mermaid-cli 0.6.0

Open-source AI pair programmer with agentic capabilities. Local-first with Ollama, native tool calling, and beautiful TUI.
Documentation
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
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
use anyhow::Result;
use serde::{Deserialize, Serialize};
use std::sync::Arc;
use tokio::sync::RwLock;

use crate::utils::MutexExt;

use crate::{
    agents::{ActionResult as AgentActionResult, AgentAction},
    app::Config,
    cli::OutputFormat,
    models::{
        ChatMessage, Model, ModelConfig, ModelFactory, StreamCallback, StreamEvent, ToolCall,
    },
    prompts,
};

use super::agent_loop::{self, AgentObserver, LoopControl, MAX_AGENT_ITERATIONS};

/// Result of a non-interactive run
#[derive(Debug, Serialize, Deserialize)]
pub struct NonInteractiveResult {
    /// The prompt that was executed
    pub prompt: String,
    /// The model's response
    pub response: String,
    /// Actions that were executed (if any)
    pub actions: Vec<ActionResult>,
    /// Any errors that occurred
    pub errors: Vec<String>,
    /// Metadata about the execution
    pub metadata: ExecutionMetadata,
}

#[derive(Debug, Serialize, Deserialize)]
pub struct ActionResult {
    /// Type of action (file_write, command, etc.)
    pub action_type: String,
    /// Target (file path or command)
    pub target: String,
    /// Whether the action was executed successfully
    pub success: bool,
    /// Output or error message
    pub output: Option<String>,
}

#[derive(Debug, Serialize, Deserialize)]
pub struct ExecutionMetadata {
    /// Model used
    pub model: String,
    /// Total tokens used
    pub tokens_used: Option<usize>,
    /// Execution time in milliseconds
    pub duration_ms: u128,
    /// Whether actions were executed
    pub actions_executed: bool,
}

/// Non-interactive runner for executing single prompts
pub struct NonInteractiveRunner {
    model: Arc<RwLock<Box<dyn Model>>>,
    no_execute: bool,
    model_config: ModelConfig,
    /// Project instructions auto-loaded from MERMAID.md (Step 5h).
    /// Non-interactive mode is one-shot — load once at construction,
    /// no auto-reload (single execute() call, no per-turn loop).
    instructions: Option<crate::app::instructions::LoadedInstructions>,
}

impl NonInteractiveRunner {
    /// Create a new non-interactive runner
    pub async fn new(
        model_id: String,
        config: Config,
        no_execute: bool,
        max_tokens: Option<usize>,
        reasoning: Option<crate::models::ReasoningLevel>,
    ) -> Result<Self> {
        // Create model instance
        let model = ModelFactory::create(&model_id, Some(&config)).await?;

        // Build base config from app config, then apply CLI overrides
        let mut model_config = ModelConfig::from_app_config(&config, &model_id);
        if let Some(mt) = max_tokens {
            model_config.max_tokens = mt;
        }
        // CLI `--reasoning` wins over the config-file default. Without
        // it, `from_app_config` already populated `reasoning` from
        // `[default_model].reasoning` (Wave 2). Non-interactive mode
        // historically forced thinking off; users now pick that
        // explicitly via `--reasoning none` if that's what they want.
        if let Some(level) = reasoning {
            model_config.reasoning = level;
        }

        // Step 5h: discover MERMAID.md once at construction. The
        // run-once nature of this path means there's no per-turn
        // refresh — you get whatever's on disk when you start.
        let cwd = std::env::current_dir().unwrap_or_else(|_| std::path::PathBuf::from("."));
        let instructions = crate::app::instructions::find_mermaid_md(&cwd)
            .and_then(|p| crate::app::instructions::load_from_path(&p));

        Ok(Self {
            model: Arc::new(RwLock::new(model)),
            no_execute,
            model_config,
            instructions,
        })
    }

    /// Execute a single prompt and return the result
    pub async fn execute(&self, prompt: String) -> Result<NonInteractiveResult> {
        let start_time = std::time::Instant::now();
        let mut errors = Vec::new();
        let mut total_tokens = 0;

        // Build initial messages
        let system_message = ChatMessage::system(prompts::get_system_prompt());
        let user_message = ChatMessage::user(prompt.clone());
        let mut messages = vec![system_message, user_message];

        // Use pre-built model config; inject the MERMAID.md suffix
        // (Step 5h) loaded at construction. One-shot mode means no
        // mid-run reload — suffix is whatever was on disk at startup.
        let mut effective_config = self.model_config.clone();
        effective_config.dynamic_system_suffix =
            self.instructions.as_ref().map(|i| i.content.clone());
        let model_config = &effective_config;
        let model_name = model_config.model.clone();

        // First model call. Accumulate text + tool calls from typed events;
        // ignore reasoning chunks (`ModelResponse.thinking` is still
        // populated and surfaced via the result if needed).
        let response_text = Arc::new(std::sync::Mutex::new(String::new()));
        let typed_tool_calls = Arc::new(std::sync::Mutex::new(Vec::<ToolCall>::new()));
        let text_clone = Arc::clone(&response_text);
        let tool_clone = Arc::clone(&typed_tool_calls);
        let callback: StreamCallback = Arc::new(move |event| match event {
            StreamEvent::Text(chunk) => {
                text_clone.lock_mut_safe().push_str(&chunk);
            },
            StreamEvent::ToolCall(tc) => {
                tool_clone.lock_mut_safe().push(tc);
            },
            StreamEvent::Reasoning(_) | StreamEvent::Done { .. } => {},
        });

        let result = {
            let model = self.model.read().await;
            model.chat(&messages, model_config, Some(callback)).await
        };

        let (content, initial_tool_calls) = match result {
            Ok(response) => {
                let streamed_text = response_text.lock_mut_safe().clone();
                let content = if !streamed_text.is_empty() {
                    streamed_text
                } else {
                    response.content
                };
                total_tokens += response.usage.map(|u| u.total_tokens).unwrap_or(0);
                let streamed_tool_calls = std::mem::take(&mut *typed_tool_calls.lock_mut_safe());
                let tool_calls = if !streamed_tool_calls.is_empty() {
                    streamed_tool_calls
                } else {
                    response.tool_calls.unwrap_or_default()
                };
                (content, tool_calls)
            },
            Err(e) => {
                errors.push(format!("Model error: {}", e));
                let content = response_text.lock_mut_safe().clone();
                (content, vec![])
            },
        };

        // If no tool calls, return immediately
        if initial_tool_calls.is_empty() {
            let duration_ms = start_time.elapsed().as_millis();
            return Ok(NonInteractiveResult {
                prompt,
                response: content,
                actions: vec![],
                errors,
                metadata: ExecutionMetadata {
                    model: model_name,
                    tokens_used: Some(total_tokens),
                    duration_ms,
                    actions_executed: false,
                },
            });
        }

        // Add assistant message with tool calls to history
        let assistant_msg =
            ChatMessage::assistant(content.clone()).with_tool_calls(initial_tool_calls.clone());
        messages.push(assistant_msg);

        // Handle --no-execute mode: record tool calls but don't execute them
        if self.no_execute {
            let actions = build_no_execute_actions(&initial_tool_calls, &mut messages);
            let duration_ms = start_time.elapsed().as_millis();
            return Ok(NonInteractiveResult {
                prompt,
                response: content,
                actions,
                errors,
                metadata: ExecutionMetadata {
                    model: model_name,
                    tokens_used: Some(total_tokens),
                    duration_ms,
                    actions_executed: false,
                },
            });
        }

        // Delegate to shared agent loop for tool execution + model re-calling
        let mut observer = SilentObserver;
        let loop_result = agent_loop::run_agent_loop(
            Arc::clone(&self.model),
            model_config,
            &mut messages,
            initial_tool_calls,
            &mut observer,
            MAX_AGENT_ITERATIONS,
        )
        .await?;

        // Build result from the agent loop
        total_tokens += loop_result.total_tokens;
        let final_response = if loop_result.final_response.is_empty() {
            content
        } else {
            loop_result.final_response
        };

        let actions: Vec<ActionResult> = loop_result
            .tool_results
            .iter()
            .map(|tr| {
                let (action_type, target) = extract_action_info(&tr.action);
                ActionResult {
                    action_type,
                    target,
                    success: tr.success,
                    output: Some(tr.output.clone()),
                }
            })
            .collect();

        if loop_result.interrupted {
            errors.push("Agent loop was interrupted".to_string());
        }

        let duration_ms = start_time.elapsed().as_millis();
        let actions_executed = !actions.is_empty();
        Ok(NonInteractiveResult {
            prompt,
            response: final_response,
            actions,
            errors,
            metadata: ExecutionMetadata {
                model: model_name,
                tokens_used: Some(total_tokens),
                duration_ms,
                actions_executed,
            },
        })
    }
}

/// Format a non-interactive result according to the output format
pub fn format_result(result: &NonInteractiveResult, format: OutputFormat) -> String {
    match format {
        OutputFormat::Json => serde_json::to_string_pretty(result)
            .unwrap_or_else(|e| format!("{{\"error\": \"Failed to serialize result: {}\"}}", e)),
        OutputFormat::Text => {
            let mut output = String::new();
            output.push_str(&result.response);

            if !result.actions.is_empty() {
                output.push_str("\n\n--- Actions ---\n");
                for action in &result.actions {
                    output.push_str(&format!(
                        "[{}] {} - {}\n",
                        if action.success { "OK" } else { "FAIL" },
                        action.action_type,
                        action.target
                    ));
                    if let Some(ref out) = action.output {
                        output.push_str(&format!("  {}\n", out));
                    }
                }
            }

            if !result.errors.is_empty() {
                output.push_str("\n--- Errors ---\n");
                for error in &result.errors {
                    output.push_str(&format!("{}\n", error));
                }
            }

            output
        },
        OutputFormat::Markdown => {
            let mut output = String::new();

            output.push_str("## Response\n\n");
            output.push_str(&result.response);
            output.push_str("\n\n");

            if !result.actions.is_empty() {
                output.push_str("## Actions Executed\n\n");
                for action in &result.actions {
                    let status = if action.success { "SUCCESS" } else { "FAILED" };
                    output.push_str(&format!(
                        "- {} **{}**: `{}`\n",
                        status, action.action_type, action.target
                    ));
                    if let Some(ref out) = action.output {
                        output.push_str(&format!("  ```\n  {}\n  ```\n", out));
                    }
                }
                output.push('\n');
            }

            if !result.errors.is_empty() {
                output.push_str("## Errors\n\n");
                for error in &result.errors {
                    output.push_str(&format!("- {}\n", error));
                }
                output.push('\n');
            }

            output.push_str("---\n");
            output.push_str(&format!(
                "*Model: {} | Tokens: {} | Duration: {}ms*\n",
                result.metadata.model,
                result.metadata.tokens_used.unwrap_or(0),
                result.metadata.duration_ms
            ));

            output
        },
    }
}

/// Extract action type and target description from an AgentAction
fn extract_action_info(action: &AgentAction) -> (String, String) {
    let (label, target) = action.display_info();
    (label.to_lowercase().replace(' ', "_"), target)
}

/// Build ActionResult entries for --no-execute mode (records tool calls without executing)
fn build_no_execute_actions(
    tool_calls: &[crate::models::ToolCall],
    messages: &mut Vec<ChatMessage>,
) -> Vec<ActionResult> {
    let mut actions = Vec::new();
    for tc in tool_calls {
        let tool_call_id = tc.id.clone().unwrap_or_else(|| "call_noexec".to_string());
        let tool_name = tc.function.name.clone();

        let (action_type, target) = match tc.to_agent_action() {
            Ok(action) => extract_action_info(&action),
            Err(_) => (tool_name.clone(), String::new()),
        };

        let msg = "Not executed (--no-execute mode)".to_string();
        messages.push(ChatMessage::tool(&tool_call_id, &tool_name, &msg));
        actions.push(ActionResult {
            action_type,
            target,
            success: false,
            output: Some(msg),
        });
    }
    actions
}

/// Observer that does nothing -- used by non-interactive mode
struct SilentObserver;

impl AgentObserver for SilentObserver {
    fn check_interrupt(&mut self) -> LoopControl {
        LoopControl::Continue
    }
    fn on_status(&mut self, _: &str) {}
    fn on_tool_result(&mut self, _: &str, _: &str, _: &AgentAction, _: &AgentActionResult) {}
    fn on_error(&mut self, _: &str) {}
    fn on_generation_start(&mut self) {}
    fn on_generation_complete(&mut self, _: usize) {}
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::agents::AgentAction;

    fn sample_result() -> NonInteractiveResult {
        NonInteractiveResult {
            prompt: "Fix the bug".to_string(),
            response: "I fixed the bug.".to_string(),
            actions: vec![ActionResult {
                action_type: "write_file".to_string(),
                target: "src/main.rs".to_string(),
                success: true,
                output: Some("File written".to_string()),
            }],
            errors: vec![],
            metadata: ExecutionMetadata {
                model: "test-model".to_string(),
                tokens_used: Some(100),
                duration_ms: 1234,
                actions_executed: true,
            },
        }
    }

    fn sample_result_with_errors() -> NonInteractiveResult {
        NonInteractiveResult {
            prompt: "Do something".to_string(),
            response: "Tried but failed.".to_string(),
            actions: vec![ActionResult {
                action_type: "bash".to_string(),
                target: "cargo test".to_string(),
                success: false,
                output: Some("tests failed".to_string()),
            }],
            errors: vec!["Command failed".to_string()],
            metadata: ExecutionMetadata {
                model: "test-model".to_string(),
                tokens_used: Some(50),
                duration_ms: 500,
                actions_executed: true,
            },
        }
    }

    #[test]
    fn test_extract_action_info_read() {
        let action = AgentAction::ReadFile {
            paths: vec!["foo.rs".to_string()],
        };
        let (action_type, target) = extract_action_info(&action);
        assert_eq!(action_type, "read");
        assert_eq!(target, "foo.rs");
    }

    #[test]
    fn test_extract_action_info_bash() {
        let action = AgentAction::ExecuteCommand {
            command: "cargo test".to_string(),
            working_dir: None,
            timeout: None,
        };
        let (action_type, target) = extract_action_info(&action);
        assert_eq!(action_type, "bash");
        assert_eq!(target, "cargo test");
    }

    #[test]
    fn test_extract_action_info_web_search() {
        let action = AgentAction::WebSearch {
            queries: vec![("rust async".to_string(), 5)],
        };
        let (action_type, target) = extract_action_info(&action);
        assert_eq!(action_type, "web_search");
        assert_eq!(target, "rust async");
    }

    #[test]
    fn test_extract_action_info_write() {
        let action = AgentAction::WriteFile {
            path: "out.txt".to_string(),
            content: "hello".to_string(),
        };
        let (action_type, target) = extract_action_info(&action);
        assert_eq!(action_type, "write");
        assert_eq!(target, "out.txt");
    }

    #[test]
    fn test_format_result_json() {
        let result = sample_result();
        let json = format_result(&result, OutputFormat::Json);
        assert!(json.contains("\"prompt\": \"Fix the bug\""));
        assert!(json.contains("\"success\": true"));
        assert!(json.contains("\"model\": \"test-model\""));
    }

    #[test]
    fn test_format_result_text() {
        let result = sample_result();
        let text = format_result(&result, OutputFormat::Text);
        assert!(text.contains("I fixed the bug."));
        assert!(text.contains("[OK] write_file - src/main.rs"));
        assert!(text.contains("--- Actions ---"));
    }

    #[test]
    fn test_format_result_text_with_errors() {
        let result = sample_result_with_errors();
        let text = format_result(&result, OutputFormat::Text);
        assert!(text.contains("[FAIL] bash - cargo test"));
        assert!(text.contains("--- Errors ---"));
        assert!(text.contains("Command failed"));
    }

    #[test]
    fn test_format_result_markdown() {
        let result = sample_result();
        let md = format_result(&result, OutputFormat::Markdown);
        assert!(md.contains("## Response"));
        assert!(md.contains("I fixed the bug."));
        assert!(md.contains("## Actions Executed"));
        assert!(md.contains("SUCCESS **write_file**"));
        assert!(md.contains("*Model: test-model"));
    }

    #[test]
    fn test_format_result_text_no_actions() {
        let result = NonInteractiveResult {
            prompt: "hi".to_string(),
            response: "hello".to_string(),
            actions: vec![],
            errors: vec![],
            metadata: ExecutionMetadata {
                model: "m".to_string(),
                tokens_used: None,
                duration_ms: 10,
                actions_executed: false,
            },
        };
        let text = format_result(&result, OutputFormat::Text);
        assert_eq!(text, "hello");
        assert!(!text.contains("Actions"));
    }
}