genaicode 2.39.0 → 2.40.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/agents/pack.d.ts +18 -1
- package/dist/agents/pack.js +91 -56
- package/dist/agents/pack.js.map +1 -1
- package/dist/agents/session.d.ts +11 -0
- package/dist/agents/session.js +76 -3
- package/dist/agents/session.js.map +1 -1
- package/dist/agents/sessions/answer.js +2 -3
- package/dist/agents/sessions/answer.js.map +1 -1
- package/dist/agents/sessions/approval-previews.d.ts +7 -0
- package/dist/agents/sessions/approval-previews.js +83 -0
- package/dist/agents/sessions/approval-previews.js.map +1 -0
- package/dist/agents/sessions/files.d.ts +2 -0
- package/dist/agents/sessions/files.js +10 -0
- package/dist/agents/sessions/files.js.map +1 -1
- package/dist/agents/sessions/hook-input.d.ts +3 -0
- package/dist/agents/sessions/hook-input.js +21 -1
- package/dist/agents/sessions/hook-input.js.map +1 -1
- package/dist/agents/sessions/index.js +9 -1
- package/dist/agents/sessions/index.js.map +1 -1
- package/dist/agents/sessions/live.d.ts +5 -1
- package/dist/agents/sessions/live.js +13 -5
- package/dist/agents/sessions/live.js.map +1 -1
- package/dist/agents/sessions/stores.js +4 -3
- package/dist/agents/sessions/stores.js.map +1 -1
- package/dist/agents/sessions/tools.js +2 -1
- package/dist/agents/sessions/tools.js.map +1 -1
- package/dist/agents/sessions/types.d.ts +5 -0
- package/dist/agents/types.d.ts +5 -0
- package/dist/agents.d.ts +1 -1
- package/dist/agents.js.map +1 -1
- package/dist/cli/args.d.ts +6 -1
- package/dist/cli/args.js +21 -2
- package/dist/cli/args.js.map +1 -1
- package/dist/cli/main.d.ts +4 -1
- package/dist/cli/main.js +47 -10
- package/dist/cli/main.js.map +1 -1
- package/dist/react/agent-chat.css +122 -0
- package/dist/react/agent-chat.d.ts +3 -0
- package/dist/react/agent-chat.js +14 -4
- package/dist/react/agent-chat.js.map +1 -1
- package/dist/react/format.d.ts +3 -0
- package/dist/react/format.js +20 -0
- package/dist/react/format.js.map +1 -1
- package/dist/react/pack-tasks.d.ts +12 -0
- package/dist/react/pack-tasks.js +49 -0
- package/dist/react/pack-tasks.js.map +1 -0
- package/dist/ui/index.js +98 -89
- package/dist/ui/web-client.js +137 -15
- package/docs/CHANGELOG.md +8 -0
- package/docs/agents.md +23 -5
- package/docs/cli.md +34 -0
- package/docs/react.md +1 -0
- package/package.json +1 -1
package/docs/CHANGELOG.md
CHANGED
|
@@ -8,6 +8,14 @@ fails. Merging to master refreshes the `release` PR; merging that publishes to n
|
|
|
8
8
|
|
|
9
9
|
## Unreleased
|
|
10
10
|
|
|
11
|
+
## 2.40.0 — 2026-11-20
|
|
12
|
+
|
|
13
|
+
### Added
|
|
14
|
+
|
|
15
|
+
- `genaicode tower` opens an indexed chat with waiting sessions across projects. Use `/waiting [project]` to refresh the queue and `/read <id>` to preview a conversation locally (#217)
|
|
16
|
+
- The waiting queue and browser Sessions view show up to 200 characters of a shell command awaiting approval in GitHub Copilot CLI or Gemini CLI (#219)
|
|
17
|
+
- `genaicode chat` and `ui` accept `--pack claude:codex,gemini` to follow delegated tasks, answer approvals, steer a worker and review or merge its changes from either view (#220)
|
|
18
|
+
|
|
11
19
|
## 2.39.0 — 2026-11-20
|
|
12
20
|
|
|
13
21
|
### Added
|
package/docs/agents.md
CHANGED
|
@@ -253,6 +253,12 @@ its own. The lead needs `capabilities.mcp` and sees these tools on an MCP server
|
|
|
253
253
|
- `resume`, `steer()` and `result.sessionId` are the lead's. Tasks still running when the lead
|
|
254
254
|
finishes are stopped. Unmerged worktrees stay for the next turn; `agent.close()` removes them.
|
|
255
255
|
|
|
256
|
+
In a live chat, `turn.pack.tasks` holds the delegated work under its original turn.
|
|
257
|
+
`session.actOnPackTask?.(turn.id, id, { type: 'diff' })` returns the unified patch.
|
|
258
|
+
The same method accepts `merge`, `discard`, `stop` and `{ type: 'steer', text }`.
|
|
259
|
+
Each run also exposes these controls as `run.pack`. A chat can still review and merge
|
|
260
|
+
finished tasks after a later turn starts. Adopted transcripts are read-only.
|
|
261
|
+
|
|
256
262
|
## Sessions across agents
|
|
257
263
|
|
|
258
264
|
Every agent keeps its conversations somewhere of its own. `listSessions()` reads those stores and
|
|
@@ -304,10 +310,9 @@ prompt, asks for approval, gets its answer, finishes a turn and ends. It keeps o
|
|
|
304
310
|
per session in `~/.genaicode/live` with the state, the time, the directory and the name of the
|
|
305
311
|
tool asked for: nothing of what was said, and nothing leaves the machine.
|
|
306
312
|
|
|
307
|
-
|
|
308
|
-
|
|
309
|
-
|
|
310
|
-
asking. Gemini and Cursor wait for their hooks (genaicode's takes a few milliseconds), and Gemini
|
|
313
|
+
Claude Code, Codex and Copilot identify the tool in an approval request. Gemini CLI identifies
|
|
314
|
+
shell commands in its permission notifications and can also send a generic request for attention.
|
|
315
|
+
Cursor reports working sessions and finished turns. Gemini and Cursor wait for their hooks, and Gemini
|
|
311
316
|
shows that a hook runs unless `hooksConfig.notifications` is off.
|
|
312
317
|
|
|
313
318
|
`listSessions()` puts that over what the stores suggest: `state` is exact, and a `waiting`
|
|
@@ -322,6 +327,15 @@ the command that continues it. It looks a week back unless `--since` says otherw
|
|
|
322
327
|
browser's Sessions view has the same as a "N wait for you" button; in code,
|
|
323
328
|
`waitingSessions(sessions)` picks and orders them.
|
|
324
329
|
|
|
330
|
+
For Copilot and Gemini shell approvals, `waitingCommands` contains the tool name and up to
|
|
331
|
+
200 characters of each matching command. Genaicode keeps a digest in its private hook record
|
|
332
|
+
and reads the text on demand from the agent's own local session journal. A session with a
|
|
333
|
+
missing journal or incomplete permission details still reports the approval. The CLI, the
|
|
334
|
+
browser and the opt-in `sessions_waiting` MCP tool show the preview. Desktop alerts contain
|
|
335
|
+
only the agent, project and tool name. The user answers these requests in the native agent.
|
|
336
|
+
Copilot records the decision as soon as it is made. Gemini CLI 0.70 reports the result when
|
|
337
|
+
the tool finishes, so a refresh can still show its last request while the command runs.
|
|
338
|
+
|
|
325
339
|
In code: `installHooks('claude')`, `hooksStatus()`, `uninstallHooks()`, and `recordHook(agent,
|
|
326
340
|
input)` with `readLive()` and `withLive(sessions, records)` to do the same for an agent of your
|
|
327
341
|
own whose hooks give the same JSON (`session_id`, `hook_event_name`, `cwd`); `hookInput(agent,
|
|
@@ -482,6 +496,10 @@ and shows up to 30 sessions by default. For questions held open by `hooks instal
|
|
|
482
496
|
the agent can read the question ID, tool name and preview. Archived sessions stay out of this
|
|
483
497
|
view and appear with `[archived]` in `sessions_list` and `sessions_read`.
|
|
484
498
|
|
|
499
|
+
The [session tower](cli.md#open-the-session-tower) opens an indexed chat with a local preview
|
|
500
|
+
of the queue. Its `/waiting [project]` and `/read <id>` commands call the same read-only
|
|
501
|
+
operations and display the results to the user. A regular chat message starts an agent turn.
|
|
502
|
+
|
|
485
503
|
`state` is `active`, `waiting`, `unfinished` or `approval` (a session that asks whether it may use
|
|
486
504
|
a tool, known once the [hooks](#which-sessions-need-you-hooks) are installed). The tools only read
|
|
487
505
|
the local stores. They are off unless you ask for them, because the agent that has them reads what
|
|
@@ -1209,4 +1227,4 @@ way the built-in drivers are tested.
|
|
|
1209
1227
|
3. MCP server injection per driver, an env scrub helper, and an opt-in verify/repair helper. Done.
|
|
1210
1228
|
4. More CLIs (gemini, cursor, opencode) and a hosted coding-agent seam (`hostedAgent`, `managedAgent`). Done.
|
|
1211
1229
|
5. A Copilot CLI driver (`copilot`). Done.
|
|
1212
|
-
6. `pack()`: a lead agent
|
|
1230
|
+
6. `pack()`: a lead agent delegates to installed agents. The terminal and browser chat expose it with `--pack`.
|
package/docs/cli.md
CHANGED
|
@@ -97,6 +97,12 @@ must finish first. The check uses the latest store and hook information availabl
|
|
|
97
97
|
`npx genaicode notify on` then shows a desktop notification when a session asks for approval
|
|
98
98
|
(`--done`: also when one finishes its turn; see [notify](agents.md#being-told-notify)).
|
|
99
99
|
|
|
100
|
+
With Copilot or Gemini hooks installed, `sessions --waiting`, the tower's `/waiting` command
|
|
101
|
+
and the browser's Sessions view show up to 200 characters of a shell command awaiting approval.
|
|
102
|
+
Refresh the queue to see which commands still need a decision. Answer in the agent that asked.
|
|
103
|
+
Copilot reports a decision immediately. Gemini CLI may show the last request until the tool
|
|
104
|
+
finishes. Desktop notifications show the project and tool name.
|
|
105
|
+
|
|
100
106
|
With `hooks install --answer`, Claude Code's approval questions can also be answered from here:
|
|
101
107
|
`sessions --waiting` lists them, `sessions approve <q>` / `sessions deny <q> -m <why>` answer, and
|
|
102
108
|
the browser's Sessions view has Approve and Deny (see
|
|
@@ -117,6 +123,21 @@ the project, open approval questions and the latest reply. `npx genaicode sessio
|
|
|
117
123
|
the same read-only tools over stdio. The agent can read what the other sessions said and what
|
|
118
124
|
their open questions ask. See [session tools](agents.md#asking-an-agent-about-the-sessions-sessions-mcp).
|
|
119
125
|
|
|
126
|
+
### Open the session tower
|
|
127
|
+
|
|
128
|
+
Run `npx genaicode tower` to open a terminal chat with up to five sessions waiting for you.
|
|
129
|
+
Approvals come first. Each entry includes the project and the session ID. Use `-a <agent>` to
|
|
130
|
+
choose an installed agent with MCP support, or `--web` to open the same chat in a browser.
|
|
131
|
+
|
|
132
|
+
| In the chat | What it does |
|
|
133
|
+
| ----------------- | --------------------------------------------------------------------------------------------------- |
|
|
134
|
+
| `/waiting` | Refresh the waiting queue across projects, showing up to 30 sessions from the last seven days. |
|
|
135
|
+
| `/waiting harbor` | Show waiting sessions whose project directory or repository contains `harbor`. |
|
|
136
|
+
| `/read <id>` | Preview up to three recent turns in place. A unique prefix of at least eight characters also works. |
|
|
137
|
+
|
|
138
|
+
These commands also work in `chat --sessions` and `ui --sessions`. Their results appear locally
|
|
139
|
+
in the chat. To ask the agent for a summary, type your own question with the session ID.
|
|
140
|
+
|
|
120
141
|
## Chat with an agent: terminal and browser
|
|
121
142
|
|
|
122
143
|
```bash
|
|
@@ -124,6 +145,7 @@ npx genaicode # chat in this terminal (same as "genaicode c
|
|
|
124
145
|
npx genaicode chat -a claude -C ../api
|
|
125
146
|
npx genaicode ui # the same chat in your browser, served on 127.0.0.1
|
|
126
147
|
npx genaicode chat --web # both: the terminal and the browser on one session
|
|
148
|
+
npx genaicode chat --pack claude:codex,gemini --web
|
|
127
149
|
```
|
|
128
150
|
|
|
129
151
|
In the browser, **Sessions** in the header shows what every agent on this machine has going on:
|
|
@@ -137,6 +159,18 @@ the browser (`/web off` stops the server), and Enter in the terminal that runs
|
|
|
137
159
|
`genaicode ui` brings it into that terminal. A prompt, an approval or a stop from either
|
|
138
160
|
side shows up in both.
|
|
139
161
|
|
|
162
|
+
With `--pack lead:member,member`, the lead chooses which tasks to delegate. **Pack tasks**
|
|
163
|
+
under its turn shows each worker's prompt, activity, result and changed files. The slot counter
|
|
164
|
+
includes workers waiting for approval. `--max-parallel` sets the slot limit (default 4).
|
|
165
|
+
`--max-tokens` and `--max-cost` include the lead and its workers.
|
|
166
|
+
|
|
167
|
+
In the browser, select a task or open **Approvals**. In the terminal, press Tab on an empty
|
|
168
|
+
prompt, use ↑/↓ to select a task, and press `q` for the shared approval queue. The inspector
|
|
169
|
+
shows which keys are available. You can steer a live worker, stop it, inspect a finished
|
|
170
|
+
worktree's diff, merge the changes into the lead's directory, or discard the worktree.
|
|
171
|
+
Merging leaves the files available for review and commit. Unmerged worktrees stay on disk
|
|
172
|
+
when you exit. The task controls belong to the chat process that started them.
|
|
173
|
+
|
|
140
174
|
Each prompt after the first continues the agent's own session (`AgentTask.resume`), so you
|
|
141
175
|
can say "now add a test for that". Prompts sent while the agent works wait in a queue and
|
|
142
176
|
run next; Esc stops the agent. When the agent asks for permission, `y` approves, `n`
|
package/docs/react.md
CHANGED
|
@@ -76,6 +76,7 @@ sends, Shift+Enter adds a line and Esc stops. Props:
|
|
|
76
76
|
| `state` | `SessionState`, or null before there is a session |
|
|
77
77
|
| `onSend(text)`, `onStop()` | Enter or Send; Stop or Esc. No `onStop`, no Stop button |
|
|
78
78
|
| `onApprove(id, decision)` | answers `state.approvals`; without it they show as waiting |
|
|
79
|
+
| `onPackAction(turnId, id, action)` | pass the session's `actOnPackTask` to enable worker controls and return a diff or result text |
|
|
79
80
|
| `renderPrompt(text, turn?)` | how a sent prompt shows (return null to hide it) |
|
|
80
81
|
| `renderAfterTurn(turn, i)`, `renderFooter()` | your content after a turn, or at the end |
|
|
81
82
|
| `notices` | `{ id, afterTurn, text, tone? }[]`: your own lines between turns |
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "genaicode",
|
|
3
|
-
"version": "2.
|
|
3
|
+
"version": "2.40.0",
|
|
4
4
|
"description": "Drive the coding-agent CLIs you already have (Claude Code, Codex, Copilot, Cursor, Gemini, opencode...) behind one API and one CLI, plus a thin provider-neutral LLM layer.",
|
|
5
5
|
"author": "Grzegorz Tańczyk",
|
|
6
6
|
"repository": {
|