@listeningkit/log 0.2.2-nightly.202611202102.cf338e8 → 0.3.0
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/README.md +20 -12
- package/hermesPlugin/dashboard/manifest.json +1 -1
- package/package.json +2 -2
package/README.md
CHANGED
|
@@ -3,8 +3,8 @@
|
|
|
3
3
|
</p>
|
|
4
4
|
|
|
5
5
|
<p align="center">
|
|
6
|
-
<a href="https://x.com/
|
|
7
|
-
<br><sub>Introducing log, in collaboration with Hermes Agent. <a href="https://x.com/
|
|
6
|
+
<a href="https://x.com/i/status/2120938715820474690"><img src="media/readme/hermesIntro.gif" alt="Introducing log by ListeningKit, in collaboration with Hermes Agent: Hermes asks who is waiting for a reply, log reads every inbox, and the answer lands on Discord in your own voice" width="100%" /></a>
|
|
7
|
+
<br><sub>Introducing log, in collaboration with Hermes Agent. <a href="https://x.com/i/status/2120938715820474690">Watch the launch video with sound on X →</a></sub>
|
|
8
8
|
</p>
|
|
9
9
|
|
|
10
10
|
<h3 align="center">One command for every social account you run.</h3>
|
|
@@ -152,7 +152,10 @@ Then ask in plain words: *"read my X DMs and tell me who's waiting on me"*. log'
|
|
|
152
152
|
## Highlights
|
|
153
153
|
|
|
154
154
|
- **Start from a play.** Each [use case](docs/useCases/README.md) is a workflow written step by step: what to run, and what tells you each step worked. With `--skills all`, your agent picks the right one from what you say.
|
|
155
|
-
- **Your voice, checked.** A draft is read against how you really write to that person,
|
|
155
|
+
- **Your voice, checked.** A draft is read against how you really write to that person, grounded in your own past replies, and scored once against a bar set on your own messages. It's also checked for hidden characters, and for an agent copying the other person's style instead of yours ([#238](https://github.com/matthewdonsemail-lab/log/pull/238)).
|
|
156
|
+
- **AI writing caught, trained on you.** A small detector learns the difference between your messages and drafts that frontier models wrote for your chats, with a separate bar for short, medium and long messages. It runs on your machine in about 6 ms; `log voice train` builds it ([#245](https://github.com/matthewdonsemail-lab/log/pull/245)).
|
|
157
|
+
- **Who each person is, and where they are.** Every chat is labelled with who the person is (community, lead, client, partner, friend, bot) and where they are in your sales cycle. A small model labels them on your CPU, and asks a larger model only when it's unsure. The same person on different platforms is suggested as one, and you approve each link ([#242](https://github.com/matthewdonsemail-lab/log/pull/242), [#249](https://github.com/matthewdonsemail-lab/log/pull/249)).
|
|
158
|
+
- **What actually gets replies.** `log inbox outcomes` compares the reply rate of messages log sent with yours, matched by person and stage, and shows which plays work ([#246](https://github.com/matthewdonsemail-lab/log/pull/246)).
|
|
156
159
|
- **One session per account.** Each account keeps one connection open, the way the platform's own app does, never a new one per command.
|
|
157
160
|
- **Live inbox.** New messages are pushed into your agent's context as they arrive, and two agents never answer the same chat.
|
|
158
161
|
- **Kept human.** Every account has its own daily limits, pacing and local waking hours. A refusal from the platform stops the run, and nothing is retried.
|
|
@@ -428,15 +431,17 @@ A video with no sound, like motion graphics, says `no sound` and is told by what
|
|
|
428
431
|
|
|
429
432
|
### The models it uses
|
|
430
433
|
|
|
431
|
-
|
|
434
|
+
Five models run on your own machine, with no key. Four are hosted, and each needs its key:
|
|
432
435
|
|
|
433
436
|
| Model | Runs | What it does |
|
|
434
437
|
|---|---|---|
|
|
435
438
|
| all-MiniLM-L6-v2 (23 MB) | your machine | meaning search over every conversation, your voice index, matching what you typed to a play |
|
|
436
439
|
| LFM2.5-VL-450M (352 MB) | your machine | describes every picture and video frame as text |
|
|
437
440
|
| wolf-defender prompt injection small | your machine | the judge's model: flags prompt injection in what comes in |
|
|
438
|
-
|
|
|
439
|
-
|
|
|
441
|
+
| Your AI-writing detector | your machine, trained by `log voice train` | the voice gate's AI reading: your messages against drafts frontier models wrote for your chats, a bar per message length |
|
|
442
|
+
| Your pipeline labeller | your machine, trained by `log inbox label train` | who each person is and where they are in your sales cycle, on MiniLM's vectors |
|
|
443
|
+
| SubQ `subq-snap-1.1` | hosted, `SUBQ_API_KEY` | the voice gate: does a draft sound like you; and the teacher for the labeller, asked when the local one is unsure |
|
|
444
|
+
| Jina embeddings v5 and reranker v3.5 | hosted, `JINA_API_KEY` | ranking a chat's history for the gate (capped at 4 s, with a local fallback), and a meaning vector for each piece of media |
|
|
440
445
|
| AssemblyAI | hosted, `ASSEMBLYAI_API_KEY` | what was said in a voice note or a video |
|
|
441
446
|
| Inworld or Fish Audio | hosted, `INWORLD_API_KEY` or `FISH_API_KEY` | a voice note in your cloned voice, only when you ask |
|
|
442
447
|
|
|
@@ -459,7 +464,8 @@ flowchart LR
|
|
|
459
464
|
**Building the indexes** on a new machine, in this order:
|
|
460
465
|
1. `log search index` reads every conversation once and builds the keyword and meaning indexes.
|
|
461
466
|
2. `log voice index` builds your voice from what step 1 saved.
|
|
462
|
-
3. `log voice
|
|
467
|
+
3. `log voice train` trains your AI-writing detector.
|
|
468
|
+
4. `log inbox label teach --apply`, then `log inbox label train`, labels your chats and trains the local labeller.
|
|
463
469
|
|
|
464
470
|
After that, new messages are added as they arrive, and pictures are described as they land.
|
|
465
471
|
|
|
@@ -530,21 +536,23 @@ What's built, what's next and what's planned is kept in one pinned issue: **[Roa
|
|
|
530
536
|
| The record ([soron](docs/proposals/soron.md), [#197](https://github.com/matthewdonsemail-lab/log/issues/197)) | search by DM, group or channel | keeping edits and deletions, capturing the conversation around a personal mention (kept after a ban), judging each message the moment it arrives |
|
|
531
537
|
| Turn awareness ([spec](https://github.com/matthewdonsemail-lab/log/blob/turnAware/docs/proposals/turnAware.md)) | typing on Discord and Slack | X Chat |
|
|
532
538
|
| Tools | tool health | recording which tools the router offered, so unused tools can be told apart from unseen ones |
|
|
533
|
-
| Your day | | a morning debrief: who is waiting, what you owe, what was held overnight |
|
|
534
|
-
| Voice | cloned voice notes | voice-note replies in your voice, then calls, with your yes per call |
|
|
539
|
+
| Your day | the hottest deals in your DMs, from the pipeline labels | a morning debrief: who is waiting, what you owe, what was held overnight |
|
|
540
|
+
| Voice | cloned voice notes; the grounding pack and one calibrated score; an AI-writing detector trained on you | a higher catch rate on short messages, voice-note replies in your voice, then calls, with your yes per call |
|
|
541
|
+
| Pipeline ([#236](https://github.com/matthewdonsemail-lab/log/issues/236)) | who and where per chat on the CPU, the larger model when unsure, one person across platforms (suggested), reply outcomes | your own labels to measure the labeller against, the next step suggested per stage |
|
|
535
542
|
| Browser | human mouse: Fitts's-law timing, corrective moves, tremor ([#201](https://github.com/matthewdonsemail-lab/log/pull/201)) | sign-ins moving from AdsPower to Camoufox, browser-fingerprinted HTTP for X and Discord |
|
|
536
543
|
| Every harness ([spec](https://github.com/matthewdonsemail-lab/log/blob/genuiEverywhere/docs/proposals/genuiEverywhere.md), [#196](https://github.com/matthewdonsemail-lab/log/pull/196)) | views in Cursor, Claude and ChatGPT where the host draws MCP Apps | a ChatGPT connector, a smaller inbox tool profile, a Claude Desktop extension, a render test on every PR |
|
|
537
|
-
| Speed | | the prompt hook under 150 ms |
|
|
544
|
+
| Speed | hooks off the slow paths, one push loop per session, Discord catch-up in 4 calls, watchers that stay up | the prompt hook under 150 ms |
|
|
538
545
|
|
|
539
|
-
Built
|
|
546
|
+
Built means merged into `main`. It reaches npm in the next release (see [Releases](#releases)).
|
|
540
547
|
|
|
541
548
|
## Releases
|
|
542
549
|
|
|
543
|
-
Newest first. Each release pins the local models it runs with (`log models`), and `log update` moves code, harness plugins and models to the release together ([docs/updates.md](docs/updates.md)).
|
|
550
|
+
Newest first. Work merged into `main` collects in the open release PR, and merging that PR tags the version and publishes it to npm. Each release pins the local models it runs with (`log models`), and `log update` moves code, harness plugins and models to the release together ([docs/updates.md](docs/updates.md)).
|
|
544
551
|
|
|
545
552
|
<!-- releases:start: written by `bun run docs:releases` from CHANGELOG.md and src/models/manifest, not edited by hand -->
|
|
546
553
|
| Version | Date | What's in it | Models |
|
|
547
554
|
| --- | --- | --- | --- |
|
|
555
|
+
| [0.3.0](https://github.com/matthewdonsemail-lab/log/pull/212) | 2026-11-20 | one live browser view for console and chat, hidden window when streamed (+41 more) | first pinned: wolf-defender-prompt-injection-small@2484cc9, all-MiniLM-L6-v2@751bff3, roberta-base-go_emotions-onnx@90ee0c1, LFM2.5-VL-450M-ONNX@24e6300, myuri@5.0.0 |
|
|
548
556
|
| [0.2.1](https://github.com/matthewdonsemail-lab/log/pull/179) | 2026-11-20 | The relate graph, the live inbox push (messages during tool calls, burst batching, late copies dropped), inbox labels and ignoring a server or chat (+2 more) | not pinned (newest by name) |
|
|
549
557
|
| [0.2.0](https://github.com/matthewdonsemail-lab/log/pull/166) | 2026-11-15 | The one-command Hermes pack, the catalog review, the README rewrite and the catch-up machine (+1 more) | not pinned (newest by name) |
|
|
550
558
|
<!-- releases:end -->
|
|
@@ -3,7 +3,7 @@
|
|
|
3
3
|
"label": "log",
|
|
4
4
|
"description": "Your accounts, who is waiting on a reply, and what log sent today, read from log's own files on this machine.",
|
|
5
5
|
"icon": "MessageSquare",
|
|
6
|
-
"version": "0.
|
|
6
|
+
"version": "0.3.0",
|
|
7
7
|
"tab": {
|
|
8
8
|
"path": "/log",
|
|
9
9
|
"position": "after:skills"
|
package/package.json
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "@listeningkit/log",
|
|
3
|
-
"version": "0.
|
|
4
|
-
"description": "ListeningKit's go-to-market tool, open-sourced: one CLI and MCP server that runs your accounts on X, Reddit,
|
|
3
|
+
"version": "0.3.0",
|
|
4
|
+
"description": "ListeningKit's go-to-market tool, open-sourced: one CLI and MCP server that runs your accounts on X, Discord, Slack, Instagram, Reddit, LinkedIn and Facebook from your terminal or any agent, self-hosted.",
|
|
5
5
|
"type": "module",
|
|
6
6
|
"bin": {
|
|
7
7
|
"log": "bin/log.ts",
|