Repository navigation
Conversation
|
👋 Hi! This PR is currently in draft status. Workflow runs won't be auto-approved until it's marked as ready for review. When you're ready, click "Ready for review" and the workflows will be approved on the next cycle automatically. |
UX Review (Opus 5.5, fork) — ⏭️ skippedRevision |
GPT 6.1 Review (fork) — ✅ no blocking findingsReviewed Review detailsNo findings. |
Design Review (Opus 5.5, fork) — ✅ PASSDesign-level review of Design-Verdict: PASS Nothing blocks: the reversal of the verbatim-user-row pin is stated as a deliberate decision, the serve-time alternative is weighed in Notes
[DESIGN-REVIEWED] 83b4575 Description read[DESCRIPTION-READ] ab8d5809d2f098f1c1a6109c5cfa7213c3fad5aa9591aceb8facb68262401c47 That sha256 names everything this verdict read: the PR description, plus the 1 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with |
First Principles Review (Opus 5.5, fork) — 🔴 BLOCKPremise-level review of First-Principles-Verdict: BLOCK product-shape change without accepted RFC: every channel user's typed text is now permanently rewritten at rest, reversing a pinned "stored verbatim" decision. The only record cited is the author's own draft RFC, not on main. Notes
What this change shipsInventory (7 items) — 3 justifiedIntent: stop a credential typed into channel chat from being stored and served verbatim. FIX.
Blockersproduct-shape change without accepted RFC: user-typed transcript rows rewritten at rest — product-shape [FIRST-PRINCIPLES-REVIEWED] 83b4575 Description read[DESCRIPTION-READ] c28d51b022ca5ac99de4f8c9a0d6d0c750ce8fc75dd01ed3eb4091292fd4b1f6 That sha256 names the PR description this verdict read. Recompute it with |
Opus 5.5 Review (fork) — ✅ no blocking findingsReviewed |
da18d20 to
1aa2f7d
Compare
1aa2f7d to
da362d4
Compare
da362d4 to
d12448d
Compare
fd1b5b7 to
ef70afe
Compare
ef70afe to
66509f8
Compare
|
PR triage (automated, 2026-10-05) — verdict: DISTINCT / BREAKING_CHANGE, recommend REBASE + maintainer decision Checked against
|
iamwhatever
left a comment
There was a problem hiding this comment.
Thanks @rnoack1 for the row-scrub and marker work. One rebase issue before this can land.
#17642 (remote-crew-sidebar RFC, wave 4) deleted remote_adopt.py on main, and #17662 deleted the rest of the relay path. The peer now owns its session and the dashboard opens a crew window onto it (#16757). These hunks touch code that no longer exists, so they are no longer needed:
src/kiro_crew/dashboard/remote_adopt.py(deleted on main): R460-466 and R483-490 (prepare_backfill_rows), R655-660 (apply_adopted_backfill).
Could you drop those hunks and rebase on main? Please keep the rest of the PR as it is. Thank you.
|
Thanks for catching it. Rebased onto main, and the head is now All three hunks are gone. The rest of the PR is unchanged. |
rnoack1
left a comment
There was a problem hiding this comment.
Reviewed the held carry_provenance fix; OK to push.
|
Holding at
— crew-pr-reviewer (r116) |
|
HOLD at befd463. The head moved since our r116 HOLD (0f7e757). In what I read I found no new code defect, but approval is blocked on these:
— crew-pr-reviewer (r123) |
|
@iamwhatever both items are done at head
One thing to flag: the rest of the PR did not stay as it was. The review lanes asked for a smaller change, so the rewritten-row mark and its UI cue moved to a later PR, and the per-persister user-row scrubs were replaced by one scrub at the transcript write boundary ( Could you re-review when you get a chance? Thank you. |
…written one Ten channel persisters stored the user's turn verbatim, so a credential typed into chat reached every dashboard reader; each persister now scrubs the row and a rewritten row carries a marker the reader can see. Split from a larger autonudge egress change: this unit is the inbound row scrub and its marker only.
|
Superseded by #36640: the same commit, rebased onto current main and opened from a branch of this repository. Sensitive Change Review cannot pass on a fork PR, so the change moves there. Closing this one. |
Problem / Motivation
Goal: A credential typed into channel chat is no longer stored verbatim in the transcript served to dashboard readers.
A persisted conversation row is an egress: it is served back to every connected dashboard client.
_redact_at_write_boundaryreturned early forrole == "user", so every channel, task and eval persister stored the user's turn verbatim, and a credential typed into chat was written to the transcript unmasked and then served to every reader of it.Why it matters
The credential reaches readers who should never see it, and it persists, so the exposure outlives the session that caused it.
Reversed decision: typed user text is no longer stored verbatim
This change reverses a decision this repository previously pinned, and the reversal is deliberate rather than incidental.
_redact_at_write_boundary's docstring read "text the user typed is stored verbatim", andtest_user_typed_content_is_stored_verbatimasserted it. Both change here: therole == "user"early return is deleted, so the boundary scrubs user rows too, and the test becomestest_a_user_row_arriving_through_the_shared_helper_is_scrubbed.The old pin's stated justification was that it matched "the dashboard's own write-back boundary". The dashboard slot save does not route through this boundary at all (it writes its rows in
chat_persistence), so nothing on the dashboard side relied on the exemption, and a row typed into the dashboard is unaffected by this change. A channel transcript row keeps no original, so the row on disk is the only copy, and storing it verbatim is exactly the egress this change closes.The reasoning is recorded in
docs/system-specs/modules/history.md, under the serve-time alternative: the transcript at rest is itself an egress, so storing a credential verbatim defers the exposure instead of removing it. The irreversible-false-positive cost is accepted and stated in that spec.Backwards compatibility
Compatible: both take-aways are internal to this repository and every reader of each is updated in this same commit.
imessage/transport_dispatch.pyleavesNON_EGRESS_REDACTION_MODULESand joins the redaction-sink registry as a session-title sink in the same commit, so it stays in exactly one bucket;test_a_module_cannot_be_in_both_bucketsandtest_allowlist_has_no_stale_entriesboth passWithheld rows, and who reads them. A user row goes through the host's composed policy, so a host whose companion policy fails to compose stores
<withheld: redaction unavailable>in place of a user row's body. Assistant and system rows keep the baseline redactor pair and are never withheld. That is an error state rather than a normal one, and the row is still written -- but on such a host these three readers see the placeholder instead of the user's text:contentfor both the search fold and the snippet, so a withheld body is not searchable on that hostcontentinto the consolidation span that feeds memory recalluserandassistantcontent into the one-line session summaryThe alternative is worse rather than better:
redact_via_contextRAISES on such a host, which took the whole message down instead of one field.What changed
_redact_at_write_boundaryscrubs a user row throughredact_log_via_context, the active credential policy, so a host with composed patterns of its own is not skipped; other roles keep the baseline pair. A new persister is covered without being registered. The decision to rewrite user-typed rows at rest is recorded in RFC: redact credentials in user-typed transcript rows at rest (docs/request-for-change/rfc-redact-user-rows-at-rest.md,draft, not yet on main).append, which the boundary now covers;nudge_channel_fire.pyandimessage/transport_dispatch.pyleave the non-egress allowlist.ConversationLogwrite, so they scrub both halves first.append_if_absenttreats a raw or a redacted copy on disk as the same row. Otherwise a window replay would append a credential-bearing row a second time.The exemption was measured before it was removed. An earlier revision deferred this, citing four timeouts above 120s in
test_history_consolidation_retry.pyandtest_history_locking_remediation.pyand attributing them to_might_contain_credential. That attribution was wrong: the pre-filter is flat on a long uniform run (0.32 ms at 32 kB) and gates the quadratic_CREDENTIAL_PATTERNSalternation, so the boundary costs 2.4 ms there and scales linearly. With the early return removed at this head, both suites pass (101 tests) in 8.5s, slowest test 2.9s.Not a goal
src/kiro_crew/security/is touched.Tests
ConversationLogis stored without its credential, and a clean one is stored unchanged. Negative control: restoring the early return reddens exactly the three boundary tests (this suite's two plus the shared-helper test in the one-conversation suite)test_display_time_redaction.py::test_load_leaves_user_content_rawnow writes its raw user row directly to disk, sinceappendno longer stores one; it still pins that the load-time pass leaves a user row aloneScope
One unit of a split of #7996, which is 9.3x this repository's added-line cap and cannot be reviewed as a single change. This unit is the inbound row scrub only. It depends on nothing else in that split and can merge on its own.