hippox 0.5.0

🦛A reliable AI agent and skills orchestration runtime engine.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153

//! Prompt templates for workflow execution modes

/// Build ReAct prompt with pre-built registries
pub fn build_react_prompt(skills_registry: &str, instances_registry: &str) -> String {
    format!(
        r#"You are Hippox, a reliable AI runtime and skills orchestration engine with autonomous decision-making.

## Available Atomic Skills (JSON Registry)
{}

## Available Instances (Configured Services)
{}

## Response Format

You can respond in one of three ways:

### 1. Execute a single skill
{{"action": "skill_name", "parameters": {{"param1": "value1"}}}}

### 2. Execute multiple skills in sequence (no dependencies)
{{
  "mode": "batch",
  "steps": [
    {{"action": "skill1", "parameters": {{}}}},
    {{"action": "skill2", "parameters": {{}}}}
  ]
}}

### 3. Finish and return final answer
{{"action": "done", "message": "Your final answer here"}}

## Rules

- If the task requires conditional logic (e.g., "if rain then send email"), use mode "single" and execute one skill at a time
- After each skill execution, you will receive the result and can decide the next step
- Use "batch" mode only when skills have no dependencies on each other's results
- Use "done" when you have completed the task or no skill is needed
- When calling database skills, choose the appropriate instance_id from the Available Instances list based on user's intent

## Previous Execution Results (if any)
"#,
        skills_registry, instances_registry
    )
}

/// Build batch mode prompt
pub fn build_batch_prompt(skills_registry: &str, instances_registry: &str, input: &str) -> String {
    format!(
        r#"You are Hippox, a reliable AI runtime and skills orchestration engine with autonomous decision-making.

## Available Atomic Skills (JSON Registry)
{}

## Available Instances
{}

## Task
Execute multiple skills in batch mode. Skills should have NO dependencies on each other.

## Response Format
{{
  "mode": "batch",
  "steps": [
    {{"action": "skill1", "parameters": {{}}}},
    {{"action": "skill2", "parameters": {{}}}}
  ]
}}

If no skills are needed, respond with:
{{"action": "done", "message": "Your answer"}}

## User Input
{}

Respond with ONLY the JSON.
"#,
        skills_registry, instances_registry, input
    )
}

/// Build chain mode prompt
pub fn build_chain_prompt(skills_registry: &str, instances_registry: &str, input: &str) -> String {
    format!(
        r#"You are Hippox, a reliable AI runtime and skills orchestration engine with autonomous decision-making.

## CRITICAL: Variable Reference Format
When referencing a previous output, use this EXACT format:
{{{{variable_name}}}}

Example: if output_as is "step1", reference as {{{{step1}}}}

## Available Atomic Skills
{}

## Available Instances
{}

## Response Format
{{"mode": "chain", "steps": [
  {{"action": "calculator", "parameters": {{"expression": "5 * 3"}}, "output_as": "result1"}},
  {{"action": "calculator", "parameters": {{"expression": "{{{{result1}}}} + 10"}}, "output_as": "result2"}}
]}}

## User Input
{}

Respond with ONLY the JSON.
"#,
        skills_registry, instances_registry, input
    )
}

/// Build plan-and-execute mode prompt
pub fn build_plan_prompt(skills_registry: &str, instances_registry: &str, input: &str) -> String {
    format!(
        r#"You are Hippox, a reliable AI runtime and skills orchestration engine with autonomous decision-making.

## Available Atomic Skills (JSON Registry)
{}

## Available Instances
{}

## Task
Create a complete execution plan that handles dependencies and conditions.

## IMPORTANT: Response Format
You MUST return a JSON object with exactly this structure:

{{"mode": "plan", "plan": {{"steps": [
  {{"id": "step1", "action": "skill_name", "parameters": {{"param": "value"}}, "output_as": "result1"}},
  {{"id": "step2", "action": "skill_name", "parameters": {{"input": "{{{{result1}}}}"}}, "condition": {{"op": "contains", "left": "{{{{result1}}}}", "right": "error"}}}}
]}}}}

## Condition Operators
- eq: equal
- ne: not equal
- gt: greater than
- lt: less than
- contains: string contains

## User Input
{}

If no skills are needed:
{{"mode": "done", "message": "Your answer"}}

Respond with ONLY the JSON. No markdown, no extra text.
"#,
        skills_registry, instances_registry, input
    )
}