@mndrk/agx 1.4.3 → 1.4.8

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 CHANGED
@@ -1,6 +1,6 @@
1
1
  # agx
2
2
 
3
- Unified AI Agent CLI with persistent memory. Wraps Claude, Gemini, and Ollama with automatic state management via [mem](https://github.com/ramarlina/memx).
3
+ Autonomous AI agents. One command, works until done.
4
4
 
5
5
  ```bash
6
6
  npm install -g @mndrk/agx
@@ -9,166 +9,82 @@ npm install -g @mndrk/agx
9
9
  ## Quick Start
10
10
 
11
11
  ```bash
12
- # Simple prompt
13
- agx claude -p "explain this code"
12
+ agx -a -p "Build a REST API with auth"
13
+ # Created task: build-rest-api
14
+ # ✓ Daemon started (wakes every 15m)
15
+ # ✓ Working...
16
+ ```
14
17
 
15
- # Use default provider
16
- agx -p "what does this function do?"
18
+ That's it. The agent continues working autonomously until complete.
17
19
 
18
- # With persistent memory (auto-detected)
19
- agx claude -p "continue working on the todo app"
20
+ ## How It Works
20
21
 
21
- # Autonomous mode - creates task and works until done
22
- agx claude --autonomous -p "Build a todo app with React"
23
- ```
22
+ The `-a` flag starts an autonomous task:
23
+ 1. Creates a task from your prompt
24
+ 2. Starts background daemon
25
+ 3. Agent works on the task
26
+ 4. Wakes every 15 minutes to continue
27
+ 5. Stops when agent outputs `[done]` or `[blocked]`
24
28
 
25
- ## Memory Integration
29
+ No manual task management. No babysitting. Just results.
26
30
 
27
- agx integrates with [mem](https://github.com/ramarlina/memx) for persistent state across sessions:
31
+ ## Checking Progress
28
32
 
29
33
  ```bash
30
- # If ~/.mem has a task mapped to cwd, context is auto-loaded
31
- cd ~/Projects/my-app
32
- agx claude -p "continue" # Knows where it left off
33
-
34
- # Create task with explicit criteria
35
- agx claude --task todo-app \
36
- --criteria "CRUD working" \
37
- --criteria "Tests passing" \
38
- --criteria "Deployed to Vercel" \
39
- -p "Build a todo app"
34
+ agx status # Current task
35
+ agx tasks # All tasks
36
+ agx progress # % complete
37
+ agx daemon logs # Recent activity
40
38
  ```
41
39
 
42
- ## Output Markers
40
+ ## One-Shot Mode
43
41
 
44
- Agents control state via markers in their output:
42
+ For quick questions without persistence:
45
43
 
46
- ```
47
- [checkpoint: Hero section complete] # Save progress
48
- [learn: Tailwind is fast] # Record learning
49
- [next: Add auth system] # Set next step
50
- [criteria: 2] # Mark criterion #2 done
51
- [approve: Deploy to production?] # Halt for approval
52
- [blocked: Need API key from client] # Mark stuck
53
- [pause] # Stop, resume later
54
- [continue] # Keep going (daemon)
55
- [done] # Task complete
56
- [split: auth "Handle authentication"] # Create subtask
44
+ ```bash
45
+ agx -p "explain this error"
46
+ agx claude -p "refactor this function"
57
47
  ```
58
48
 
59
49
  ## Providers
60
50
 
61
- | Provider | Aliases | Description |
62
- |----------|---------|-------------|
63
- | claude | c, cl | Anthropic Claude Code |
64
- | gemini | g, gem | Google Gemini CLI |
65
- | ollama | o, ol | Local Ollama models |
66
-
67
- ## Options
51
+ | Provider | Alias | Description |
52
+ |----------|-------|-------------|
53
+ | claude | c | Anthropic Claude Code |
54
+ | gemini | g | Google Gemini |
55
+ | ollama | o | Local Ollama |
68
56
 
69
- ```
70
- --prompt, -p <text> Prompt to send
71
- --model, -m <name> Model name
72
- --yolo, -y Skip permission prompts
73
- --print Non-interactive output
74
- --interactive, -i Force interactive mode
75
- --mem Enable mem integration (auto-detected)
76
- --no-mem Disable mem integration
77
- --autonomous, -a Create task and run autonomously (starts daemon)
78
- --task <name> Specific task name
79
- --criteria <text> Success criterion (repeatable)
80
- ```
81
-
82
- ## Claude Code Plugin
57
+ ## Output Markers
83
58
 
84
- Install as a Claude Code plugin:
59
+ Agents control their state via markers:
85
60
 
86
- ```bash
87
- claude plugin install github:ramarlina/agx
88
61
  ```
89
-
90
- This adds:
91
- - **Skill**: Claude learns how to spawn background agents
92
- - **Commands**: `/agx:spawn <goal>`, `/agx:continue`
93
-
94
- ## Commands
95
-
96
- ```bash
97
- agx init # Setup wizard
98
- agx config # Configuration menu
99
- agx status # Show current config
100
- agx skill # View LLM skill
101
- agx skill install # Install skill to Claude/Gemini
102
-
103
- # Daemon management
104
- agx daemon start # Start background daemon
105
- agx daemon stop # Stop daemon
106
- agx daemon status # Check if running
107
- agx daemon logs # Show recent logs
62
+ [checkpoint: Built login page] # Save progress
63
+ [learn: JWT works better here] # Record insight
64
+ [next: Add signup flow] # Set next step
65
+ [done] # Task complete
66
+ [blocked: Need API key] # Can't proceed
108
67
  ```
109
68
 
110
- ## Autonomous Mode
69
+ ## Manual Control (Optional)
111
70
 
112
- Start a task that runs autonomously until complete:
71
+ For power users who want fine control:
113
72
 
114
73
  ```bash
115
- agx claude --autonomous -p "Build a React todo app with auth"
116
- # Created task: build-react-todo
117
- # Mapped: ~/Projects/app → task/build-react-todo
118
- # Daemon started (pid 12345)
119
- # ✓ Autonomous mode: daemon will continue work every 15m
120
- ```
121
-
122
- The daemon:
123
- - Runs in background (survives terminal close)
124
- - Wakes every 15 minutes
125
- - Continues work on active tasks
126
- - Stops when task is `[done]` or `[blocked]`
127
-
128
- ## Loop Control
129
-
130
- The agent controls execution flow via markers:
131
-
132
- - `[done]` → Task complete, exit
133
- - `[pause]` → Save state, exit (resume later with same command)
134
- - `[blocked: reason]` → Mark stuck, notify human, exit
135
- - `[continue]` → Keep going (daemon mode loops)
136
- - `[approve: question]` → Halt until human approves
137
-
138
- ## Task Splitting
139
-
140
- Break large tasks into subtasks:
141
-
142
- ```
143
- Agent output:
144
- This is too big. Breaking it down.
145
-
146
- [split: setup "Project scaffolding"]
147
- [split: auth "Authentication system"]
148
- [split: crud "CRUD operations"]
149
- [next: Start with setup subtask]
150
- [pause]
74
+ agx run [task] # Run task now
75
+ agx pause [task] # Pause scheduled runs
76
+ agx stop [task] # Mark done
77
+ agx stuck <reason> # Mark blocked
151
78
  ```
152
79
 
153
- agx creates subtask branches in ~/.mem linked to the parent.
80
+ ## Setup
154
81
 
155
- ## Example: Full Workflow
82
+ First run walks you through setup:
156
83
 
157
84
  ```bash
158
- # Day 1: Start project
159
- mkdir ~/Projects/my-app && cd ~/Projects/my-app
160
- agx claude --auto-task -p "Build a React todo app with auth"
161
-
162
- # Agent works, outputs markers
163
- # [checkpoint: Scaffolded with Vite]
164
- # [learn: Vite is faster than CRA]
165
- # [next: Add todo list component]
166
- # [pause]
167
-
168
- # Day 2: Continue
169
- cd ~/Projects/my-app
170
- agx claude -p "continue"
171
- # Context auto-loaded, agent picks up where it left off
85
+ agx setup # Install providers, set defaults
86
+ agx add claude # Add a provider
87
+ agx login claude # Authenticate
172
88
  ```
173
89
 
174
90
  ## License