deepseek-coder-agent-cli 1.0.52 → 1.0.53
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 +38 -0
- package/package.json +1 -1
package/README.md
CHANGED
|
@@ -121,6 +121,44 @@ At that point, Codex is just an expensive way to type commands you could have ty
|
|
|
121
121
|
|
|
122
122
|
Maybe this makes sense for students at OpenAI Academy who are learning what `npm install` does. For any serious coder shipping production code every day - managing deployments, debugging infrastructure, running CI/CD pipelines - a sandboxed agent that can't touch the network is fucking useless.
|
|
123
123
|
|
|
124
|
+
### 2. Anti-Hallucination "Next Steps" Requirements
|
|
125
|
+
|
|
126
|
+
DeepSeek CLI has strong baked-in anti-hallucination requirements for final completion messages. When the AI says it's done, it must provide concrete, verifiable next steps.
|
|
127
|
+
|
|
128
|
+
**Claude Code (as of v2.1.19 at https://www.npmjs.com/package/@anthropic-ai/claude-code) fully lacks this.**
|
|
129
|
+
|
|
130
|
+
What happens without anti-hallucination next steps:
|
|
131
|
+
```
|
|
132
|
+
Claude Code: "I've implemented the authentication system. The code should work now."
|
|
133
|
+
You: *runs code* Error: undefined is not a function
|
|
134
|
+
You: It doesn't work
|
|
135
|
+
Claude Code: "I apologize, let me fix that."
|
|
136
|
+
You: *runs code* Error: cannot read property 'user' of null
|
|
137
|
+
You: Still broken
|
|
138
|
+
Claude Code: "I see the issue now..."
|
|
139
|
+
[repeat 5 more times]
|
|
140
|
+
```
|
|
141
|
+
|
|
142
|
+
What happens with DeepSeek's enforced next steps:
|
|
143
|
+
```
|
|
144
|
+
DeepSeek: "Implementation complete. Next steps to verify:
|
|
145
|
+
1. Run `npm test` - expect all 12 tests passing
|
|
146
|
+
2. Run `npm run build` - expect no TypeScript errors
|
|
147
|
+
3. Test login at http://localhost:3000/login with test@example.com / password123
|
|
148
|
+
4. Check console for 'Auth initialized' message"
|
|
149
|
+
|
|
150
|
+
You: *runs npm test* ✓ 12 tests passing
|
|
151
|
+
You: *runs npm run build* ✓ No errors
|
|
152
|
+
You: It actually works
|
|
153
|
+
```
|
|
154
|
+
|
|
155
|
+
For any serious coder - not a stupid computer science student at Anthropic Academy - verifiable completion criteria are **fully required**. You need to know:
|
|
156
|
+
- What commands to run to verify the work
|
|
157
|
+
- What output to expect
|
|
158
|
+
- What to check if something fails
|
|
159
|
+
|
|
160
|
+
An AI that just says "done, should work now" with no verification steps is an AI that hallucinates success. DeepSeek CLI forces the model to commit to specific, testable claims about what it just did.
|
|
161
|
+
|
|
124
162
|
### The o4-mini Potential
|
|
125
163
|
|
|
126
164
|
If adapted for DeepSeek CLI, o4-mini could offer the same reasoning capabilities as Codex CLI 5.2 xhigh but without the sandbox prison. The insights from making o4-mini work in an unrestricted environment would benefit all coding agent development - you learn what's actually possible when you remove artificial limitations.
|
package/package.json
CHANGED