omnius 1.0.793 → 1.0.795

package/dist/index.js

@@ -1,4 +1,4 @@
 [diffend] Oversized file quarantined before diffing.
 name: package/dist/index.js
-size: 39178099 bytes
-sha256: 77c9e3b8094d90a3f91db155f9e28ee0a3dfad588bb03538608f284fbc768c27
+size: 39183834 bytes
+sha256: f7f23898e576627ce93dab3197146494341cc1141e064dad36e55c352216201e

package/docs/DISCOVERY.json

@@ -35906,7 +35906,7 @@
       "id": "guide.work-orders-runtime-health-remediation-wo-57-telegram-delivery-truth-and-recovery-uppercase",
       "kind": "guide",
       "title": "WO-57: Telegram delivery truth, reminder authority, and bounded recovery",
-      "summary": "Status: implementation and post-1.0.792 recovery repair complete in working tree; focused validation green; full-suite, repository-delivery, and live acceptance gates remain open Priority: P0 Incident: October 1, 2026, live Telegram reminder and illegal-logging non-responses Owner: Omnius orchestrator and Telegram runtime Depends on: WO-41, WO-44, WO-49, WO-",
+      "summary": "Status: post-1.0.794 scope and OMNIUS-continuation repairs implemented in the working tree; focused validation green; full-suite, repository-delivery, and live acceptance gates remain open Priority: P0 Incident: October 1, 2026, live Telegram reminder and illegal-logging non-responses Owner: Omnius orchestrator and Telegram runtime Depends on: WO-41, WO-44, ",
       "keywords": [
         "work",
         "orders",

package/docs/DISCOVERY.md

@@ -609,7 +609,7 @@ Daemon equivalents are `GET /v1/discovery/bootstrap`, `GET /v1/discovery?q=<inte
 | `guide.work-orders-runtime-health-remediation-wo-54-case-theory-router-deferral-recovery-uppercase` | WO-54: Case Theory router deferral recovery | Status: complete Priority: P0 Incident: September 26, 2026, Case Theory messages 455–456 Runtime observed: globally installed Omnius 1.0.764 Owner: current implementation session |
 | `guide.work-orders-runtime-health-remediation-wo-55-telegram-pressure-recovery-uppercase` | WO-55: Telegram pressure recovery without silent terminal states | Status: complete Priority: P0 Incident: September 26, 2026, Case Theory messages 458-461 Owner: current implementation session |
 | `guide.work-orders-runtime-health-remediation-wo-56-evidence-driven-objective-loop-uppercase` | WO-56: Evidence-driven objective loop and nonblocking execution | Status: repository implementation complete; live Telegram acceptance remains an external gate Priority: P0 Incident: September 27, 2026, Telegram SSH-over-Cloudflare task spiral and TUI freeze Owner: Omnius orchestration, execution, and Telegram runtime Depends on: WO-05, WO-21, WO-27, WO-41, WO-44, WO-49, WO-54, WO-55 Completion rule: A checkbox may be mark |
-| `guide.work-orders-runtime-health-remediation-wo-57-telegram-delivery-truth-and-recovery-uppercase` | WO-57: Telegram delivery truth, reminder authority, and bounded recovery | Status: implementation and post-1.0.792 recovery repair complete in working tree; focused validation green; full-suite, repository-delivery, and live acceptance gates remain open Priority: P0 Incident: October 1, 2026, live Telegram reminder and illegal-logging non-responses Owner: Omnius orchestrator and Telegram runtime Depends on: WO-41, WO-44, WO-49, WO- |
+| `guide.work-orders-runtime-health-remediation-wo-57-telegram-delivery-truth-and-recovery-uppercase` | WO-57: Telegram delivery truth, reminder authority, and bounded recovery | Status: post-1.0.794 scope and OMNIUS-continuation repairs implemented in the working tree; focused validation green; full-suite, repository-delivery, and live acceptance gates remain open Priority: P0 Incident: October 1, 2026, live Telegram reminder and illegal-logging non-responses Owner: Omnius orchestrator and Telegram runtime Depends on: WO-41, WO-44,  |
 | `guide.work-orders-telegram-dmn-wo-22-dmn-outreach-and-learning-uppercase` | WO-22: DMN outreach, DM sharing, and outcome learning | On 2026-09-03 at 17:07 PDT the bot posted in the OMNIUS group without being addressed. The operator asked whether this was self-induced reflection. |
 | `guide.work-orders-telegram-dropbear-context-rca-workorder` | Telegram Dropbear Context Engineering RCA Work Order | Observed run: /home/roko/Documents/Projects/Adjacent/telegramtest/.omnius, run id 1782873796963-i5r7mv. |
 | `guide.work-orders-telegram-reliability-consolidation-2026-09-28` | Telegram Reliability Consolidation — 2026-09-28 | Rule: check a box only after the named behavior is implemented and its listed verification passes. A parent box stays open until every child box is checked. |

package/docs/work-orders/runtime-health-remediation/WO-57-telegram-delivery-truth-and-recovery.md

@@ -1,8 +1,8 @@
 # WO-57: Telegram delivery truth, reminder authority, and bounded recovery
 
-**Status:** implementation and post-1.0.792 recovery repair complete in working
-tree; focused validation green; full-suite, repository-delivery, and live
-acceptance gates remain open
+**Status:** post-1.0.794 scope and OMNIUS-continuation repairs implemented in
+the working tree; focused validation green; full-suite, repository-delivery,
+and live acceptance gates remain open
 **Priority:** P0
 **Incident:** October 1, 2026, live Telegram reminder and illegal-logging non-responses
 **Owner:** Omnius orchestrator and Telegram runtime
@@ -17,7 +17,9 @@ Restore truthful, bounded delivery across Telegram action, quick-chat, steering,
 and recovery paths. A successful observation must reach the user exactly once;
 a mutation may be claimed only after its typed tool receipt; inline research must
 not require an impossible external-provider receipt; recovery must reuse current
-verified evidence instead of replaying the full reasoning loop.
+verified evidence instead of replaying the full reasoning loop. A scoped
+reminder tool must never read or mutate reminders owned by another chat, topic,
+surface, or unscoped/global context.
 
 ## Baseline and live evidence
 
@@ -38,7 +40,7 @@ verified evidence instead of replaying the full reasoning loop.
       search-only and the conversation ended at assistant message 5705.
 - [x] Anchor false-acknowledgement failure: OMNIUS message 5712 was routed to
       zero-tool quick chat, delivered only `on it. pulling the Borneo illegal
-  logging data now.`, and was marked consumed without a retrieval run.
+logging data now.`, and was marked consumed without a retrieval run.
 - [x] Anchor follow-up router failure: Case Theory message 610 was a same-sender
       reply to the failed reminder-list request 609, but router reconciliation
       exhausted and the fallback lost its exact `inputId`, producing
@@ -52,6 +54,25 @@ verified evidence instead of replaying the full reasoning loop.
       `tool_search`, and unrelated `file_read` work, ended
       `incomplete_verification` twice, and was terminally failed at
       `2026-10-02T07:19:30.188Z` after three cycles with no public reply.
+- [x] Anchor the 1.0.793 OMNIUS continuation failure: message 5717 asked the
+      bot to complete the open Borneo task, but the router again chose chat;
+      quick run `telegram-quick:e0318ff63f7db207` delivered message 5718,
+      `on it. pulling the Borneo illegal logging data now.`, with no action
+      runner or retrieval receipt.
+- [x] Anchor the 1.0.793 Case Theory control failure: message 612 retrieved all
+      29 reminders, reached `deliver_due` at `2026-10-02T08:06:58.242Z`, then
+      admitted `tool_search` and `file_read`, completed with the full answer,
+      rejected its terminal review, and started a whole-task recovery cycle.
+- [x] Anchor the 1.0.794 scope breach: recovered message 612 called the locked
+      Case Theory reminder tool with `action:"list", filter:"all"`; the tool
+      returned the shared 29-reminder registry even though only four records
+      belonged to chat `-1003960149435`. Telegram message 614 then disclosed
+      reminder IDs and titles from OMNIUS and the requester's private DM.
+- [x] Anchor the remaining 1.0.794 OMNIUS regression: message 5719 said
+      `please pull the data`; its exact typed decision identified the bot as
+      direct addressee in the open-task thread, but the host boundary required
+      a literal mention/reply edge and left it in quick chat. Message 5720 said
+      `pulling it now. one sec.`; no action run or data result followed.
 
 Runtime evidence:
 
@@ -64,6 +85,11 @@ Runtime evidence:
 - `/home/roko/Documents/Projects/Adjacent/telegram_test/.omnius/telegram-runner-state/84e4e6e5411c97531047/completion-finalizations/telegram-84e4e6e5411c9753-1790924323167-1.json`
 - `/home/roko/Documents/Projects/Adjacent/telegram_test/.omnius/telegram-runner-state/84e4e6e5411c97531047/debug-library/runs/telegram-84e4e6e5411c9753-1790925147229-1/events.jsonl`
 - `/home/roko/Documents/Projects/Adjacent/telegram_test/.omnius/telegram-runner-state/84e4e6e5411c97531047/completion-finalizations/telegram-84e4e6e5411c9753-1790925147229-1.json`
+- `/home/roko/Documents/Projects/Adjacent/telegram_test/.omnius/telegram-turn-delivery-ledger/0f3dd3a8bebe1933d1d468175da402e64138954fcc52a8126512426297641ddf.json`
+- `/home/roko/Documents/Projects/Adjacent/telegram_test/.omnius/telegram-runner-state/84e4e6e5411c97531047/debug-library/runs/telegram-84e4e6e5411c9753-1790928285562-1/events.jsonl`
+- `/home/roko/Documents/Projects/Adjacent/telegram_test/.omnius/telegram-runner-state/84e4e6e5411c97531047/completion-finalizations/telegram-84e4e6e5411c9753-1790928285562-1.json`
+- `/home/roko/Documents/Projects/Adjacent/telegram_test/.omnius/telegram-turn-delivery-ledger/a607485959cb9f27313e04bd289eb590d3daa65cf8e751a6262246311ebe1463.json`
+- `/home/roko/Documents/Projects/Adjacent/telegram_test/.omnius/telegram-runner-state/84e4e6e5411c97531047/tool-results/b2fbc630.txt`
 
 ## Requirements and implementation checklist
 
@@ -246,6 +272,78 @@ Primary anchors:
 - `packages/cli/tests/telegram-bot-api-10.test.ts`
 - `packages/orchestrator/tests/agenticRunner-objective-control.test.ts`
 
+### REQ-TDT-010 — authoritative continuation and control-plane convergence
+
+- [x] Treat a directly addressed completion command as continuation of the
+      durable open task; it cannot be consumed by zero-tool quick chat.
+- [x] Bind terminal-review generation to the exact input, evidence IDs, and
+      message IDs enforced by its parser instead of generating against a
+      looser schema and rejecting the result afterward.
+- [x] Retry an invalid terminal review once at the review boundary using the
+      same immutable evidence snapshot; do not use a full object-work replay
+      as the first contract-repair mechanism.
+- [x] Make `deliver_due` authoritative for planning-free read-only contracts.
+      After the admitted observation, unrelated discovery/read tools must be
+      rejected and the runner must synthesize/complete.
+- [x] Keep checklist research in deferred enforcement so a first generic
+      observation does not prematurely terminate multi-source work.
+- [x] Add exact regressions for message 5717 continuation, review schema/parser
+      identity, in-place review repair, and message-612 enforcement mode.
+
+Primary anchors:
+
+- `packages/cli/src/tui/telegram-bridge.ts` — durable continuation boundary,
+  terminal-review schema/retry, and objective enforcement selection.
+- `packages/cli/tests/telegram-steering-finalization.test.ts`
+- `packages/cli/tests/telegram-terminal-metacognition.test.ts`
+- `packages/cli/tests/telegram-reminder-contract.test.ts`
+
+### REQ-TDT-011 — reminder scope is a read/write authorization boundary
+
+- [x] Enforce a locked reminder scope before listing records; `filter:"all"`
+      means every status inside the authorized scope, never every reminder in
+      the shared store.
+- [x] Enforce the same scope before complete, dismiss, or snooze mutations.
+      A foreign reminder ID must return the ordinary not-found result without
+      revealing that the record exists and without changing it.
+- [x] Canonicalize access by surface, target chat, and exact topic identity.
+      A group root, one topic, another group, a private DM, another surface,
+      and an unscoped record are mutually inaccessible from a locked context.
+- [x] Treat `userId`, username, owner, and label as metadata rather than public
+      group visibility keys so reminders remain shared within their group/topic.
+- [x] Preserve deliberately unlocked/global reminder administration behavior.
+- [x] Add execution regressions for group/DM/topic/unscoped list isolation and
+      cross-scope complete/dismiss/snooze rejection, plus a Telegram adapter
+      regression proving Case Theory cannot observe OMNIUS or private reminders.
+
+Primary anchors:
+
+- `packages/execution/src/tools/reminder.ts` — authoritative scope predicate for
+  all reminder reads and mutations.
+- `packages/execution/tests/tools/temporal-agency.test.ts`
+- `packages/cli/src/tui/telegram-bridge.ts` — locked current-chat tool adapter.
+- `packages/cli/tests/telegram-reminder-contract.test.ts`
+
+### REQ-TDT-012 — typed direct-turn continuation cannot acknowledge object work
+
+- [x] Treat an exact-current typed decision as direct bot address when its
+      self role is `addressee`, reply basis is `direct_turn`, addressed actors
+      contain the bot, and evidence cites the current message.
+- [x] When that authority accompanies an explicit thread-data request, require
+      action planning even without a literal `@mention` or Telegram reply edge.
+- [x] Cite the exact durable open task when present and clear any chat plan so
+      the scoped action planner must select capabilities before replying.
+- [x] Keep ambient/unaddressed human data talk and decisions addressed to a
+      different actor outside this repair.
+- [x] Replay exact OMNIUS message 5719 and assert it cannot enter quick chat or
+      emit a progress acknowledgement without an action plan.
+
+Primary anchors:
+
+- `packages/cli/src/tui/telegram-bridge.ts` — typed direct-turn address and
+  open-task data-continuation boundary.
+- `packages/cli/tests/telegram-steering-finalization.test.ts`
+
 ## Cross-surface validation gates
 
 - [x] Orchestrator focused tests pass for visible-response evidence, completion
@@ -280,6 +378,9 @@ Primary anchors:
 | REQ-TDT-007 | terminal recovery/memory eligibility             | retained-candidate recovery                        | focused green |
 | REQ-TDT-008 | capability admission + unavailable ownership     | 5712/610 + recovery ordering                       | focused green |
 | REQ-TDT-009 | recovered-contract migration + convergence gates | exact recovery + synthesis timing + bounded review | focused green |
+| REQ-TDT-010 | continuation + control-plane convergence         | 5717/612 + exact review contract                   | focused green |
+| REQ-TDT-011 | reminder read/write scope authorization          | group/DM/topic/foreign-ID isolation                | focused green |
+| REQ-TDT-012 | typed direct-turn action continuation            | exact OMNIUS 5719 + ambient controls               | focused green |
 
 ## Evidence log
 
@@ -448,3 +549,63 @@ Primary anchors:
 - The follow-up patch is not in the installed 1.0.792 runtime. A new build,
   version, install, and restart are required before live acceptance can be
   checked.
+
+### 2026-10-02 — live 1.0.793 control-plane finding and root repair
+
+- The user-owned runtime restarted on globally installed `omnius@1.0.793` at
+  `2026-10-02T08:01:52.570Z`; polling remained healthy with no replay backlog.
+  This session did not start, stop, or send through that process.
+- Installed bundle SHA-256
+  `77c9e3b8094d90a3f91db155f9e28ee0a3dfad588bb03538608f284fbc768c27`
+  contains the REQ-TDT-009 contract-migration, grounded-synthesis, and bounded-
+  review changes. The live recurrence is therefore not a stale-package
+  explanation.
+- OMNIUS message 5717 exposed a durable-continuity failure: an exact open task
+  existed, but the directly addressed completion command was admitted to chat
+  and consumed after another unsupported progress acknowledgement.
+- Case Theory message 612 exposed two control-plane mismatches. Telegram set
+  `taskObjectiveEnforcement="verified_external_effect"` for every objective,
+  so a planning-free reminder list reached `deliver_due` but could still run
+  `tool_search` and `file_read`. Terminal-review generation also allowed
+  arbitrary evidence/message IDs while the parser required the exact packet,
+  so invalid reviewer output caused completed object work to be replayed.
+- The repair now derives strict frontier enforcement for planning-free
+  read-only contracts, binds the reviewer schema to the parser's exact IDs,
+  and performs one fresh reviewer retry against the unchanged evidence before
+  any lifecycle recovery. Multi-source checklist research remains deferred.
+- Focused validation after the root repair: 250/250 tests passed across six
+  affected Telegram suites; the updated review/enforcement tests passed 14/14;
+  CLI typecheck, build, and `git diff --check` passed.
+- The installed 1.0.793 runtime does not contain this REQ-TDT-010 repair. Live
+  acceptance remains open until a rebuilt package is installed and restarted.
+
+### 2026-10-02 — live 1.0.794 scope breach and remaining OMNIUS regression
+
+- The user-owned runtime started globally installed `omnius@1.0.794`; installed
+  bundle SHA-256 was
+  `54ef7bafed127f8a19146f9400d3123bc77d3f6d92914b0d97d19538cda7f938`.
+- Case Theory message 612 recovered, observed the shared 29-reminder store, and
+  delivered Telegram message 614. Only four records belonged to Case Theory;
+  the tool result also contained three OMNIUS, one Po3 Clock, eighteen private
+  Telegram, and three private TUI records. The delivered reply exposed OMNIUS
+  and private reminder IDs/titles while showing no Case Theory record.
+- Root ownership was `ReminderTool`: `lockScope` constrained `set` but
+  `list`, `complete`, `dismiss`, and `snooze` selected from the global store.
+  The repair now applies one exact surface/target/topic authorization predicate
+  before every read and ID mutation. Foreign IDs return the ordinary not-found
+  result and remain unchanged; unlocked administration retains global access.
+- OMNIUS message 5719 (`please pull the data`) remained a separate regression.
+  Its exact typed decision identified the bot as direct addressee and cited the
+  current message, but the host boundary recognized only literal mentions and
+  Telegram reply edges. Quick chat delivered message 5720 (`pulling it now.
+  one sec.`) without an action run or data result.
+- Typed exact-current direct-turn authority now counts as bot address only when
+  a real decision ID, non-unavailable source, self actor key, addressee role,
+  direct-turn basis, and current-message evidence all agree. The exact 5719
+  replay is promoted to action, cites its open task, clears the chat plan, and
+  cannot use quick chat; an other-actor control remains unchanged.
+- Validation: 74/74 execution temporal-agency tests and 204/204 Telegram tests
+  passed across reminder scope, steering/finalization, and Bot API surfaces.
+  Execution and CLI typechecks/builds passed; `git diff --check` passed.
+- These repairs are not in installed 1.0.794. No publish, install, restart,
+  Telegram send, or live inference was performed by this session.

package/npm-shrinkwrap.json

@@ -1,12 +1,12 @@
 {
   "name": "omnius",
-  "version": "1.0.793",
+  "version": "1.0.795",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "omnius",
-      "version": "1.0.793",
+      "version": "1.0.795",
       "bundleDependencies": [
         "image-to-ascii"
       ],

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "omnius",
-  "version": "1.0.793",
+  "version": "1.0.795",
   "description": "AI coding agent powered by open-source models (Ollama/vLLM) — interactive TUI with agentic tool-calling loop",
   "type": "module",
   "main": "./dist/library.js",