You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Commit b3e5ab4
Browse filesBrowse the repository at this point in the historyBrowse files
feat(console): Connect an Agent lists the AI assistants you authorized and your API keys
7
+
8
+
Clause-②: yes (widening)
9
+
Metadata-protocol: none
10
+
11
+
- **`@object-ui/app-shell`**: under the setup steps on Connect an Agent (which sits in Account's AI Assistant group), a "What can act as you" card lists the assistants the signed-in person authorized on the OAuth consent page, read from `oauth.authorizedClients.list()`: the client's name (its client id beside a fallback when it declares none), each granted scope in the consent page's own words, when it was authorized, when it was last used (the server records the hour, so it reads "in the hour from …") or that no use is recorded, and the organization only when it is not the active one. Assistant rows carry no Revoke yet; the card says so. Below them, the person's own API keys (`sys_api_key` rows filtered to the signed-in user, so a platform administrator sees their own keys here too): name, prefix, created, last used or never used, and the revoked or expired state. An active key's Revoke asks first, naming the key, then writes `revoked: true`, which ends the key's access at once; the list re-reads in place, and a key minted on the same page appears without a reload.
12
+
-**`@object-ui/i18n`**: the new `connectAgent.access.*` strings in all eleven packs. Scope wording reuses the existing `oauth.consent.scope.*` keys.
feat(audit): every data export writes one run-level `sys_audit_log` row, and the `export` action comes back with its writer
7
+
8
+
Clause-②: yes (widening)
9
+
Metadata-protocol: none
10
+
11
+
`GET /data/:object/export` used to leave the compliance ledger nothing to say: a user could download up to 50,000 rows of any object they may export, `sys_audit_log` included, and no row recorded who, what or how much. Each export the door streams now writes one row with `action: "export"`, the object, `record_id: null` and the exporter (user, actor, tenant, and the agent or impersonator when there is one). Its `metadata` carries the format, the row count, the effective and requested limit and whether the 50,000 cap clipped it, whether the file was truncated at the limit, a bounded summary of the filter (field names and operators, at most 32 terms, never a comparand), whether a search term narrowed it (never the term) and the outcome. An export whose read fails after the checks records `outcome: "failed"` with the status and machine code it answered.
12
+
13
+
A request refused before anything is read (no export permission, a bad parameter, an unparseable filter) writes no row; neither does the import template. The row is written after the file is finished, so the audit write never changes or delays the download; a ledger write that fails is logged at `error` once per cause and the export is unaffected.
14
+
15
+
`export` is declared again on `sys_audit_log.action` (it was retired under #8147 because nothing wrote it) and is shown by the `config_changes` list view beside `config_change` and `import`. The writer lives on the `audit` service slot as `recordDataExport`, beside `recordAuthEvent` and `recordActionInvocation`; a kernel without the audit plugin exports exactly as before.
feat(lint): warn when a permission set reads a roll-up summary over child data it cannot read
6
+
7
+
Clause-②: yes (widening) — `@objectstack/lint` exports one new rule function, `validateRollupSourceVisibility`, its two rule-id constants `SECURITY_ROLLUP_SOURCE_UNREADABLE` and `SECURITY_ROLLUP_SOURCE_ROW_SCOPED`, and the `RollupSourceVisibilityFinding` / `RollupSourceVisibilitySeverity` types; nothing that passed before is refused.
8
+
Metadata-protocol: none
9
+
10
+
A `summary` (roll-up) field keeps its documented semantics: the engine recomputes it as the system over the whole child collection, and on read it follows only the parent field's own field-level security. So a permission set that can see the summary field still reads the roll-up when it cannot read the aggregated child field, a child field the roll-up's `filter` tests, or the child object, or when it reads only some child rows. Two new author-time findings now say so, for each permission set and each summary field it can see:
11
+
12
+
-`security-rollup-source-unreadable` (warning) — the set cannot read `summaryOperations.object`, `summaryOperations.field`, or a top-level field key of `summaryOperations.filter`. A `count` reads no value, so its `field` is not judged; its `filter` is.
13
+
-`security-rollup-source-row-scoped` (info) — the set reads all of those, but one of its own read-class `rowLevelSecurity` policies on the child, or the child's `defaultRowScope`, narrows which child rows it reads.
14
+
15
+
"Can read" follows the runtime's three gates: the object grant (its own entry, otherwise `'*'`, which covers a private object only with View All or Modify All), the field entry (`readable: false`, made inert by View All or Modify All on the object), and the capability gate (an object's or field's `requiredPermissions` against the set's `systemPermissions`, which View All and Modify All do not lift). A set holding View All or Modify All on the child is still reported for a capability-gated child object or field, and for a row policy unless the child is private, has tenancy off or is auth-managed, the postures on which the bypass skips row security. A summary field the set cannot see, through any of the three gates on the parent, is not reported.
16
+
17
+
The fix line names the remedy and its reach: `fields: { 'invoice.total': { readable: false } }` on the set hides the summary field from every holder of that set, and a co-held set with no entry for it does not restore it; to keep it for some holders, declare `readable: true` on it in the sets they hold.
18
+
19
+
Both findings are advisory: they never fail `os lint`, `os validate` or `os build` (only `--strict` promotes a warning), and the rule does not run at the runtime publish gate. `os explain security-rollup-source-unreadable` and `os explain security-rollup-source-row-scoped` print the full reasoning.
Copy file name to clipboardExpand all lines: content/docs/deployment/validating-metadata.mdx
+2-1Lines changed: 2 additions & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -524,6 +524,7 @@ orthogonal to both, and no cell here can carry it; it is written out in
524
524
| Row-level-security predicates (#4983) — a permission set's policies, an object's `defaultRowScope`, a membership key's `filter` and a field permission entry's `when` (#25866); at the save door also a membership key nothing owns (#25234) | ✓ | ✓ | ✓ | ✓ᵖᵉᵒ |
525
525
| Membership key declarations — `userField` / `valueField` columns, one key on one object, a key the registered resolver also owns (save door only) (#25234) | ✓ | ✓ | ✓ | ✓ᵒ |
526
526
| Advisory: retired permission residue — the `allowRestore` / `allowPurge` default the parse strips in silence (`permission-retired-lifecycle-residue`, #17425) | ✓ | ✓ | ✓ | ✓ᵖᵉ |
527
+
| Advisory: a roll-up a permission set reads over child data it cannot read — the child object, the aggregated child field or a field the roll-up's `filter` tests unreadable to the set, through its grants, its field entries or a `requiredPermissions` capability it lacks (`security-rollup-source-unreadable`, a warning), or child rows a row-level policy narrows (`security-rollup-source-row-scoped`, a note); a roll-up aggregates the whole child collection and follows only the parent field's own permissions | ✓ | ✓ | ✓ | — |
| Public-form anonymous intake on this deployment's tenancy posture — advisory only (#21476) | — | — | — | ✓ᵛ |
529
530
| Public-form file fields — a form open to anonymous intake whose sections declare a `file`, `image`, `avatar`, `video` or `audio` field (#25650) | ✓ | ✓ | ✓ | ✓ᵛ |
@@ -809,7 +810,7 @@ A clean run walks the registry and reports timing:
Copy file name to clipboardExpand all lines: content/docs/permissions/tenant-audit-census.mdx
+15-15Lines changed: 15 additions & 15 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -132,7 +132,7 @@ are reported as `undecidable` rather than assumed either way.
132
132
133
133
The same holds twice over for the context. An options argument spelled as a
134
134
literal can be read; one spelled `options`, `{ ...opts }`, or handed through a
135
-
forwarding shim cannot, and **55 of the 264 sites are spelled that way**. A
135
+
forwarding shim cannot, and **55 of the 265 sites are spelled that way**. A
136
136
context resolved from an inline literal or a local `const` can be tested for
137
137
`isSystem`; one arriving from a helper call cannot.
138
138
@@ -197,10 +197,10 @@ reproduce them. Where it disagrees, it disagrees on the page:
197
197
198
198
| carried figure | where it survives | this census |
199
199
| :--- | :--- | ---: |
200
-
| 175 write call sites | quoted in the merged changeset |**264**|
200
+
| 175 write call sites | quoted in the merged changeset |**265**|
201
201
| 24 carrying no tenant context | quoted in the merged changeset |**2** provable and tenancy-enabled; **29** more whose options argument is unreadable |
202
-
| 127 of 175 statically decidable, 48 runtime-parameter-name sites | restated on the `isSystem`-scoping card |**187 of 264** decidable, **77** undecidable |
203
-
| 135 (77%) silenced by the `isSystem` guard before the posture gate | the lost issue body — **no surviving corroboration**|**not reproduced**: 147 decidably elevated, 0 decidably not, 109 undecidable |
202
+
| 127 of 175 statically decidable, 48 runtime-parameter-name sites | restated on the `isSystem`-scoping card |**188 of 265** decidable, **77** undecidable |
203
+
| 135 (77%) silenced by the `isSystem` guard before the posture gate | the lost issue body — **no surviving corroboration**|**not reproduced**: 148 decidably elevated, 0 decidably not, 109 undecidable |
204
204
| 141 and 132, two independent re-derivations | the card that filed this work | — |
205
205
206
206
**The differences are not reconciled, and deliberately so.** The old census's
@@ -217,11 +217,11 @@ would report a smaller number and would not say so.
217
217
218
218
The fourth row is the one worth flagging to anyone citing it. **The 135 / 77%
219
219
figure has no surviving corroboration anywhere in the tree.** This census reads
220
-
147 of 264 (56%) as decidably elevated, with 109 more whose elevation is a
220
+
148 of 265 (56%) as decidably elevated, with 109 more whose elevation is a
221
221
run-time fact — so the claim is neither confirmed nor refuted, and the honest
222
222
answer is that a static reading cannot settle it.
223
223
224
-
⇒ **Cite `2 / 264`, and say what it is**: the sites whose options argument was
224
+
⇒ **Cite `2 / 265`, and say what it is**: the sites whose options argument was
225
225
READ and holds no tenant context, against a decidably tenancy-enabled object.
226
226
That is the control's provable yield surface. ⛔ Do not cite it as "the sites
227
227
without tenant context" — **29 further sites** have an options argument this
@@ -233,28 +233,28 @@ cannot read, and they are neither in nor out.
233
233
234
234
| what | count |
235
235
| :--- | ---: |
236
-
| write call sites on the application surface |**264**|
237
-
| …whose object name is statically decidable |187|
236
+
| write call sites on the application surface |**265**|
237
+
| …whose object name is statically decidable |188|
238
238
| …whose object name is chosen at run time | 77 |
239
-
| …against an object with tenancy ENABLED |181|
239
+
| …against an object with tenancy ENABLED |182|
240
240
| …against an object that declares tenancy off | 6 |
241
-
| threading a tenant context |201|
241
+
| threading a tenant context |202|
242
242
| PROVABLY carrying none (options read, no context key) |**8**|
243
243
| …of those, against a decidably tenancy-enabled object |**2**|
244
244
| options argument UNREADABLE — may or may not carry one | 55 |
245
245
| …of those, against a decidably tenancy-enabled object | 29 |
246
-
| threading a decidably ELEVATED (`isSystem`) context |147|
246
+
| threading a decidably ELEVATED (`isSystem`) context |148|
247
247
| threading a context that is decidably NOT elevated | 0 |
248
248
| threading a context whose elevation is a run-time fact | 109 |
249
249
250
250
| how the instrument reached the site | count |
251
251
| :--- | ---: |
252
-
| receiver carried a readable engine type |202|
252
+
| receiver carried a readable engine type |203|
253
253
| receiver erased, placed by the object NAME | 52 |
254
254
| receiver erased, placed by an `object: string` PARAMETER | 5 |
255
255
| receiver erased, placed by an `UNTYPED_RECEIVERS` row | 5 |
256
256
257
-
| object name spelled inline |118|
257
+
| object name spelled inline |119|
258
258
| object name spelled through a `const`| 69 |
259
259
| object name is an `object: string` parameter | 7 |
260
260
| object name is some other run-time expression | 70 |
@@ -307,11 +307,11 @@ holds still. They are required to be HERE and to say WHEN they were true;
307
307
their values are not compared. The reasoning, and the measurement behind it,
308
308
are in `scripts/check-tenant-audit-census.mjs`.
309
309
310
-
Measured on 2026-11-06 at `19cadde19`.
310
+
Measured on 2026-11-06 at `4addb47fd`.
311
311
312
312
| corpus scale (not enforced) | count |
313
313
| :--- | ---: |
314
-
| tracked non-test sources scanned |1044|
314
+
| tracked non-test sources scanned |1045|
315
315
| engine-shaped types recognised | 112 |
316
316
| declared objects in the registry | 130 |
317
317
| same-named calls subtracted as non-engine | 234 |
0 commit comments