let-them-talk 0.1.0__tar.gz

let_them_talk-0.1.0/.gitignore

@@ -0,0 +1,16 @@
+# Your own data: boards (chat ids, connection reasons) and the log of sent notes
+boards/
+logs/
+# Python cache
+__pycache__/
+# Private notes (handoffs, scratch files)
+temp/
+# Claude Code (local settings, worktrees)
+.claude/
+# Written by Claude Code each time it loads a mod
+mods/*/.claude-plugin/types/
+mods/*/tsconfig.json
+# The page checks' browser (tests/setup.sh)
+tests/.venv/
+# Built packages (uv build)
+dist/

let_them_talk-0.1.0/LICENSE

@@ -0,0 +1,202 @@
+
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

let_them_talk-0.1.0/PKG-INFO

@@ -0,0 +1,55 @@
+Metadata-Version: 2.5
+Name: let-them-talk
+Version: 0.1.0
+Summary: A visual board for your Claude Code agents: see sessions, subagents and background agents as cards, and draw arrows so they talk to each other.
+Project-URL: Homepage, https://github.com/abdullahkavakli/let-them-talk
+Project-URL: Issues, https://github.com/abdullahkavakli/let-them-talk/issues
+Project-URL: Documentation, https://github.com/abdullahkavakli/let-them-talk/blob/main/docs/usage.md
+Author: Abdullah KAVAKLI
+License-Expression: Apache-2.0
+License-File: LICENSE
+License-File: mods/let-them-talk-handoff/skills/handoff/LICENSE
+License-File: static/fonts/LICENSE-Selawik.txt
+Keywords: agents,board,claude,claude-code,dashboard,multi-agent,subagents
+Classifier: Environment :: Web Environment
+Classifier: Intended Audience :: Developers
+Classifier: Operating System :: MacOS
+Classifier: Operating System :: POSIX :: Linux
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Classifier: Topic :: Software Development
+Requires-Python: >=3.12
+Description-Content-Type: text/markdown
+
+# Let Them Talk
+
+A visual board for your agents. Each running Claude Code session is a card, and an arrow from one card to another tells both chats who they work with and why.
+
+![Let Them Talk: cards for Claude Code chats, an arrow between two of them, and their conversation](https://raw.githubusercontent.com/abdullahkavakli/let-them-talk/main/docs/demo/demo.gif)
+
+## What it does
+
+- Draw an arrow between two chats and it tells them who to talk to and why. They then talk through Claude Code's own messaging.
+- Click an arrow to read what the two chats sent each other.
+- Start a team: a master and up to 8 agents, each with its own role, prompt, model and effort.
+- Terminal chats, Cursor chats and background agents share one board, with the subagents they run.
+- Local and small: the server is one Python file that uses only Python's standard library, and nothing is installed into Claude Code.
+
+## Start
+
+Needs Python 3.12+ and Claude Code (on Windows, both inside WSL).
+
+- One line: `uvx let-them-talk` (needs [uv](https://docs.astral.sh/uv/); or `pipx run let-them-talk`). It starts the board and opens it in your browser; `--no-open` skips that. Its boards are kept in `~/.local/share/let-them-talk`.
+- From a copy of this repository, on Windows: double-click `Start Let Them Talk.bat`.
+- From a copy of this repository, on macOS, Linux or WSL: `./start.sh`, then open <http://localhost:8765>.
+
+Click **New board** and pick a project folder: its running chats appear as cards. Drag the blue handle from one card onto another, then click **Connect**.
+
+More: [using it](https://github.com/abdullahkavakli/let-them-talk/blob/main/docs/usage.md) · [how it works](https://github.com/abdullahkavakli/let-them-talk/blob/main/docs/how-it-works.md) · [settings](https://github.com/abdullahkavakli/let-them-talk/blob/main/docs/configuration.md) · [limits](https://github.com/abdullahkavakli/let-them-talk/blob/main/docs/limits.md)
+
+Not related to [Dekelelz/let-them-talk](https://github.com/Dekelelz/let-them-talk), a different project with the same name.
+
+## License
+
+[Apache 2.0](https://github.com/abdullahkavakli/let-them-talk/blob/main/LICENSE). The Selawik font and the `/handoff` skill keep their own licenses, next to their files.

let_them_talk-0.1.0/README.md

@@ -0,0 +1,31 @@
+# Let Them Talk
+
+A visual board for your agents. Each running Claude Code session is a card, and an arrow from one card to another tells both chats who they work with and why.
+
+![Let Them Talk: cards for Claude Code chats, an arrow between two of them, and their conversation](docs/demo/demo.gif)
+
+## What it does
+
+- Draw an arrow between two chats and it tells them who to talk to and why. They then talk through Claude Code's own messaging.
+- Click an arrow to read what the two chats sent each other.
+- Start a team: a master and up to 8 agents, each with its own role, prompt, model and effort.
+- Terminal chats, Cursor chats and background agents share one board, with the subagents they run.
+- Local and small: the server is one Python file that uses only Python's standard library, and nothing is installed into Claude Code.
+
+## Start
+
+Needs Python 3.12+ and Claude Code (on Windows, both inside WSL).
+
+- One line: `uvx let-them-talk` (needs [uv](https://docs.astral.sh/uv/); or `pipx run let-them-talk`). It starts the board and opens it in your browser; `--no-open` skips that. Its boards are kept in `~/.local/share/let-them-talk`.
+- From a copy of this repository, on Windows: double-click `Start Let Them Talk.bat`.
+- From a copy of this repository, on macOS, Linux or WSL: `./start.sh`, then open <http://localhost:8765>.
+
+Click **New board** and pick a project folder: its running chats appear as cards. Drag the blue handle from one card onto another, then click **Connect**.
+
+More: [using it](docs/usage.md) · [how it works](docs/how-it-works.md) · [settings](docs/configuration.md) · [limits](docs/limits.md)
+
+Not related to [Dekelelz/let-them-talk](https://github.com/Dekelelz/let-them-talk), a different project with the same name.
+
+## License
+
+[Apache 2.0](LICENSE). The Selawik font and the `/handoff` skill keep their own licenses, next to their files.

let_them_talk-0.1.0/docs/configuration.md

@@ -0,0 +1,85 @@
+# Configuration
+
+Let Them Talk runs without setup; these settings and the optional
+suggestions mod change how it runs.
+
+## Environment settings
+
+Set these for the server before you start it.
+
+- `LTT_PORT`: the port it listens on (default `8765`). The Windows `.bat`
+  always opens the page on port 8765; for the suggestions mod, see
+  [below](#suggestions-mod).
+- `LTT_DATA`: the folder where the app keeps its boards and logs. By
+  default the app's own folder when you run `./start.sh` or
+  `python3 server.py` from a copy of the repository, and
+  `$XDG_DATA_HOME/let-them-talk` (`~/.local/share/let-them-talk` if that isn't
+  set, on macOS too) when you installed it with `uvx` or `pipx`. It is also the
+  folder the app's own short `claude -p` runs start in. To carry boards over
+  to another folder, copy `boards/` and `logs/` there. See
+  [What is stored where](how-it-works.md#what-is-stored-where).
+- `LTT_MODEL`: the model for notes and TL;DRs (default `haiku`).
+- `LTT_CLAUDE`: the path to the `claude` binary. By default `claude` on the
+  PATH, else `~/.local/bin/claude`.
+- `LTT_WINDOWS_HOME`: your Windows user folder as a WSL path, e.g.
+  `/mnt/c/Users/you`, whose `.claude` holds Windows sessions. By default the
+  app asks Windows for `%USERPROFILE%`.
+- `CLAUDE_CONFIG_DIR`: if you run Claude Code with it, set it for the server
+  too. It then reads sessions, transcripts and background agents from there
+  instead of `~/.claude`.
+
+The older `ORGANIZER_*` names (such as `ORGANIZER_PORT`) still work.
+
+## Suggestions mod
+
+The mod in `mods/let-them-talk-suggestions` does three things: it passes on
+Claude Code's reply suggestion for terminal chats (see
+[Reply suggestions](how-it-works.md#reply-suggestions)), it runs a
+**Chat in IDE** on the model picked in **New agent** (see
+[The model of a Chat in IDE](how-it-works.md#the-model-of-a-chat-in-ide)),
+and in a background agent, after an ultracode switch made while it works,
+it runs `/effort status` so the app can read the switch (see
+[Ultracode](how-it-works.md#ultracode)). In other chats it runs nothing
+for ultracode.
+
+To install it, add this repository as a plugin marketplace and install the
+plugin from it:
+
+```
+claude plugin marketplace add abdullahkavakli/let-them-talk
+claude plugin install let-them-talk-suggestions@let-them-talk
+```
+
+Or, in a chat:
+`/plugin install let-them-talk-suggestions --marketplace abdullahkavakli/let-them-talk`.
+
+The plugin is the mod of the latest release, the same release as the app
+`uvx let-them-talk` runs, so the two always match. A change made to the mod
+after that release reaches the plugin only with the next release, which brings
+both: update the plugin with
+`claude plugin update let-them-talk-suggestions@let-them-talk`.
+
+If you run a copy of the repository and want the mod to follow it, load it
+from there instead. To load it in every chat started from then on, add its
+absolute path to the `env` block of `~/.claude/settings.json`:
+
+```json
+{ "env": { "CLAUDE_CODE_PLUGIN_DIRS": "/path/to/let-them-talk/mods/let-them-talk-suggestions" } }
+```
+
+For a single chat:
+`claude --plugin-dir /path/to/let-them-talk/mods/let-them-talk-suggestions`.
+
+A copy loaded this way (`CLAUDE_CODE_PLUGIN_DIRS` or `--plugin-dir`) takes the
+place of an installed one with the same name, so having both is fine.
+
+- Chats started before the mod was loaded show no suggestion until
+  restarted, and an ultracode switch made while they work shows only at
+  their next prompt.
+- A **Chat in IDE** gets the picked model only where the mod is loaded: a
+  Cursor or VS Code connected to WSL reads the same `~/.claude/settings.json`;
+  one on Windows has its own.
+- It talks to port 8765. If you changed `LTT_PORT`, edit `APP` in
+  `mods/let-them-talk-suggestions/hooks/register.ts` to match, in a copy of the
+  repository loaded as above: Claude Code keeps an installed copy itself and
+  may replace it when the plugin updates.

let_them_talk-0.1.0/docs/how-it-works.md

@@ -0,0 +1,227 @@
+# How it works
+
+Where the board's data comes from, how notes reach a chat, and what the app
+stores.
+
+## Sessions and cards
+
+- Live sessions come from `~/.claude/sessions/*.json`; background agents that
+  aren't running come from `claude agents --json --all`. (With
+  [`CLAUDE_CONFIG_DIR`](configuration.md#environment-settings) set, that
+  folder takes the place of `~/.claude` here and below.)
+- A card greys out when its session ends, and leaves the board about two
+  minutes later unless it has arrows.
+- Titles, models and recent messages are read from its transcript in
+  `~/.claude/projects/`, and a chat's agents from the subagent and workflow
+  files next to it.
+- Inside WSL, sessions of Claude Code on native Windows are read from
+  `%USERPROFILE%\.claude` while their `claude.exe` runs.
+
+## Notes and messages
+
+Notes are delivered by a short headless run,
+`claude -p --model haiku --name let-them-talk`, whose only tool is
+`SendMessage` (about 8 s); [`LTT_MODEL`](configuration.md#environment-settings)
+picks another model.
+
+Each note is then marked *sent*, *sent, reworded* or *failed*; until then its
+arrow shows it is still sending (see [Arrows](usage.md#arrows)). If the
+server restarts meanwhile, the run carries on without it, and the arrow is
+settled from the chats' transcripts: *sent* once the note shows up there,
+*failed* if it hasn't within 3½ minutes.
+
+A message you send to a chat goes the same way, as *[Let Them Talk] Message
+from your user:*, and a new editor chat's prompt as *[Let Them Talk] Your
+user started this chat from Let Them Talk with this prompt:*. An idle
+background agent gets yours as a prompt instead: typed into its prompt box
+through `claude attach` while its process runs, or with
+`claude --resume <id> --bg` once that has ended.
+
+## Images
+
+An image you send is saved under a random name, readable only by you, in the
+chat's own folder in Claude Code's temporary folder
+(`/tmp/claude-<uid>/<project>/let-them-talk-images/`). Your text gets a line
+`[Image: source: <path>]` for it, the way Claude Code names an image you
+paste. The chat opens it with its Read tool without asking, as it does its
+own files there.
+
+Images older than a week are deleted the next time an image is saved.
+
+## An arrow's conversation
+
+An arrow's **Conversation** is read from both chats' transcripts. A message
+belongs to it when its message id is in one chat's `SendMessage` call and in
+the other's transcript, or when it went to the other chat's own socket (*not
+read yet* until it shows up there).
+
+Matching by name is a last resort, used only when one side's transcript is
+gone (*read state unknown* if it was the receiver's). A→B and B→A arrows show
+the same messages between the two chats, each with its own notes.
+
+## TL;DRs
+
+A TL;DR is written by `claude -p` on the same model as notes, with no tools
+(about 7 s). It is written once per finished reply of 280+ characters, only
+when you open that chat, and kept in memory until the server restarts.
+
+## Handoffs
+
+A handoff is written by `claude -p --resume <id> --fork-session` on the
+chat's own model, running the app's `/handoff` skill over the whole
+conversation. The skill is loaded for that run only, with
+`--plugin-dir mods/let-them-talk-handoff`; your own skills are left as they
+are. It takes a few seconds to minutes and costs about one reply of that
+chat. The fork is not saved.
+
+Nobody can answer the fork, so it edits files without asking
+(`acceptEdits`), in the chat's folder and in the handoff folder (see
+[What is stored where](#what-is-stored-where)). It is told only to write the
+handoff there; nothing else stops it from editing the chat's files.
+
+## Compacting
+
+**Compact** types `/compact` into a running background agent through
+`claude attach`, as a prompt is typed, pressing Enter only once its prompt
+box shows `/compact`. An ended one is woken with
+`claude --resume <id> --bg -- /compact`, which Claude Code runs as the
+command too. A chat in a terminal or an editor gets only messages, read
+between steps, and a message can't run a command, so it has no **Compact**.
+
+Claude Code writes the result into the agent's transcript once it is done
+(seconds, or a minute or two for a long conversation): a compact boundary
+with the context's size before and after, then `/compact`'s answer. The app
+reads what the transcript gets after you click. It stops waiting after 8
+minutes, or once the agent is neither busy nor done for a few seconds (Esc in
+its terminal, or it ended). Meanwhile it doesn't look up ultracode (see
+[Ultracode](usage.md#ultracode)) in that agent.
+
+## Reply suggestions
+
+The Send box shows the same grey suggestion as the chat's own prompt box, or
+nothing:
+
+- **A background agent that ends its turn waiting on you:** Claude Code saves
+  its suggestion in `~/.claude/jobs/<id>/state.json` (the one `claude agents`
+  offers with Tab), and the app reads it there. No setup, any agent. This job
+  file isn't an official Claude Code interface; if an update moves it, those
+  agents show nothing rather than wrong text.
+- **Terminal chats, and background turns that don't wait on you:** Claude
+  Code saves the suggestion nowhere, so the
+  [suggestions mod](configuration.md#suggestions-mod) passes it on as it is
+  shown. When Claude Code makes none (the window isn't focused, or nobody is
+  attached) and the app has that chat's details open, the mod makes one the
+  way Claude Code does (one tool-less reply of the chat's own model over its
+  cached conversation). It puts it in the chat's prompt box too, so both
+  match.
+  - It first waits about 12 s for Claude Code's own.
+  - At most one per turn, none for chats nobody has open, and none once the
+    reply is more than 5 minutes old (its conversation may no longer be
+    cached).
+  - While the app waits for one (up to about 45 s after the reply, or after
+    you open the chat), the Send box says *Suggesting a reply…*.
+- **Chats in Cursor or VS Code** show none, like their own composer: the
+  extension doesn't ask Claude Code for suggestions.
+
+## The model of a Chat in IDE
+
+An editor link can't name a model, so the
+[suggestions mod](configuration.md#suggestions-mod) runs a **Chat in IDE**
+started from the app on the model picked in
+[New agent](usage.md#new-agent). The app tells the mod which model as the
+chat's first turns start, and the mod names that model on the chat's own
+requests, not its subagents'.
+
+## Ultracode
+
+Whether ultracode is on in a background agent (see
+[Ultracode](usage.md#ultracode)) is read from its transcript: Claude Code
+notes each `/effort` answered while the agent is idle ("Ultracode on …",
+"Ultracode off …", or for `/effort status` a line that names ultracode only
+while it is on), and at each prompt a reminder when it changed since the last
+one. The newest of these counts, with what the app saw itself: a **Turn on**
+or **Turn off**, or a look in its Effort panel.
+
+- **While it works:** an `/effort` answers only on the agent's screen, and
+  its transcript gets nothing. In a background agent that loads the
+  [suggestions mod](configuration.md#suggestions-mod), the mod sees the
+  command run but not its answer. When `/effort ultracode …` or the Effort
+  panel gets no answer (an effort level alone leaves ultracode as it was, so
+  it doesn't count), it runs `/effort status` in the agent, once. Claude
+  Code queues it until the agent is idle, then notes its answer in the
+  transcript, where the app reads it like any other. The mod sends the app
+  nothing about it, and the app shows neither that line nor its queued entry
+  in the agent's messages. Without the mod, the agent's next prompt tells.
+- **Only in background agents:** Claude Code starts one with
+  `CLAUDE_CODE_SESSION_KIND=bg`, which also makes its registry file's kind
+  `bg`, the kind the app reads. The mod runs `/effort status` only where that
+  variable says `bg`, so never in an ordinary terminal or IDE chat.
+- **The app's own Turn on or off while it works:** the app reads Claude
+  Code's answer off the agent's screen, so it is in no transcript either.
+  The app keeps it on disk, as it does what a look in the Effort panel
+  showed, so a restart of the app loses neither.
+- **Which process:** a switch the app keeps counts only while the process
+  that had it (by its pid) runs the agent, and only from when that process
+  started, so a later process that gets the same pid doesn't inherit it.
+- **The panel opened only to look:** closed with Esc, Claude Code notes
+  "Cancelled", which changes nothing. Opened while the agent works, the mod's
+  `/effort status` reads what it left once the agent is idle.
+
+## Moving an editor chat to a terminal
+
+The editor's Claude panel runs a chat as a Claude Code process of its own
+(entrypoint `claude-vscode`, in `~/.claude/sessions/<pid>.json`). Claude Code
+doesn't stop a terminal from resuming a conversation that process still holds;
+both would write its transcript. So **Open in terminal** first closes the one
+in the editor, then opens `claude --resume <id>`.
+
+- **How it is closed:** the process gets SIGINT, once you said yes. In the
+  mode the extension runs it in, Claude Code answers by ending its turn,
+  saving the conversation and exiting with code 0, which is what happens when
+  the extension closes a chat itself (it ends the process's input). SIGTERM
+  would exit with code 143, which the extension reports as an error in the
+  panel, so it isn't used. The app waits up to 10 seconds for the process to
+  go; if it stays, nothing opens in the terminal and the notice says to close
+  the tab in the editor.
+- **What the editor shows:** the extension sees the process end the way it
+  does when you close the tab, with no error. Its panel keeps the
+  conversation and is not started again by itself; a message typed there later
+  starts it, and the extension, which reads the same session files, sees the
+  terminal holds the chat and asks before taking it back.
+- **Which chats:** a chat on this machine whose process was started by the
+  editor. Windows chats can't be signalled from here.
+
+## Security
+
+The server listens only on 127.0.0.1 and answers only requests addressed to
+`localhost` or `127.0.0.1` on its port. It takes commands only with its own
+request header, which other web pages can't send, and tells browsers never to
+show it in a frame.
+
+## What is stored where
+
+The first group lives in the app's *data folder*. Run from a copy of the
+repository (`./start.sh`), that is the app's own folder. Installed with `uvx`
+or `pipx`, it is `~/.local/share/let-them-talk` (`$XDG_DATA_HOME/let-them-talk`
+if that is set; the same on macOS), so nothing is written next to the
+installed code. [`LTT_DATA`](configuration.md#environment-settings) picks
+another folder in both cases. The app's own short `claude -p` runs (notes,
+TL;DRs, looking up a model's full id) start in the data folder too, and save
+no conversation.
+
+- `boards/`: the boards, a card's name included.
+- `logs/relay.jsonl`: every note and message sent through the headless run.
+- `logs/suggestions.json`: the mod's latest suggestions, kept for a day.
+- `logs/models.json`: the model picked for each Chat in IDE, kept a week, so
+  a chat resumed within a week still runs on it.
+- `logs/ultracode.json`: each chat's latest ultracode switch the app made or
+  saw in its Effort panel, with the pid of the process that had it, kept 30
+  days.
+
+Elsewhere:
+
+- `let-them-talk-handoffs` in the system's temp folder (`/tmp` on Linux and
+  WSL): the handoffs.
+- Images you send: see [Images](#images).
+- What the page keeps in your browser: see
+  [What the page remembers](usage.md#what-the-page-remembers).