@aitne-sh/aitne 0.1.7 → 0.1.8

package/agent-assets/docs/concepts/safety-model.md

@@ -33,7 +33,7 @@ ask_examples:
   - Where do I see what the agent has been doing?
 locale: en-US
 created: 2026-04-26
-updated: 2026-04-26
+updated: 2026-05-17
 keywords:
   - deniedTools
   - safety floor
@@ -121,21 +121,30 @@ of "report to me" events. Information about what the agent did is
 - **`matchToolPattern`** — pattern matcher used everywhere `deniedTools`
   is enforced. Exact match (`send_email`), prefix glob (`send_*`), or
   bare `*`.
-- **`/api/integrations/:key/invoke`** — the cross-backend chokepoint.
-  Enforces `deniedTools` server-side before spawning a subprocess.
+- **`/api/integrations/:key/exec`** — the cross-backend chokepoint
+  (task mode; the RPC-style `/invoke` route was retired 2026-05-01,
+  see `docs/design/17-delegated-mode-v2.md` §4.2). Enforces `deniedTools`
+  server-side by filtering the integration's `capabilityTools` through
+  the deny list before spawning the delegated backend, so the
+  task-mode planner can only pick from the allowed surface. A
+  fully-denied surface short-circuits with `errorClass: "denied_tool"`
+  before any subprocess spawn.
 - **`agent_actions`** — SQLite table of every agent action. Direct +
-  cross-backend rows are full-fidelity. Same-backend native MCP rolls
-  up to `mcp_tool_calls` + the parent session row.
+  cross-backend rows are full-fidelity (current cross-backend writes
+  emit `delegated_task.run` / `delegated_task.exec` /
+  `delegated_task.tool_step`; legacy rows from before 2026-05-01 carry
+  `delegated_proxy.invoke`). Same-backend native MCP rolls up to
+  `mcp_tool_calls` + the parent session row.
 
 ## Where the Defenses Apply
 
 | Path | Enforcement |
 |---|---|
 | Direct mode (`/api/mail/*`, `/api/calendar/*`) | Route handler middleware checks `deniedTools` against the materialized skill body's `allowed-tools` list (frontmatter). |
-| Cross-backend (`/api/integrations/:key/invoke`) | Endpoint handler runs `matchToolPattern(deniedPattern, tool)` before spawning. Returns `403 denied_tool` on hit. |
-| Same-backend native MCP — Claude | `collectSessionDeniedTools` merges the deny patterns into the SDK's `disallowedTools` array at `query()` time. |
-| Same-backend native MCP — Gemini | Patterns are folded into `generateAdminPolicy`'s TOML deny rules (priority 1000). |
-| Same-backend native MCP — Codex | **Prose-only.** Codex bundles its connector apps into the binary; there is no per-tool deny config and the workspace-write sandbox does not match MCP tool calls. Skill prose lists the denied tools explicitly. Operators who require strict deny on Gmail / Calendar should pick a non-Codex DM backend or run those integrations through the proxy. |
+| Cross-backend (`/api/integrations/:key/exec`) | Invoker filters the integration's `capabilityTools` through `deniedTools` before spawning the delegated backend so the task-mode planner can only pick from the allowed surface. A fully-denied surface short-circuits with `errorClass: "denied_tool"`; individual tool denials surface as the same error from the invoker's `resolveAllowedToolPatterns`. |
+| Same-backend / native MCP — Claude | `collectSessionDeniedTools` merges the deny patterns into the SDK's `disallowedTools` array at `query()` time. Same code path covers both delegated same-backend and native — they share the in-session MCP surface. |
+| Same-backend / native MCP — Gemini | Patterns are folded into `generateAdminPolicy`'s TOML deny rules (priority 1000). |
+| Same-backend / native MCP — Codex | **Prose-only.** Codex bundles its connector apps into the binary; there is no per-tool deny config and the workspace-write sandbox does not match MCP tool calls. Skill prose lists the denied tools explicitly. Operators who require strict deny on Gmail / Calendar should pick a non-Codex DM backend or route those integrations through `delegated` cross-backend mode (which IS deny-enforced at `/exec`). |
 
 ## Recommended Starter Denylists
 
@@ -211,18 +220,26 @@ When the operator asks "what did you do yesterday?" / "have you sent
 anything from Gmail this week?" the agent calls:
 
 ```bash
-curl 'http://localhost:8321/api/agent/actions?since=2026-04-25T00:00:00Z&kind=delegated_proxy.invoke&limit=50'
+curl 'http://localhost:8321/api/agent/actions?since=2026-04-25T00:00:00Z&kind=delegated_task.run&kind=delegated_task.tool_step&limit=50'
 ```
 
 and answers in conversation. The endpoint:
 
 - Lives at `Autonomous` tier — the agent reads only its own audit
   trail, no operator data.
-- Accepts `since`, `kind`, `limit` (default 50, max 200).
+- Accepts `since`, `kind` (repeat for multiple values, e.g.
+  `?kind=a&kind=b`), `limit` (default 50, max 200).
 - Redacts values via the standard secret-redaction utility before
   serializing.
 - Returns rows from `agent_actions`, optionally joined with
-  `mcp_tool_calls` when `kind=mcp` (same-backend native MCP).
+  `mcp_tool_calls` when `kind=mcp` (same-backend / native MCP).
+
+Common `kind` values for the cross-backend proxy: `delegated_task.run`
+(one row per `/exec` call), `delegated_task.exec` (the planner's
+chosen tool), `delegated_task.tool_step` (each individual tool call
+inside the task). The legacy `delegated_proxy.invoke` rows persist
+from before 2026-05-01 — include them if the `since` window crosses
+that date.
 
 This **replaces** the rejected daily-digest pattern. Reasons:
 
@@ -270,5 +287,5 @@ Approve still gates:
   Allow modes.
 - Integration Delegation Framework (design) — `docs/design/14-integration-delegation.md`
   §14.12, the deniedTools spec.
-- Delegated Mode v2 (design) — `DELEGATED-MODE-V2-DESIGN.md` §4.5,
+- Delegated Mode v2 (design) — `docs/design/17-delegated-mode-v2.md` §4.5,
   the rationale for Notify-tier abolition + the starter denylist.

package/agent-assets/docs/concepts/skills.md

@@ -17,14 +17,16 @@ tags:
   - core
   - skills
   - safety
+  - knowledge
 status: stable
 ask_examples:
   - What skills does the agent have?
   - How do I add a new skill?
   - Why does the agent refuse to run a tool?
+  - How does skill self-optimization work?
 locale: en-US
 created: 2026-04-25
-updated: 2026-05-04
+updated: 2026-05-15
 keywords:
   - SKILL.md
   - allowed-tools
@@ -33,11 +35,12 @@ keywords:
   - self-optimization
   - skill curation
   - overlays
+  - skill
+  - skills
 related:
   - concepts/safety-and-execution
   - concepts/process-keys
   - reference/skills
-  - features/operations/skill-self-optimization
 ui_anchors:
   - /knowledge
   - /connections/mcp
@@ -69,7 +72,8 @@ visible to the model.
 - **SKILL.md**: the Markdown file that defines a single skill. Lives
   under `agent-assets/skills/<slug>/SKILL.md` in the repo and is
   materialized into each session workdir as `.claude/skills/<slug>/SKILL.md`
-  (or the per-backend equivalent).
+  (Codex uses `.codex/skills/`, Gemini uses `.gemini/skills/`, OpenCode
+  reuses `.claude/skills/` per V2 of docs/design/appendices/opencode-backend.md).
 - **`allowed-tools`**: a YAML list in the skill's frontmatter naming
   tools and patterns the session may use. Patterns like
   `Bash(curl http://localhost:8321/api/context/*)` are the daemon's
@@ -110,8 +114,9 @@ scoped toolset (`Bash(curl http://localhost:8321/api/skill-curation/*)`,
 drift signals than the previous overlay generated, the section is
 reverted and frozen for two cycles.
 
-See [Skill Self-Optimization](../features/operations/skill-self-optimization.md)
-for cadence, manual run, and the dashboard surface.
+Curation cadence, manual-run trigger, and the per-skill exclusion
+list are surfaced at **Settings → Self-learning**
+(`/settings/self-learning`).
 
 ## Where You See It in the Dashboard
 
@@ -131,5 +136,5 @@ for cadence, manual run, and the dashboard surface.
   `allowed-tools` is too permissive.
 - [Process Keys](process-keys.md) — the dispatch identity that picks
   which skill manifest to load.
-- [Skill Self-Optimization](../features/operations/skill-self-optimization.md) —
-  how overlays are generated and applied.
+- [Skills (Reference)](../reference/skills.md) — index of every
+  built-in skill.

package/agent-assets/docs/features/integrations/calendar.md

@@ -17,7 +17,7 @@ summary: |
   when the integration runs in direct mode.
 section: integrations
 tags:
-  - integration
+  - integrations
   - calendar
   - core
 status: stable

package/agent-assets/docs/features/integrations/git.md

@@ -14,9 +14,9 @@ summary: |
   notification spam.
 section: integrations
 tags:
-  - integration
+  - integrations
   - git
-  - observation
+  - observations
 status: stable
 ask_examples:
   - How do I add a git repo to watch?

package/agent-assets/docs/features/integrations/github.md

@@ -15,9 +15,9 @@ summary: |
   security alerts, assignments) become direct DMs.
 section: integrations
 tags:
-  - integration
+  - integrations
   - github
-  - observation
+  - observations
 status: stable
 ask_examples:
   - How do I connect GitHub?
@@ -26,6 +26,13 @@ ask_examples:
 locale: en-US
 created: 2026-04-25
 updated: 2026-04-28
+keywords:
+  - github
+  - issue
+  - pull request
+  - PR review
+  - workflow run
+  - security alert
 related:
   - features/integrations/git
   - concepts/observations

package/agent-assets/docs/features/integrations/mail.md

@@ -17,7 +17,7 @@ summary: |
   transport, not a separately exposed provider kind.)
 section: integrations
 tags:
-  - integration
+  - integrations
   - mail
   - core
 status: stable

package/agent-assets/docs/features/integrations/notion.md

@@ -12,9 +12,9 @@ summary: |
   Changes record observations consumed by the hourly check.
 section: integrations
 tags:
-  - integration
+  - integrations
   - knowledge
-  - observation
+  - observations
 status: stable
 ask_examples:
   - How do I connect Notion?
@@ -22,6 +22,12 @@ ask_examples:
 locale: en-US
 created: 2026-04-25
 updated: 2026-04-25
+keywords:
+  - notion
+  - page
+  - database
+  - notion observer
+  - notion poller
 related:
   - features/integrations/obsidian
   - features/routines/hourly-check
@@ -46,23 +52,45 @@ records observations on change.
 
 ## When It Runs / How It Is Triggered
 
-Continuous polling.
+Continuous polling in `direct` mode. In `delegated` / `native` / `disabled`
+modes the poller does not run — see Integration Modes below.
+
+## Integration Modes
+
+Notion supports all four integration modes (`direct` / `delegated` /
+`native` / `disabled`), selected from **Connections → Knowledge**.
+
+- **`direct`** — the daemon polls Notion itself, records change
+  observations, and the hourly check consumes them.
+- **`delegated`** — a delegated-sync worker runs on opt-in cadences;
+  observations are recorded the same way but on a different
+  schedule (see [Delegated Mode](../../concepts/delegated-mode.md)).
+- **`native`** — the main backend reaches Notion through its own MCP
+  connector on demand; the daemon does no polling. Observations
+  flow in only when the main backend POSTs them in-turn via
+  `/api/observations`. Available with Claude (`mcp__notion__*`).
+- **`disabled`** — silence; no observations, no daemon access.
+
+Switching modes requires the integration flip-lock probe to pass
+(connector reachable, capabilities reported). See `POST /api/integrations/notion/probe`.
 
 ## Where in the Dashboard
 
-- **Connections → Knowledge** holds the integration token and target
-  databases.
+- **Connections → Knowledge** holds the integration token, target
+  databases, and the mode picker.
 
 ## Configuration
 
 | Setting | Default | Notes |
 |---|---|---|
-| `notionPollIntervalSeconds` | 300 | Polling interval. |
+| `notionPollIntervalSeconds` | 300 | Polling interval (direct mode only). |
 
 ## When Something Goes Wrong
 
 - Notion's API rate-limits aggressively at high poll frequency. Stay
   ≥ 300 seconds.
+- "No observations" in `native` mode is expected — observations flow
+  only when the main backend POSTs them during a session.
 
 ## Related
 

package/agent-assets/docs/features/integrations/obsidian.md

@@ -13,9 +13,9 @@ summary: |
   notes on demand.
 section: integrations
 tags:
-  - integration
+  - integrations
   - knowledge
-  - observation
+  - observations
 status: stable
 ask_examples:
   - How do I connect my Obsidian vault?
@@ -24,6 +24,11 @@ ask_examples:
 locale: en-US
 created: 2026-04-25
 updated: 2026-04-25
+keywords:
+  - obsidian
+  - vault
+  - vault watcher
+  - obsidian observer
 related:
   - features/integrations/notion
   - features/routines/hourly-check

package/agent-assets/docs/features/lifestyle/git.md

@@ -27,7 +27,7 @@ tags:
   - git
   - github
   - automation
-  - observation
+  - observations
   - architecture
 status: stable
 ask_examples:
@@ -46,7 +46,7 @@ ask_examples:
   - How much does one Architecture refresh cost?
 locale: en-US
 created: 2026-05-05
-updated: 2026-05-07
+updated: 2026-05-17
 keywords:
   - my life
   - git
@@ -63,7 +63,6 @@ related:
   - features/integrations/git
   - features/integrations/github
   - features/routines/hourly-check
-  - guides/refresh-repository-architecture
   - concepts/observations
 ui_anchors:
   - /git
@@ -137,8 +136,8 @@ a Sonnet-class agent that reads the actual code.
   re-copies the source README into the slug directory before queueing
   the agent run, so the mirror stays in sync with what the agent
   actually reads.
-- **Never automatically** — there is no weekly/monthly cron that
-  refreshes Architecture. This is by design: each refresh costs one
+- **Never automatically** — there is no recurring cron that refreshes
+  Architecture. This is by design: each refresh costs one
   Sonnet session, and the analysis is durable across daily journal
   appends. If you want a recurring refresh, click the button on
   whatever cadence makes sense for the repo (after a big refactor,
@@ -314,8 +313,6 @@ repository has a `localPath`:
 
 ## Related
 
-- [Refresh repository architecture](../../guides/refresh-repository-architecture.md)
-  — focused walkthrough of the Refresh architecture button.
 - [Git integration](../integrations/git.md) — the underlying observer
   contract and `gitPollIntervalSeconds` semantics.
 - [GitHub integration](../integrations/github.md) — remote-side data

package/agent-assets/docs/features/lifestyle/receipts.md

@@ -3,6 +3,11 @@ schema_version: 1
 slug: features/lifestyle/receipts
 title: Receipts
 id: receipts
+aliases:
+  - receipts
+  - invoices
+  - expenses
+  - receipt files
 category: features
 summary: |
   A SQLite-backed log of receipt and invoice attachments the Gmail
@@ -12,6 +17,7 @@ section: lifestyle
 tags:
   - lifestyle
   - receipts
+  - mail
 status: stable
 ask_examples:
   - Where are my receipts stored?
@@ -19,7 +25,14 @@ ask_examples:
   - What does the receipts table track?
 locale: en-US
 created: 2026-04-25
-updated: 2026-04-26
+updated: 2026-05-17
+keywords:
+  - receipt
+  - invoice
+  - expense
+  - attachment
+  - PDF
+  - gmail receipts
 related:
   - features/lifestyle/reading
   - features/lifestyle/travel-bookings
@@ -56,7 +69,9 @@ the SQLite row plus the optional Obsidian copy.
 
 There is no dedicated tab today. Receipts surface inline when you ask
 for them in chat (e.g. "what receipts haven't I saved yet?"). The
-underlying API is documented in `agent-assets/skills/receipts/SKILL.md`.
+underlying API is documented in
+`agent-assets/skills/gmail-lifestyle/references/receipts-api.md`
+(loaded by the `gmail-lifestyle` skill).
 
 ## Configuration
 

package/agent-assets/docs/features/lifestyle/travel-bookings.md

@@ -3,6 +3,12 @@ schema_version: 1
 slug: features/lifestyle/travel-bookings
 title: Travel Bookings
 id: travel-bookings
+aliases:
+  - travel
+  - bookings
+  - flights
+  - hotels
+  - trains
 category: features
 summary: |
   A lightweight log of trips, flights, hotels, and confirmation
@@ -11,6 +17,7 @@ section: lifestyle
 tags:
   - lifestyle
   - travel
+  - mail
 status: stable
 ask_examples:
   - How do I log a flight?
@@ -18,6 +25,14 @@ ask_examples:
 locale: en-US
 created: 2026-04-25
 updated: 2026-04-25
+keywords:
+  - flight
+  - hotel
+  - booking
+  - trip
+  - itinerary
+  - trains
+  - restaurants
 related:
   - features/lifestyle/travel-time
 ---

package/agent-assets/docs/features/lifestyle/travel-time.md

@@ -15,7 +15,7 @@ section: lifestyle
 tags:
   - lifestyle
   - travel
-  - skill
+  - skills
 status: stable
 ask_examples:
   - How long will it take me to get to the airport?
@@ -23,6 +23,12 @@ ask_examples:
 locale: en-US
 created: 2026-04-25
 updated: 2026-04-25
+keywords:
+  - travel time
+  - departure time
+  - google maps
+  - commute
+  - ETA
 related:
   - features/lifestyle/travel-bookings
   - features/operations/schedule-approaching

package/agent-assets/docs/features/memory-files/agent-journal.md

@@ -60,7 +60,7 @@ first person. Each entry has:
 
 Unlike Activity, the journal is opinion-shaped. The evening review
 adds the day's reflections; the weekly retro reads them back to
-look for patterns; the monthly retro distills them into trend notes.
+look for patterns.
 
 ## When It Runs / How It Is Triggered
 
@@ -82,7 +82,7 @@ look for patterns; the monthly retro distills them into trend notes.
 
 - **Connections → Journal** is the read view of the file.
 - **Settings → Journal** controls retention and which routines
-  contribute (currently always-on for evening / weekly / monthly).
+  contribute (currently always-on for evening / weekly).
 
 ## Configuration
 

package/agent-assets/docs/features/memory-files/projects.md

@@ -15,6 +15,7 @@ section: memory-files
 tags:
   - memory
   - projects
+  - core
 status: stable
 ask_examples:
   - Where do I find the per-project files?
@@ -22,6 +23,11 @@ ask_examples:
 locale: en-US
 created: 2026-04-25
 updated: 2026-04-25
+keywords:
+  - projects.md
+  - project file
+  - projects/<slug>.md
+  - active project
 related:
   - features/memory-files/roadmap
 context_files:

package/agent-assets/docs/features/memory-files/roadmap.md

@@ -25,6 +25,11 @@ ask_examples:
 locale: en-US
 created: 2026-04-25
 updated: 2026-04-25
+keywords:
+  - roadmap.md
+  - roadmap
+  - long-term goals
+  - preparation timeline
 related:
   - features/routines/morning-routine
   - features/memory-files/projects

package/agent-assets/docs/features/memory-files/today.md

@@ -15,6 +15,7 @@ section: memory-files
 tags:
   - core
   - memory
+  - today
 status: stable
 ask_examples:
   - What is in today.md?

package/agent-assets/docs/features/memory-files/user-profile.md

@@ -26,6 +26,12 @@ ask_examples:
 locale: en-US
 created: 2026-04-25
 updated: 2026-04-25
+keywords:
+  - user profile
+  - user/profile.md
+  - user slices
+  - people
+  - expertise
 related:
   - getting-started/04-first-day
   - concepts/memory-model

package/agent-assets/docs/features/messaging/bang-commands.md

@@ -68,14 +68,24 @@ Pair your messaging app first; see
 | Command | What it does |
 |---|---|
 | `!help` | Lists every command currently registered — built-ins plus any custom user commands. |
-| `!stop` | Pauses cron-driven autonomous work (hourly check, morning / evening / weekly / monthly routines, scheduled tasks). In-flight runs are **not** aborted. |
+| `!stop` | Pauses cron-driven autonomous work (hourly check, morning / evening / weekly routines, scheduled tasks). In-flight runs are **not** aborted. |
 | `!start` | Resumes autonomous work after `!stop`. |
+| `!close` | Closes the active DM session for the current routing tuple so the next DM starts a fresh conversation. Returns `null` if no session was open. |
 | `!cost` | Last-7-day spend across all backends. |
-| `!cost claude` / `!cost codex` / `!cost gemini` | Spend for a single backend. |
+| `!cost claude` / `!cost codex` / `!cost gemini` / `!cost opencode` | Spend for a single backend (one row per registered backend). |
 | `!report` | Recent agent failures (last 7 days, top groups, most recent sample). |
-
-Custom commands added at `/settings/commands` show up in `!help`
-automatically — no restart needed.
+| `!wiki` | Wiki status snapshot (workspaces, recent ingests / compiles, queue depth). |
+| `!ingest <url>` | Ingest a URL into the wiki raw layer. Enqueues `wiki.ingest_url`. |
+| `!compile [full]` | Run wiki compile (raw → wiki synthesis). `full` mode crosses the per-workspace cost threshold and routes through the dashboard approval queue. |
+| `!ask <question>` | Q&A against the compiled wiki. Reply lands back on the same channel. |
+| `!lint` | Wiki health pass — writes `90_meta/health/<date>.md`. |
+| `!trace <idea>` | Chronological evolution of an idea across raw / wiki / outputs layers. |
+| `!connect <A> <B>` | Bridge two domains; writes `30_outputs/<date>-connect-<slug>.md`. |
+
+Wiki commands accept an optional `@<workspace>` suffix
+(e.g. `!compile @work`) to target a specific workspace when multiple
+are configured. Custom commands added at `/settings/commands` show
+up in `!help` automatically — no restart needed.
 
 ## How They Look
 
@@ -127,8 +137,8 @@ enabled user commands at `/settings/commands`.
   opened. Every invocation writes one `bang_command` row to
   `agent_actions` for the activity log.
 - **While paused**, any DM (bang or not) replies with the paused
-  notice; only `!start`, `!cost`, `!report`, and `!help` continue to
-  run.
+  notice; only `!start`, `!cost`, `!report`, `!help`, and `!close`
+  continue to run.
 - **Replies land where the command did** — same platform, same
   channel, same thread.
 
@@ -149,9 +159,9 @@ window version.
 - **The reply lists fewer commands than you expect.** `!help` only
   lists *enabled* user commands. Toggle the row on at
   `/settings/commands`.
-- **`!cost foo` says "unknown command".** Only `claude`, `codex`,
-  `gemini` are valid backend arguments today. Plain `!cost` (no
-  argument) covers all three.
+- **`!cost foo` says "unknown command".** Valid backend arguments
+  match the registered backends — `claude`, `codex`, `gemini`,
+  `opencode`. Plain `!cost` (no argument) covers all of them.
 
 ## Related
 

package/agent-assets/docs/features/messaging/discord.md

@@ -3,6 +3,10 @@ schema_version: 1
 slug: features/messaging/discord
 title: Discord
 id: discord
+aliases:
+  - discord bot
+  - discord adapter
+  - discord pairing
 category: features
 summary: |
   Pair Discord by creating a bot in the developer portal, pasting the
@@ -11,13 +15,20 @@ summary: |
 section: messaging
 tags:
   - messaging
-  - integration
+  - integrations
+  - discord
 status: stable
 ask_examples:
   - How do I pair Discord?
 locale: en-US
 created: 2026-04-25
 updated: 2026-04-25
+keywords:
+  - discord
+  - bot token
+  - DM
+  - owner channel
+  - messaging adapter
 related:
   - features/messaging/overview
   - features/messaging/pairing-and-magic-phrase

package/agent-assets/docs/features/messaging/overview.md

@@ -18,6 +18,7 @@ section: messaging
 tags:
   - core
   - messaging
+  - integrations
 status: stable
 ask_examples:
   - How do I pair my phone with the agent?
@@ -43,7 +44,6 @@ related:
   - features/messaging/discord
   - features/messaging/whatsapp
   - features/messaging/dashboard-chat
-  - features/messaging/voice-attachments
   - features/messaging/bang-commands
   - features/operations/notifications
   - features/operations/quiet-hours
@@ -65,14 +65,19 @@ notification fires.
 ## What It Does
 
 - **Reactive DMs**: the agent answers every direct message you send.
-- **Mentions**: in shared channels, @-mentioning the agent answers
-  the same way (group chat is filtered out by design — DMs only).
+- **@-mentions**: in Slack, @-mentioning the agent inside a shared
+  channel routes to `message.mention` and is answered the same way
+  as a DM. Telegram, Discord, and WhatsApp drop all non-DM traffic
+  (no group support); multi-user DMs (Slack `mpim`) are also
+  filtered out — Aitne is single-owner by design.
 - **Outbound notifications**: routines and observations fire alerts
   back through the same channel.
 - **Voice attachments**: when the platform attaches audio (Telegram
   voice notes, WhatsApp voice messages, etc.), the daemon transcribes
-  it locally with Whisper before handing the turn to the agent. See
-  [Voice Attachments](voice-attachments.md).
+  it locally with Whisper before handing the turn to the agent.
+  Transcription runs entirely on your machine — no audio is shipped
+  to a cloud API. Configurable via the `PA_VOICE_TRANSCRIPTION_*`
+  env vars (model, language, duration cap).
 - **Magic-phrase pairing**: the operator types a one-time phrase to
   bind their account to the agent (anti-impersonation).
 
@@ -117,7 +122,5 @@ adapter and the dispatcher both check).
 - [Pairing & Magic Phrase](pairing-and-magic-phrase.md)
 - [Bang Commands](bang-commands.md) — `!stop` / `!start` / `!cost` /
   `!report` / `!help` for mobile-first daemon control.
-- [Voice Attachments](voice-attachments.md) — local Whisper
-  transcription for inbound audio.
 - [Dashboard Chat](dashboard-chat.md) — the in-browser equivalent.
 - [Notifications](../operations/notifications.md)