ineedcodes 0.1.1 → 0.1.2
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 +64 -103
- package/package.json +7 -2
- package/packages/cli/dist/index.js +11384 -515
package/README.md
CHANGED
|
@@ -2,160 +2,121 @@
|
|
|
2
2
|
|
|
3
3
|
<img src="docs/assets/ineed-logo.svg" alt="ineed: your terminal, now autonomous" width="640"/>
|
|
4
4
|
|
|
5
|
-
|
|
5
|
+
**ineed**
|
|
6
6
|
|
|
7
|
-
[](https://ineed.web.id)
|
|
7
|
+
[](https://www.npmjs.com/package/ineedcodes)
|
|
9
8
|
[](LICENSE)
|
|
10
9
|
[](https://nodejs.org)
|
|
11
10
|
[]()
|
|
12
11
|
|
|
13
|
-
|
|
12
|
+
[ineed.codes](https://ineed.codes) · [ineed.web.id](https://ineed.web.id) · [Contributing](CONTRIBUTING.md)
|
|
14
13
|
|
|
15
14
|
</div>
|
|
16
15
|
|
|
17
16
|
---
|
|
18
17
|
|
|
19
|
-
|
|
20
|
-
|
|
21
|
-
|
|
18
|
+
## What is ineed?
|
|
19
|
+
|
|
20
|
+
A CLI AI agent. You say what you want in plain language. It does the work:
|
|
21
|
+
reads code, runs commands, edits files, uses Git, runs tests, and shows you
|
|
22
|
+
real proof that it worked.
|
|
22
23
|
|
|
23
24
|
```bash
|
|
24
25
|
ineed "fix the build errors in this project"
|
|
25
26
|
```
|
|
26
27
|
|
|
27
|
-
|
|
28
|
-
|
|
29
|
-
|
|
30
|
-
Most AI coding tools are built for people who already know Git, terminals,
|
|
31
|
-
and debugging. **ineed flips that**: it is built so that *beginners can learn
|
|
32
|
-
by watching an agent work*.
|
|
28
|
+
Made for beginners. Every step is narrated in plain English, nothing is
|
|
29
|
+
hidden, and success is never claimed without evidence. Watch how it works,
|
|
30
|
+
learn the workflow: that is vibecoding.
|
|
33
31
|
|
|
34
|
-
|
|
35
|
-
came out of it.
|
|
36
|
-
- Nothing is hidden. You see the real commands, the real diffs, the real
|
|
37
|
-
test output. Never fabricated.
|
|
38
|
-
- After finishing, the agent explains what happened and one thing worth learning.
|
|
39
|
-
- If something confuses you, just ask: *"explain that more simply."*
|
|
32
|
+
## Install
|
|
40
33
|
|
|
41
|
-
|
|
42
|
-
|
|
34
|
+
```bash
|
|
35
|
+
npm install -g ineedcodes
|
|
36
|
+
```
|
|
43
37
|
|
|
44
|
-
|
|
45
|
-
## Quick start
|
|
38
|
+
Package name is `ineedcodes`. The command you type is `ineed`.
|
|
46
39
|
|
|
47
|
-
|
|
40
|
+
## First run
|
|
48
41
|
|
|
49
42
|
```bash
|
|
50
|
-
|
|
51
|
-
cd ineedcodes
|
|
52
|
-
npm install
|
|
53
|
-
npm run build
|
|
54
|
-
npm link
|
|
43
|
+
ineed
|
|
55
44
|
```
|
|
56
45
|
|
|
57
|
-
|
|
46
|
+
A setup wizard asks three things: pick a provider, paste an API key, choose
|
|
47
|
+
a model from a live list. Saved once in `~/.ineedcodes/config.json`. Any
|
|
48
|
+
OpenAI-compatible API works: OpenAI, iNeed API, OmniRoute, LM Studio,
|
|
49
|
+
Ollama, vLLM, and more.
|
|
50
|
+
|
|
51
|
+
## Use it
|
|
58
52
|
|
|
59
53
|
```bash
|
|
60
|
-
|
|
54
|
+
ineed # interactive session in this folder
|
|
55
|
+
ineed "explain this repo" # one-shot task
|
|
56
|
+
ineed --resume # continue your last session
|
|
61
57
|
```
|
|
62
58
|
|
|
63
|
-
|
|
64
|
-
choose a model from a live list. Saved once, done forever.
|
|
65
|
-
|
|
66
|
-
Any OpenAI-compatible endpoint works: OpenAI, iNeed API, OmniRoute,
|
|
67
|
-
LM Studio, Ollama, vLLM, …
|
|
59
|
+
Then just talk:
|
|
68
60
|
|
|
69
|
-
|
|
70
|
-
|
|
61
|
+
- *Fix the authentication bug.*
|
|
62
|
+
- *Get this GitHub project running.*
|
|
63
|
+
- *Why is this API failing? Fix it and prove it.*
|
|
64
|
+
- *Make this README sound human.*
|
|
71
65
|
|
|
72
|
-
|
|
66
|
+
## Learn while you work
|
|
73
67
|
|
|
74
68
|
| You type | You learn |
|
|
75
69
|
|---|---|
|
|
76
|
-
|
|
|
70
|
+
| a plain-language goal | how agents plan and verify work |
|
|
71
|
+
| `/plan` then `/build` | why reading before changing matters |
|
|
77
72
|
| `/context` | how token budgets and compaction work |
|
|
78
|
-
| `/plan` before `/build` | why reading before changing matters |
|
|
79
73
|
| `/think high` on a hard bug | how reasoning depth changes results |
|
|
74
|
+
| `/init` | how AGENTS.md instructions shape an agent |
|
|
80
75
|
| `/doctor` | what a healthy setup looks like |
|
|
81
|
-
| `/init` | how project instructions (AGENTS.md) shape agent behavior |
|
|
82
|
-
| watching the tool output | real commands, real diffs, real evidence |
|
|
83
|
-
|
|
84
|
-
Explanations default to beginner-friendly. Change anytime:
|
|
85
|
-
|
|
86
|
-
```bash
|
|
87
|
-
/explain beginner # more plain-English narration
|
|
88
|
-
/explain balanced
|
|
89
|
-
/explain terse # for when you already know what's happening
|
|
90
|
-
```
|
|
91
|
-
|
|
92
|
-
> 📚 **New to AI coding?** Free tutorials on how to use AI tools, plus a
|
|
93
|
-
> curated list of free AI resources, live at **[ineed.web.id](https://ineed.web.id)**.
|
|
94
|
-
> No signup, no paywall.
|
|
95
76
|
|
|
96
|
-
|
|
97
|
-
|
|
98
|
-
|
|
99
|
-
ineed # interactive session here
|
|
100
|
-
ineed "explain this repo" # one-shot
|
|
101
|
-
ineed --resume # continue your last session
|
|
102
|
-
```
|
|
103
|
-
|
|
104
|
-
Then just talk:
|
|
105
|
-
|
|
106
|
-
- *Fix the authentication bug.*
|
|
107
|
-
- *Install this GitHub project and get it running.*
|
|
108
|
-
- *Why is this API failing? Then fix it and prove it.*
|
|
109
|
-
- *Make this README sound less AI-generated.*
|
|
110
|
-
|
|
111
|
-
Slash commands are optional shortcuts (`/help`, `/model`, `/context`, `/plan`,
|
|
112
|
-
`/build`, `/think`, `/memory`, `/skills`, `/permissions`, `/init`, `/doctor`,
|
|
113
|
-
`/resume`, `/clear`, `/copy`, `/exit`). Plain language always works.
|
|
77
|
+
Slash commands are optional. Plain language always works:
|
|
78
|
+
`/help` `/model` `/plan` `/build` `/think` `/context` `/memory` `/skills`
|
|
79
|
+
`/permissions` `/init` `/doctor` `/resume` `/clear` `/copy` `/exit`
|
|
114
80
|
|
|
115
|
-
|
|
81
|
+
> New to AI coding? Free tutorials and a curated list of free AI resources:
|
|
82
|
+
> [ineed.web.id](https://ineed.web.id). No signup, no paywall.
|
|
116
83
|
|
|
117
|
-
|
|
118
|
-
Natural language
|
|
119
|
-
↓
|
|
120
|
-
Lead Agent (understands objective, recalls ICM memory, picks skills)
|
|
121
|
-
↓
|
|
122
|
-
Iterative loop: reason → tool call → observe → adapt → verify
|
|
123
|
-
↓
|
|
124
|
-
Real execution: files, shell, git, web, MCP tools
|
|
125
|
-
↓
|
|
126
|
-
Evidence-based result: what changed, what ran, what passed
|
|
127
|
-
```
|
|
84
|
+
## What is inside
|
|
128
85
|
|
|
129
|
-
- **
|
|
86
|
+
- **Agent runtime**: an iterative loop: reason, call a tool, observe the real
|
|
87
|
+
result, adapt, verify. Plan mode (read only) and build mode. Reasoning
|
|
88
|
+
effort from low to high.
|
|
89
|
+
- **Tools**: filesystem, search (ripgrep with fallback), shell, Git, web, and
|
|
90
|
+
a native MCP client. Tools take JSON, return JSON.
|
|
91
|
+
- **Memory**: [ICM](https://github.com/rtk-ai/icm), one database per project.
|
|
130
92
|
Conventions, solved errors, and decisions survive across sessions.
|
|
131
|
-
- **Skills**: portable `SKILL.md` files. Built-ins: `boost` (isolated worktree
|
|
132
|
-
independent verification) and `humanizer` (natural prose rewriting).
|
|
133
|
-
- **
|
|
134
|
-
|
|
135
|
-
- **
|
|
136
|
-
flows freely; destructive, remote, and secret operations need approval.
|
|
137
|
-
- **Secrets**: `.env`, SSH keys, tokens are never dumped into model context.
|
|
93
|
+
- **Skills**: portable `SKILL.md` files. Built-ins: `boost` (isolated worktree
|
|
94
|
+
plus independent verification) and `humanizer` (natural prose rewriting).
|
|
95
|
+
- **Permissions**: `ask`, `workspace`, or `autonomous`. Destructive, remote,
|
|
96
|
+
and secret operations always need approval.
|
|
97
|
+
- **Secrets**: `.env` files, SSH keys, and tokens never reach model context.
|
|
138
98
|
- **Providers**: any OpenAI-compatible API. No lock-in.
|
|
139
99
|
|
|
140
|
-
##
|
|
100
|
+
## Project layout
|
|
141
101
|
|
|
142
102
|
```
|
|
143
103
|
packages/
|
|
144
|
-
core/ tools, LLM providers, ICM adapter, permissions, context,
|
|
145
|
-
agent/ agent loop, lead agent, orchestrator, boost
|
|
146
|
-
cli/
|
|
104
|
+
core/ tools, LLM providers, ICM adapter, permissions, context, skills
|
|
105
|
+
agent/ agent loop, lead agent, orchestrator, boost
|
|
106
|
+
cli/ TUI, setup wizard, slash commands
|
|
147
107
|
```
|
|
148
108
|
|
|
149
109
|
## Contributing
|
|
150
110
|
|
|
151
|
-
This project
|
|
152
|
-
See [CONTRIBUTING.md](CONTRIBUTING.md).
|
|
153
|
-
explanations, new built-in skills, provider adapters.
|
|
111
|
+
This project teaches beginners, so contributions should keep that spirit.
|
|
112
|
+
See [CONTRIBUTING.md](CONTRIBUTING.md).
|
|
154
113
|
|
|
155
|
-
|
|
114
|
+
---
|
|
156
115
|
|
|
157
|
-
|
|
116
|
+
<div align="center">
|
|
158
117
|
|
|
159
|
-
|
|
118
|
+
MIT licensed.
|
|
119
|
+
[ineed.codes](https://ineed.codes) is the product.
|
|
120
|
+
[ineed.web.id](https://ineed.web.id) is where you learn AI, free.
|
|
160
121
|
|
|
161
122
|
</div>
|
package/package.json
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "ineedcodes",
|
|
3
|
-
"version": "0.1.
|
|
3
|
+
"version": "0.1.2",
|
|
4
4
|
"description": "ineed — your terminal, now autonomous. CLI-first general-purpose autonomous AI agent.",
|
|
5
5
|
"license": "MIT",
|
|
6
6
|
"type": "module",
|
|
@@ -42,5 +42,10 @@
|
|
|
42
42
|
"autonomous",
|
|
43
43
|
"coding",
|
|
44
44
|
"llm"
|
|
45
|
-
]
|
|
45
|
+
],
|
|
46
|
+
"devDependencies": {
|
|
47
|
+
"@types/node": "^22.10.2",
|
|
48
|
+
"esbuild": "^0.28.2",
|
|
49
|
+
"typescript": "^5.6.3"
|
|
50
|
+
}
|
|
46
51
|
}
|