the same capability as D1, and is a harness question. kharon's own client never enables Network | Unverified | | D5 | Observation *during* the fill window, before submit | **kharon**, in part — fill-and-submit is one in-page call on a millisecond budget, so with `submit: true` the readable window is the page's own submit handling. With `submit: false` there is no window to shrink (A4) | The budget is tested (REQ-API-009); a concurrent-observer race test does not exist — unverified as a race | | D6 | kharon's own outputs: tool results, audit log, broker/connector journals, agent transcripts | **kharon** — status and names only, no values anywhere (above) | Yes: #201 filesystem sweep (0 hits) and the #353 leak check (canary absent from audit, journals, transcripts, sessions) | +| D7 | `kharon_open` (#357) navigates a broker-owned tab to the item's saved login URL; on the shared CDP the agent can read the tab's full URL — path and query included — not just the origin the tool result returns | **host** — the login URL is item metadata, not a credential: the item's URI is already returned by `kharon_list`/`kharon_search`, and secrets live in the vault fields, never in the URI. kharon's own part is kept: the tool result carries only `{target_id, origin}` (REQ-API-012) | Unverified as a read path — no canary step reads a tab URL back; the tool-result shape is covered by the #357 tests | ## Per-host notes