pi-caveman 1.0.5 → 1.0.7
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 +10 -0
- package/extensions/caveman.ts +3 -3
- package/package.json +3 -3
package/README.md
CHANGED
|
@@ -84,6 +84,16 @@ When active, a status bar displays caveman level and an animated campfire flicke
|
|
|
84
84
|
|
|
85
85
|
The extension hooks `before_agent_start` to append caveman communication rules to the system prompt at the selected intensity. Within a session, the active level is stored as a custom session entry and restored on resume. Across sessions, persistent config (`~/.pi/agent/caveman.json`) provides the default level and status bar preference. Auto-clarity rules tell the model to drop caveman mode for security warnings or irreversible actions.
|
|
86
86
|
|
|
87
|
+
## Warning
|
|
88
|
+
|
|
89
|
+
Caveman mode only affects _output tokens_, not chain-of-thought tokens, input tokens, file read/write tool calls, etc. If your agent is in a huge codebase and you've given it a complex task, don't expect much token reduction.
|
|
90
|
+
|
|
91
|
+
<p align="center">
|
|
92
|
+
<img src="https://raw.githubusercontent.com/jonjonrankin/pi-caveman/main/shoutout.jpg" alt="pi-caveman glowing review" width="600">
|
|
93
|
+
</p>
|
|
94
|
+
|
|
95
|
+
But for the cost-conscious, every token counts ;)
|
|
96
|
+
|
|
87
97
|
## Credits
|
|
88
98
|
|
|
89
99
|
Based on [caveman](https://github.com/JuliusBrussee/caveman) by [Julius Brussee](https://github.com/JuliusBrussee).
|
package/extensions/caveman.ts
CHANGED
|
@@ -13,9 +13,9 @@
|
|
|
13
13
|
import { readFile, writeFile, mkdir } from "node:fs/promises";
|
|
14
14
|
import { homedir } from "node:os";
|
|
15
15
|
import { join } from "node:path";
|
|
16
|
-
import type { ExtensionAPI, ExtensionContext } from "@
|
|
17
|
-
import { getSettingsListTheme } from "@
|
|
18
|
-
import { Container, type SettingItem, SettingsList, Text } from "@
|
|
16
|
+
import type { ExtensionAPI, ExtensionContext } from "@earendil-works/pi-coding-agent";
|
|
17
|
+
import { getSettingsListTheme } from "@earendil-works/pi-coding-agent";
|
|
18
|
+
import { Container, type SettingItem, SettingsList, Text } from "@earendil-works/pi-tui";
|
|
19
19
|
|
|
20
20
|
// ---------------------------------------------------------------------------
|
|
21
21
|
// Levels
|
package/package.json
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "pi-caveman",
|
|
3
|
-
"version": "1.0.
|
|
4
|
-
"description": "Why use many token when few do trick. Caveman mode for pi
|
|
3
|
+
"version": "1.0.7",
|
|
4
|
+
"description": "Why use many token when few do trick. Caveman mode for pi \u2014 cuts ~75% output tokens while keeping full technical accuracy.",
|
|
5
5
|
"keywords": [
|
|
6
6
|
"pi-package",
|
|
7
7
|
"pi-extension",
|
|
@@ -27,6 +27,6 @@
|
|
|
27
27
|
"image": "https://raw.githubusercontent.com/jonjonrankin/pi-caveman/main/pi-caveman.gif"
|
|
28
28
|
},
|
|
29
29
|
"peerDependencies": {
|
|
30
|
-
"@
|
|
30
|
+
"@earendil-works/pi-coding-agent": "*"
|
|
31
31
|
}
|
|
32
32
|
}
|