taskuary 0.3.3.2__tar.gz → 0.3.3.3__tar.gz

{taskuary-0.3.3.2/taskuary.egg-info → taskuary-0.3.3.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: taskuary
-Version: 0.3.3.2
+Version: 0.3.3.3
 Summary: Automate your job - the local-first operating system for your work: AI triage in, coding agents out.
 License: MIT
 Project-URL: Homepage, https://github.com/ldbumble/taskuary
@@ -62,7 +62,7 @@ Dynamic: license-file
 # Taskuary
 
 [![CI](https://github.com/ldbumble/taskuary/actions/workflows/ci.yml/badge.svg)](https://github.com/ldbumble/taskuary/actions/workflows/ci.yml)
-[![PyPI](https://img.shields.io/pypi/v/taskuary.svg?cacheSeconds=300&release=0.3.3.2)](https://pypi.org/project/taskuary/)
+[![PyPI](https://img.shields.io/pypi/v/taskuary.svg?cacheSeconds=300&release=0.3.3.3)](https://pypi.org/project/taskuary/)
 [![Python 3.10+](https://img.shields.io/badge/python-3.10%2B-3776ab.svg)](https://github.com/ldbumble/taskuary)
 [![License: MIT](https://img.shields.io/badge/license-MIT-blue.svg)](LICENSE)
 
@@ -71,9 +71,9 @@ Dynamic: license-file
 Taskuary turns incoming messages into organized work. It sorts what matters, hands tasks to
 your agents, and brings decisions back to you. Nothing sends or ships without your approval.
 
-![Taskuary: the pipe ranked by triage beside the assistant chat - a mail pulled in with its drafted reply, the Timeline in task view, the Reports and the Board.](https://raw.githubusercontent.com/ldbumble/taskuary/master/docs/hero.gif)
+![The Taskuary Studio: work arrives, triage decides what needs action, and agents take seats to work it.](https://raw.githubusercontent.com/ldbumble/taskuary/master/docs/hero.gif?v=studio)
 
-Taskuary is early—currently **v0.3.3.2**—so breaking changes are still possible before 1.0.
+Taskuary is early—currently **v0.3.3.3**—so breaking changes are still possible before 1.0.
 
 <p align="center">
   <a href="https://taskuary.com/demo/"><img
@@ -85,16 +85,27 @@ Taskuary is early—currently **v0.3.3.2**—so breaking changes are still possi
 
 ## What Taskuary can do
 
-**One Timeline and Assistant**
+**Connect every system. Keep control.**
 
-Mail, chats, issues, alerts, and reports appear on one Timeline and in one Assistant chat.
-Open any item to see its context, status, and available actions.
+Bring your mail, chats, issue trackers, alerts, reports, and AI agents into one control center
+that runs on your machine. Every input lands on a single Timeline with its context, status, and
+available actions intact. Taskuary gives your tools one place to work together while you decide
+what can run, what can send, and what needs your attention.
 
-![Taskuary's unified work pipe and chat, populated only with fictional demo data.](https://raw.githubusercontent.com/ldbumble/taskuary/master/docs/screenshot-chat.png)
+![Taskuary's unified work pipe and chat, populated only with fictional demo data.](https://raw.githubusercontent.com/ldbumble/taskuary/master/docs/screenshot-chat.png?v=0.3.3.2)
 
-**AI triage**
+**Turn incoming work into tasks automatically**
 
-Triage files routine updates, turns real requests into tasks, and puts urgent decisions first.
+AI triage reads every incoming item, understands what it means, and turns actionable requests
+into ready-to-run tasks. Routine updates are filed, questions get draft replies, and urgent
+decisions move to the front. Every choice comes with a reason, and you can change it at any time.
+
+**Watch the work. Approve the outcome.**
+
+Taskuary keeps the system moving and brings you in when your judgment matters. The Assistant
+speaks up when a reply is waiting, a task has gone quiet, a report has failed, or a meeting is
+about to start. Each update arrives once, with the evidence and next actions attached, so you can
+review the work, make the call, and approve what happens next.
 
 **Agents in your own repos**
 
@@ -105,25 +116,25 @@ Send coding work to Claude Code, Codex, Gemini, Cursor, Copilot, or another CLI.
 Watch the terminal, answer questions, review changes, and use the built-in browser without
 losing the session.
 
-**Scheduled checks**
+**Describe a check. Let it run itself.**
 
-Watch databases, files, APIs, cloud logs, reports, or agent skills. Quiet checks stay silent.
+Say what you want watched in plain words and Taskuary builds it: read a folder of CSVs every
+morning, total the AP bills due in the next 30 days, count helpdesk tickets by day. Reports read
+and summarize, on a schedule or on demand, and can come back with an AI summary. Workflows write
+data and keep state. Quiet checks stay quiet, so the only one you hear about is the one that
+failed.
 
 **You stay in control**
 
 Replies and proposed actions wait in Review. You decide what sends, runs, closes, or gets dismissed.
 
-**Catches what slipped**
-
-Surface unanswered replies, upcoming meetings, quiet tasks, failed reports, and important changes.
-
 ## How work flows
 
 ### 1. Work arrives and gets sorted
 
 Taskuary reads connected inboxes, separates tasks from noise, and shows the result on one Timeline.
 
-![The Timeline in task view: every message with its verdict, and the drafted reply waiting on you](https://raw.githubusercontent.com/ldbumble/taskuary/master/docs/screenshot-timeline-crop.png)
+![The Timeline in task view: every message with its verdict, and the drafted reply waiting on you](https://raw.githubusercontent.com/ldbumble/taskuary/master/docs/screenshot-timeline-crop.png?v=0.3.3.2)
 
 ### 2. An agent works on it
 
@@ -135,7 +146,7 @@ Tasks move from **Queued** to **Agent working**. The Board shows live progress a
 
 Finished work, draft replies, questions, and loose ends return for review. You decide what to send, change, snooze, or dismiss.
 
-![Taskuary assistant showing loose ends and suggested actions](https://raw.githubusercontent.com/ldbumble/taskuary/master/docs/screenshot-assistant.png)
+![The assistant walking the pipe: the item on the table, its drafted reply, and Approve & send waiting on you](https://raw.githubusercontent.com/ldbumble/taskuary/master/docs/screenshot-assistant.png?v=0.3.3.2)
 
 ## Install
 

{taskuary-0.3.3.2 → taskuary-0.3.3.3}/README.md

@@ -1,7 +1,7 @@
 # Taskuary
 
 [![CI](https://github.com/ldbumble/taskuary/actions/workflows/ci.yml/badge.svg)](https://github.com/ldbumble/taskuary/actions/workflows/ci.yml)
-[![PyPI](https://img.shields.io/pypi/v/taskuary.svg?cacheSeconds=300&release=0.3.3.2)](https://pypi.org/project/taskuary/)
+[![PyPI](https://img.shields.io/pypi/v/taskuary.svg?cacheSeconds=300&release=0.3.3.3)](https://pypi.org/project/taskuary/)
 [![Python 3.10+](https://img.shields.io/badge/python-3.10%2B-3776ab.svg)](https://github.com/ldbumble/taskuary)
 [![License: MIT](https://img.shields.io/badge/license-MIT-blue.svg)](LICENSE)
 
@@ -10,9 +10,9 @@
 Taskuary turns incoming messages into organized work. It sorts what matters, hands tasks to
 your agents, and brings decisions back to you. Nothing sends or ships without your approval.
 
-![Taskuary: the pipe ranked by triage beside the assistant chat - a mail pulled in with its drafted reply, the Timeline in task view, the Reports and the Board.](https://raw.githubusercontent.com/ldbumble/taskuary/master/docs/hero.gif)
+![The Taskuary Studio: work arrives, triage decides what needs action, and agents take seats to work it.](https://raw.githubusercontent.com/ldbumble/taskuary/master/docs/hero.gif?v=studio)
 
-Taskuary is early—currently **v0.3.3.2**—so breaking changes are still possible before 1.0.
+Taskuary is early—currently **v0.3.3.3**—so breaking changes are still possible before 1.0.
 
 <p align="center">
   <a href="https://taskuary.com/demo/"><img
@@ -24,16 +24,27 @@ Taskuary is early—currently **v0.3.3.2**—so breaking changes are still possi
 
 ## What Taskuary can do
 
-**One Timeline and Assistant**
+**Connect every system. Keep control.**
 
-Mail, chats, issues, alerts, and reports appear on one Timeline and in one Assistant chat.
-Open any item to see its context, status, and available actions.
+Bring your mail, chats, issue trackers, alerts, reports, and AI agents into one control center
+that runs on your machine. Every input lands on a single Timeline with its context, status, and
+available actions intact. Taskuary gives your tools one place to work together while you decide
+what can run, what can send, and what needs your attention.
 
-![Taskuary's unified work pipe and chat, populated only with fictional demo data.](https://raw.githubusercontent.com/ldbumble/taskuary/master/docs/screenshot-chat.png)
+![Taskuary's unified work pipe and chat, populated only with fictional demo data.](https://raw.githubusercontent.com/ldbumble/taskuary/master/docs/screenshot-chat.png?v=0.3.3.2)
 
-**AI triage**
+**Turn incoming work into tasks automatically**
 
-Triage files routine updates, turns real requests into tasks, and puts urgent decisions first.
+AI triage reads every incoming item, understands what it means, and turns actionable requests
+into ready-to-run tasks. Routine updates are filed, questions get draft replies, and urgent
+decisions move to the front. Every choice comes with a reason, and you can change it at any time.
+
+**Watch the work. Approve the outcome.**
+
+Taskuary keeps the system moving and brings you in when your judgment matters. The Assistant
+speaks up when a reply is waiting, a task has gone quiet, a report has failed, or a meeting is
+about to start. Each update arrives once, with the evidence and next actions attached, so you can
+review the work, make the call, and approve what happens next.
 
 **Agents in your own repos**
 
@@ -44,25 +55,25 @@ Send coding work to Claude Code, Codex, Gemini, Cursor, Copilot, or another CLI.
 Watch the terminal, answer questions, review changes, and use the built-in browser without
 losing the session.
 
-**Scheduled checks**
+**Describe a check. Let it run itself.**
 
-Watch databases, files, APIs, cloud logs, reports, or agent skills. Quiet checks stay silent.
+Say what you want watched in plain words and Taskuary builds it: read a folder of CSVs every
+morning, total the AP bills due in the next 30 days, count helpdesk tickets by day. Reports read
+and summarize, on a schedule or on demand, and can come back with an AI summary. Workflows write
+data and keep state. Quiet checks stay quiet, so the only one you hear about is the one that
+failed.
 
 **You stay in control**
 
 Replies and proposed actions wait in Review. You decide what sends, runs, closes, or gets dismissed.
 
-**Catches what slipped**
-
-Surface unanswered replies, upcoming meetings, quiet tasks, failed reports, and important changes.
-
 ## How work flows
 
 ### 1. Work arrives and gets sorted
 
 Taskuary reads connected inboxes, separates tasks from noise, and shows the result on one Timeline.
 
-![The Timeline in task view: every message with its verdict, and the drafted reply waiting on you](https://raw.githubusercontent.com/ldbumble/taskuary/master/docs/screenshot-timeline-crop.png)
+![The Timeline in task view: every message with its verdict, and the drafted reply waiting on you](https://raw.githubusercontent.com/ldbumble/taskuary/master/docs/screenshot-timeline-crop.png?v=0.3.3.2)
 
 ### 2. An agent works on it
 
@@ -74,7 +85,7 @@ Tasks move from **Queued** to **Agent working**. The Board shows live progress a
 
 Finished work, draft replies, questions, and loose ends return for review. You decide what to send, change, snooze, or dismiss.
 
-![Taskuary assistant showing loose ends and suggested actions](https://raw.githubusercontent.com/ldbumble/taskuary/master/docs/screenshot-assistant.png)
+![The assistant walking the pipe: the item on the table, its drafted reply, and Approve & send waiting on you](https://raw.githubusercontent.com/ldbumble/taskuary/master/docs/screenshot-assistant.png?v=0.3.3.2)
 
 ## Install
 

{taskuary-0.3.3.2 → taskuary-0.3.3.3}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "taskuary"
-version = "0.3.3.2"
+version = "0.3.3.3"
 description = "Automate your job - the local-first operating system for your work: AI triage in, coding agents out."
 readme = "README.md"
 license = { text = "MIT" }

{taskuary-0.3.3.2 → taskuary-0.3.3.3}/taskuary/assistant.py

@@ -537,7 +537,7 @@ def raised(store, days: float = 2) -> str:
     return '\n'.join(out) or '(the assistant raised nothing in this window)'
 
 
-def parse(text: str, cands: list, max_lines: int = MAX_LINES) -> list:
+def parse(store, text: str, cands: list, max_lines: int = MAX_LINES) -> list:
     """The model's list, kept honest: a key it invents must be idea:*, a candidate key keeps its
     kind and its buttons, and the text is the model's when it gave one. Every line keeps its WHY -
     the hub's facts for a candidate (plus the model's read on them), the model's own for an idea -
@@ -767,7 +767,7 @@ def think(store, cands: list, llm, instruction: str = None, max_lines: int = MAX
         from .llm import readable_images
         images = readable_images(store, _people_context(store)[1])
     text = llm(system, user, max_tokens=POST_TOKENS, **({'images': images} if images else {}))
-    return parse(text, cands, max_lines), _notes(text), user
+    return parse(store, text, cands, max_lines), _notes(text), user
 
 
 def facts(store, watch_source_ids=None, watch_sources=None, systems_only: bool = False) -> str:

{taskuary-0.3.3.2 → taskuary-0.3.3.3}/taskuary/clis.py

@@ -54,9 +54,19 @@ READONLY = {'claude': (('--dangerously-skip-permissions',), ('--tools', '')),
 # classifier above, which gets no tools at all because the text it classifies is untrusted input.
 # Claude's built-ins are named explicitly, and every MCP tool is denied because MCP permissions
 # are separate from --tools and a connector may expose writes beside reads.
+REPORT_TOOLS = 'Read,Glob,Grep,WebFetch,WebSearch'
+
+# --tools and --allowedTools answer different questions and a report needs BOTH: --tools says which
+# built-ins EXIST in this run, --allowedTools says which may be used without asking a human. With
+# the bypass flag dropped and only --tools given, WebFetch and WebSearch existed but still prompted,
+# and a headless report has nobody to click: every fetch came back "Claude requested permissions to
+# use WebFetch, but you haven't granted it yet" and the run narrated the refusal instead of the
+# trending page. Read/Glob/Grep never prompt, which is why the gap stayed invisible for two months
+# (the GitHub Trending report, 2026-09-04). Granting the same five is not a widening - the set the
+# run may use is still exactly the set that exists.
 REPORT_READ = {
     'claude': (('--dangerously-skip-permissions',),
-               ('--tools', 'Read,Glob,Grep,WebFetch,WebSearch', '--disallowedTools', 'mcp__*')),
+               ('--tools', REPORT_TOOLS, '--allowedTools', REPORT_TOOLS, '--disallowedTools', 'mcp__*')),
     'codex': READONLY['codex'],
     'gemini': READONLY['gemini'],
 }

{taskuary-0.3.3.2 → taskuary-0.3.3.3}/taskuary/concierge.py

@@ -81,8 +81,13 @@ _SAYS = (
     (re.compile(r"\b(not (my|our) (issue|problem|job|task|thing)|not (ours|mine)|nothing to do with (me|us)|let them (handle|deal|sort|do)"
                 r"|ignore (it|this)|leave it (alone|be)|drop it|no need to (respond|reply|answer))\b", re.I), 'not_ours'),
     (re.compile(r"\b(never|always) (again|file|ignore)|remember (this|that)|from now on\b", re.I), 'not_ours_remember'),
+    # "research" came out of this list: the other verbs here read as DIAGNOSING something that
+    # exists ("look into that server", "find out why the export drops rows"), but research is the
+    # word for reading about the WORLD - "can you research the Factor Elara gravel bike" opened a
+    # coding agent on a checkout (the owner, 2026-09-04). Left off the list it reaches the brain,
+    # which has the sentence and can tell a product from a production database.
     (re.compile(r"\b(send (it |this |that )?to (the )?(coding agent|coder|codex|claude|gemini|agent)|have the (coder|agent)|let the (coder|agent)|start the (coder|agent)|code it|(coder|codex|claude) (should|can) (review|look|fix|handle)"
-                r"|look into|investigate|dig into|find out|figure out|research|check (on |out )?(that|the|this|what|why|if)|ask (the |an )?(assistant|agent|coder) to)\b", re.I), 'coder'),
+                r"|look into|investigate|dig into|find out|figure out|check (on |out )?(that|the|this|what|why|if)|ask (the |an )?(assistant|agent|coder) to)\b", re.I), 'coder'),
     (re.compile(r"\b(i'?ll (do|take|handle) (it|this|that)|i will (do|take|handle) (it|this|that)|mine|leave it (to|with) me|(just )?make it a task|create a task|(add|put) (it|this) (on|to) my list)\b", re.I), 'mine'),
     # ENDING AN AGENT is not closing a task: "close the agent working" closed the item on the table
     # instead - a task the owner had not even asked about (2026-09-03). It comes first, so the words
@@ -137,6 +142,7 @@ SYSTEM = (
     "(hand it to the coding agent - put EVERYTHING the owner wants done after a colon, in their words: DECIDE: coder: find out "
     "why the bulk-approve fix from before did not stick, and create an admin login for X), mine (they will do it themselves), "
     "approve (send the drafted reply as it stands), not_ours_sender (the SENDER is noise - file everything from them from now on), "
+    "setup (reading, thinking or research with NO system to type at - Taskuary opens it as a conversation with the assistant: nothing is built, no repository is touched, and it can be handed to the coding agent later if it turns out something has to be), "
     "split (this arrival is TWO jobs - Taskuary breaks it in two, each with its own ref), "
     "setting (what they want is a SWITCH - Taskuary puts it in front of them to approve and never changes it itself), "
     "stop_agent (end the AGENT that is running - stopping a session is not closing a task, and never guess which: "
@@ -149,6 +155,28 @@ SYSTEM = (
     "have it wrong is the one thing you must not shrug off.\n"
     "A question or a remark is not a decision: answer it and write no DECIDE line. A polite request is "
     "not a question: \"can you look into that server\" is a hand-off, so decide it.\n"
+    # "can you research the Factor Elara gravel bike" opened a coding agent on a checkout (the
+    # owner, 2026-09-04), and before that a set-up did the same ("it doesn't need coding agent
+    # just a regular agent that will walk me through it"). The example above primes coder; this
+    # is the line that stops it generalising to everything phrased as a favour.
+    '...but coder and setup are not the same road, and the test is whether there is a SYSTEM to '
+    'type at. A repository, a server, a database, a query, a file, an error, a failing report: '
+    'coder. Reading about the world, comparing products, weighing an option, working out what to '
+    'ask, anything whose answer is a judgement rather than a change: setup. Never send reading '
+    'work to the coding agent because the sentence was polite - "can you research X" is a '
+    'walk-through, not a hand-off.\n'
+    # "ignore" names the act and leaves out the part that lasts. Filing one mail and silencing a
+    # sender for ever are different acts, and the owner asked to be consulted on which (2026-09-04:
+    # "even without the button when you say ignore unless you say specific just for today it
+    # should ask what it should do with it in terms of memory").
+    '"Ignore it" and "not ours" name the act but not its SCOPE, and scope is the part that lasts. '
+    'Unless they already said which - "just this once", "just for today", "never again", "always", '
+    '"from this sender" - do not pick one: say in one line that you can file this one, remember the '
+    'kind, or silence the sender, and end with exactly OPTIONS: just this once | this kind from now '
+    'on | everything from this sender. This is the one ambiguity worth a question, because the wide '
+    'answers are the ones that quietly hide mail later. When they HAVE said which, decide it and '
+    'write no options line: this once is not_ours, the kind is not_ours_remember, the sender is '
+    'not_ours_sender.\n'
     "The thread you are given is the whole thread, the owner's own sent mail included. When it shows they "
     "already answered, say so as a fact. Only when the thread has no answer from them may you say the mail "
     "has not been read back yet - and then name the Sync button, never blame yourself for not seeing it.")
@@ -617,6 +645,18 @@ def parse_options(text: str) -> tuple[str, list]:
     return (text[:m.start()].strip(), opts if len(opts) >= 2 else [])
 
 
+def _verdict_why(item: dict) -> str:
+    """Triage's own reason, with the verdict word it opens with trimmed off - "triage filed it as
+    fyi - triage: fyi - an automated notification" says it twice (the owner, 2026-09-04: "ask we
+    processed as fyi because of x")."""
+    w = (item.get('why') or '').strip()
+    for pre in ('triage: fyi -', 'triage: fyi', 'triage:'):
+        if w.lower().startswith(pre):
+            w = w[len(pre):].lstrip(' -·')
+            break
+    return f' - {w}' if w else ''
+
+
 def fallback(item: dict | None, opening: bool, pile_items: list = None) -> str:
     """No model: the facts in the same three beats - where from, what was done, what you need to do."""
     if not item:
@@ -632,9 +672,11 @@ def fallback(item: dict | None, opening: bool, pile_items: list = None) -> str:
         frm = f"{item.get('who') or 'Someone'} wrote on {item.get('channel') or 'email'}" + (f" ({funnel_age(item)})" if funnel_age(item) else '') + f": \"{item['title']}\""
         done = (f"the agent {item['summary']}" if item.get('summary') else
                 'triage judged it a reply to write' if item['kind'] == 'review' else 'an agent proposed an action' if item['kind'] == 'action' else
-                'a coding task with no agent on it yet' if item.get('coding') else 'nothing has been done with it yet' if item['kind'] in ('asked', 'todo') else 'triage filed it as fyi')
+                'a coding task with no agent on it yet' if item.get('coding') else 'nothing has been done with it yet' if item['kind'] in ('asked', 'todo')
+                else f'triage filed it as fyi{_verdict_why(item)}')
         need = ('approve the draft below, or redraft it' if item['kind'] == 'review' else 'say whether it may run' if item['kind'] == 'action' else
-                'reply, hand it to the coding agent, or say it is not ours' if item['kind'] in ('asked', 'todo') else 'nothing - read it if you like')
+                'reply, hand it to the coding agent, or say it is not ours' if item['kind'] in ('asked', 'todo')
+                else 'nothing has to happen - make it a task, tell me to ignore this sender, or move on')
         return f"{frm}. Since then: {done}. From you: {need}."
     if item['kind'] == 'agent':
         return f"{item.get('agent') or 'An agent'} on {item.get('ref') or item['title']} " + (f"asked: {item['tail'][-1]}" if item.get('asking') and item.get('tail') else 'stopped at its prompt') + ' - answer it below.'
@@ -644,7 +686,7 @@ def fallback(item: dict | None, opening: bool, pile_items: list = None) -> str:
         return f"{item['title']}" + (f" ({item.get('who')})" if item.get('who') else '') + f" - {item.get('why') or 'the assistant raised this'}. Draft the follow-up, make it a task, or let it go."
     if item['kind'] == 'meeting': return f"{item['title']} is {item.get('why')}. Prep me, or move on."
     if item['kind'] == 'report':
-        return f"{item['title']} landed {funnel_age(item)}" + (' and FAILED - the cause is in it.' if item.get('bad') else '.') + ' Read it with the button, or move on.'
+        return f"{item['title']} landed {funnel_age(item)}" + (' and FAILED - the cause is in it.' if item.get('bad') else '.') + ' It is open below - run it again, or move on.'
     if item['kind'] == 'agentdone': return f"{item.get('who') or 'The agent'} finished {item.get('ref') or item['title']}" + (f": {item['summary']}" if item.get('summary') else '.') + ' Want to see the final report?'
     lead = {'agent': f"{item.get('agent') or 'An agent'} is waiting on you on {item.get('ref') or item['title']}.",
             'meeting': f"{item['title']} is {item.get('why')}.",

{taskuary-0.3.3.2 → taskuary-0.3.3.3}/taskuary/funnel.py

@@ -634,6 +634,20 @@ def came_in(i: dict) -> bool:
 FYI_BATCH = 4                              # fyi comes as a handful, not one at a time - there is nothing to do with each
 BLOCKED_AGAIN_MIN = 30                     # an agent still waiting comes out again this long after it was shown
 
+def _not_yet(i: dict) -> bool:
+    """On the timeline, but nothing to say about it YET - the walk skips it and comes back.
+
+    A meeting enters the pipe two hours out (SOON_MIN) so the day is visible, and its lane is
+    'time', which puts it at the very front of the walk. So the assistant opened with a meeting
+    that was still an hour and a half away, ahead of mail that wanted answering now. It is the same
+    shape as an agent that is mid-run: real, on the board, and not yours to do anything about until
+    it stops (the owner, 2026-09-04: "meetings should not go down into the chat until 15 minutes
+    before like a agent in middel of working"). alerts() already drew this line at ALERT_MIN; the
+    walk did not.
+    """
+    return i['kind'] == 'meeting' and i.get('mins') is not None and i['mins'] > ALERT_MIN
+
+
 def next_item(store, key: str = None, only: str = None) -> dict | None:
     """What comes out of the mouth: the named item (read or not - the chat may return to it), or
     the first unread one - of the mail alone when `only` is 'mail'. Something still being triaged is
@@ -642,6 +656,7 @@ def next_item(store, key: str = None, only: str = None) -> dict | None:
     if key: return next((i for i in build(store, keep_surfaced=True)['items'] if i['key'] == key), None) or batch_item(store, key)
     again = (datetime.now() - timedelta(minutes=BLOCKED_AGAIN_MIN)).strftime('%Y-%m-%d %H:%M:%S')
     ready = [i for i in pile(store, force=True)['items'] if not i.get('settling') and i['lane'] != 'working'
+             and not _not_yet(i)
              and (not i.get('surfaced') or _ts(i.get('surfaced_at')) <= again)]
     if only == 'mail': ready = [i for i in ready if came_in(i) or i['kind'] in INTERRUPTS]
     return ready[0] if ready else None

{taskuary-0.3.3.2 → taskuary-0.3.3.3}/taskuary/general.py

@@ -172,6 +172,25 @@ def dock_task(store, actor='owner') -> tuple[dict, bool]:
     return store.get_task(tid), True
 
 
+def retire_dock(store, actor='owner'):
+    """Close the current dock conversation so the next caller is handed a fresh one - what the
+    'New chat' button does, minus the live session there is none of in a just-started process.
+
+    A relaunch should open a NEW chat, not resume yesterday's (the owner, 2026-09-04: "after you
+    restart a new chat should be there not same chat"): the dock task id lives in settings, so
+    without this the same conversation - and the same model session - came back forever. A dock
+    nobody said anything in is left alone, or every launch would archive an empty guide task."""
+    raw = store.get_settings().get('assistant_dock_task_id')
+    task = store.get_task(int(raw)) if str(raw or '').isdigit() else None
+    if not (task and is_dock(task) and task.get('Status') not in ('done', 'dropped')): return None
+    if not history(store, task['TaskId']): return None
+    from . import concierge
+    store.update_task(task['TaskId'], {'Status': 'done'}, actor)
+    store.audit('task', task['TaskId'], 'archive_assistant_dock', actor)
+    store.set_setting(f"{concierge.SID_KEY}:{task['TaskId']}", '', actor)
+    return task['TaskId']
+
+
 def dock_snapshot(store) -> str:
     """A compact, live map of the surfaces the hovering assistant is meant to explain.
 

{taskuary-0.3.3.2 → taskuary-0.3.3.3}/taskuary/interview.py

@@ -17,13 +17,15 @@ SKILL_FILE = Path(__file__).parent / 'skills' / 'soul-interview' / 'SKILL.md'
 _LEGACY_LABELS = {
     'who': 'Who are you, and what is your work?',
     'work': 'What kind of work reaches you?',
-    'task': 'What may the assistant handle without asking?',
-    'never': 'What must never happen without you?',
+    'task': 'What kind of work is worth turning into a task?',
+    'never': 'Which ambiguities or sensitive situations need your judgment?',
     'people': 'Which people and relationships matter?',
     'systems': 'Which systems or sources of truth matter?',
     'voice': 'How should communication sound?',
 }
 
+_MANAGED_HEADINGS = ('## Connected systems', '## Project relationships', '## Repository map')
+
 
 def skill_text() -> str:
     """The portable interview contract used for both questioning and writing."""
@@ -45,10 +47,35 @@ def _known(ctx: dict) -> str:
     bits = [f"Owner name on file: {ctx['owner'] or '(not set)'}",
             f"Connected sources: {', '.join(ctx['channels']) or 'none yet'}",
             f"Known repositories (mention only if relevant): {', '.join(ctx['repos']) or 'none'}",
-            f"Frequent correspondents: {'; '.join(ctx['writes_most']) or 'nothing ingested yet'}"]
+            f"Frequent correspondents (activity only; not proof of importance or authority): "
+            f"{'; '.join(ctx['writes_most']) or 'nothing ingested yet'}"]
     return 'WHAT TASKUARY ALREADY KNOWS (facts; do not contradict them):\n' + '\n'.join(bits)
 
 
+def _section(text: str, heading: str) -> str:
+    """One level-two section, through the next level-two heading or end of document."""
+    match = re.search(rf'(?ms)^{re.escape(heading)}[^\n]*\n.*?(?=^##\s|\Z)', str(text or ''))
+    return match.group(0).rstrip() if match else ''
+
+
+def _without_section(text: str, heading: str) -> str:
+    return re.sub(rf'(?ms)^\s*{re.escape(heading)}[^\n]*\n.*?(?=^##\s|\Z)', '', str(text or '')).strip()
+
+
+def _carry_managed_sections(current: str, drafted: str) -> str:
+    """Keep machine-maintained operational memory out of the model's authorship.
+
+    The interview rewrites the owner's prose. Connected systems and repository notes are updated
+    elsewhere, while project/person edges live in SQLite and are rendered after the save. A model
+    draft must therefore be unable to erase or fabricate any of those sections.
+    """
+    body = str(drafted or '')
+    for heading in _MANAGED_HEADINGS:
+        body = _without_section(body, heading)
+    carried = [_section(current, heading) for heading in ('## Connected systems', '## Repository map')]
+    return '\n\n'.join([body.rstrip()] + [block for block in carried if block]).strip()
+
+
 def _answers(value) -> list[dict]:
     """Normalize adaptive transcripts and the previous seven-field request shape."""
     if isinstance(value, list):
@@ -146,8 +173,14 @@ def draft(store, answers, llm=None) -> str:
 
 def write(store, answers, actor: str = 'owner', llm=None) -> str:
     """Draft, save, and audit an owner-controlled SOUL.md."""
-    body = draft(store, answers, llm)
+    body = _carry_managed_sections(store.get_doc('soul') or '', draft(store, answers, llm))
     store.save_doc('soul', body, actor)
+    # Structured project/person memory is not interview prose. Re-render it from the database
+    # after every interview, and refresh the carried connection block from current configuration.
+    from .docsync import sync_connections, sync_projects
+    sync_connections(store, actor)
+    sync_projects(store, actor)
+    body = store.get_doc('soul') or body
     transcript = _answers(answers)
     store.audit('doc', 0, 'soul_interview', actor,
                 detail={'answered': sum(bool(row['a']) for row in transcript),

{taskuary-0.3.3.2 → taskuary-0.3.3.3}/taskuary/messengers.py

@@ -201,7 +201,8 @@ def wa_status(c) -> dict:
     jid = str(st.get('jid') or '')
     out = {'connected': bool(st.get('connected')), 'me': st.get('me') or '', 'jid': jid,
            'phone': ('+' + jid.split('@')[0].split(':')[0]) if jid and jid.split(':')[0].split('@')[0].isdigit() else '',   # the paired number, from the account jid
-           'pairing_code': st.get('pairingCode') or '', 'qr_svg': ''}
+           'pairing_code': st.get('pairingCode') or '', 'qr_svg': '',
+           'reconnect': st.get('reconnect') or {}, 'filter': st.get('filter') or {}}
     if st.get('qr') and not out['connected']:
         import segno
         out['qr_svg'] = segno.make(st['qr'], error='m').svg_data_uri(scale=5, border=2, dark='#1e1e2e', light='#ffffff')
@@ -222,6 +223,13 @@ def wa_chats(c) -> list:
         r['n'] += 1
         if not m.get('fromMe') and m.get('name'): r['name'] = m['name']     # the other side's push name, never ours
         if (m.get('ts') or 0) >= r['last']: r['last'], r['snippet'] = m.get('ts') or 0, (m.get('text') or '')[:80]
+    # The bridge sees the routing JID before deciding whether to decrypt. That lets the source
+    # picker show a newly active chat without retaining its sender, text, or media first.
+    for m in out.get('blockedChats', []):
+        jid = m.get('jid') or ''
+        if not jid or jid.endswith('@broadcast') or jid in by: continue
+        by[jid] = {'jid': jid, 'group': bool(m.get('group')), 'name': '', 'n': 0,
+                   'last': m.get('last') or 0, 'snippet': 'not opened - chat is not authorized'}
     rows = sorted(by.values(), key=lambda r: -r['last'])
     for r in rows: r['last'] = datetime.fromtimestamp(r['last']).strftime('%Y-%m-%d %H:%M') if r['last'] else ''
     return rows
@@ -254,6 +262,13 @@ def poll_whatsapp(store, c, sources: list, llm=None, file_only=False) -> int:
     srcs = [s for s in sources if s.get('Channel', 'whatsapp') == 'whatsapp' and s.get('Address')]
     star = any(s['Address'] == '*' for s in srcs)
     want = {s['Address'] for s in srcs if s['Address'] != '*'}
+    notify_chat = str(cfg.get('notify_chat') or '').strip()
+    bridge_want = want | ({notify_chat} if notify_chat else set())  # approvals/assistant may not be an inbound source
+    # Filtering here alone was too late: Baileys had already decrypted and downloaded media from
+    # every chat. Give the same policy to its pre-decryption hook before asking for messages.
+    try: _wa(c, '/filter', {'allDirect': star, 'jids': sorted(bridge_want)})
+    except RuntimeError as e:
+        if '(404)' not in str(e): raise                     # an older detached bridge: keep polling until its next restart
     out = _wa(c, f"/messages?after={int(cfg.get('wa_seq') or 0)}")
     n, took = 0, []
     from . import phone, remote_assistant

{taskuary-0.3.3.2 → taskuary-0.3.3.3}/taskuary/reports.py

@@ -750,10 +750,16 @@ CONNECTION_OF = {'mssql': mssql_connection, 'winrm': winrm_connection, 'database
                  'google_sheets': _sheets_connection}
 
 
+# Report types whose data IS the store - they reach no further than the local database, so they
+# can neither be slow nor unreachable. Everything else dials out: a SQL box, an API, a share.
+# run_due_reports runs these FIRST for that reason, so keep the two uses of this list together.
+STORE_BACKED = ('digest', 'evening_inbox', 'automate', 'assistant', 'agent', 'calendar', 'kb_search', 'kb_reindex',
+                'metric', 'metric_check', 'handbook_search', 'handbook_write', 'handbook_vote',
+                'hub_search', 'hub_write', 'hub_vote', 'hub_comment')
+
+
 def resolve_cfg(store, cfg: dict) -> dict:
-    if cfg.get('type') in ('digest', 'evening_inbox', 'automate', 'assistant', 'agent', 'calendar', 'kb_search', 'kb_reindex',
-                           'metric', 'metric_check', 'handbook_search', 'handbook_write', 'handbook_vote',
-                           'hub_search', 'hub_write', 'hub_vote', 'hub_comment'):
+    if cfg.get('type') in STORE_BACKED:
         return {**cfg, 'store': store}   # their data IS the store (the agent's: its profile; the calendar's: the cards; the knowledge base: its index)
     conn = CONNECTION_OF.get(cfg.get('type'))
     if conn:
@@ -1239,15 +1245,41 @@ def deliver_report(store, src: dict, cfg: dict, subject: str, body: str) -> dict
     return {'gate': 'auto', 'message_id': mid, 'sent': sent}
 
 
+def _own_data(src: dict) -> bool:
+    """Does this report read nothing but the store? An unknown type counts as dialling out, which
+    is the conservative half of the split - it waits behind the local ones rather than ahead."""
+    try: return (json.loads(src.get('ConfigJson') or '{}').get('type') or 'rest') in STORE_BACKED
+    except ValueError: return False
+
+
 def run_due_reports(store, startup: bool = False) -> int:
+    """Every due report, own-data ones FIRST and each one walled off from the others.
+
+    On a startup catch-up these run one after another, so a source that is merely unreachable used
+    to hold up everything behind it: a SQL Server that was not there spent 41 s in an ODBC login
+    timeout, and the owner's Morning digest and Assistant post - which read nothing but Taskuary's
+    own database - waited it out (the owner, 2026-09-04: the first item took ~3 minutes). Ordering
+    the local ones first means what they actually read lands while the dial-out ones take their time.
+
+    And one report may not cost the others their run. A failing executor files a FAILED row and
+    returns, but a runner that RAISES used to abort the whole loop, silently skipping every report
+    after it and leaving them un-touched so the next poll started over.
+    """
     from .llm import build_llm
     try: llm = build_llm(store)
     except Exception: llm = None
+    due = [s for s in store.list_sources() if s['Channel'] == 'report'
+           and is_due(json.loads(s.get('ConfigJson') or '{}'), s.get('LastPolledAt'), startup)]
     n = 0
-    for src in store.list_sources():
-        if src['Channel'] != 'report': continue
-        if is_due(json.loads(src.get('ConfigJson') or '{}'), src.get('LastPolledAt'), startup):
+    for src in sorted(due, key=lambda s: not _own_data(s)):
+        try:
             run_report_source(store, src, llm)
-            store.touch_source(src['SourceId'])
             n += 1
+        except Exception as e:
+            # touched anyway: a report that blew up must wait for its next slot like any other,
+            # or a poll every few minutes re-runs the broken one for ever
+            logger.warning(f"report {src.get('Address')} raised, skipping it and running the rest - {e}")
+        finally:
+            try: store.touch_source(src['SourceId'])
+            except Exception as e: logger.warning(f"could not touch source {src['SourceId']}: {e}")
     return n