gac 1.9.5__tar.gz → 1.10.1__tar.gz

{gac-1.9.5 → gac-1.10.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: gac
-Version: 1.9.5
+Version: 1.10.1
 Summary: LLM-powered Git commit message generator with multi-provider support
 Project-URL: Homepage, https://github.com/cellwebb/gac
 Project-URL: Documentation, https://github.com/cellwebb/gac#readme
@@ -33,7 +33,6 @@ Requires-Dist: tiktoken>=0.12.0
 Provides-Extra: dev
 Requires-Dist: build; extra == 'dev'
 Requires-Dist: codecov; extra == 'dev'
-Requires-Dist: pre-commit; extra == 'dev'
 Requires-Dist: pytest; extra == 'dev'
 Requires-Dist: pytest-cov; extra == 'dev'
 Requires-Dist: ruff; extra == 'dev'
@@ -69,58 +68,52 @@ Instead of generic messages like `"update stuff"`, `"fix bug"`, or `"add feature
 
 ## Quick Start
 
-### Install & Use
+### Use without installing
 
 ```bash
-# 1. Install globally
-uv tool install gac
-
-# 2. Configure your LLM provider
-gac init
-
-# 3. Use on your staged changes
-git add .
-gac
+uvx gac init  # Configure your LLM provider
+uvx gac  # Generate and commit with LLM
 ```
 
 That's it! Review the generated message and confirm with `y`.
 
-### Use without installing
+### Install gac globally
 
 ```bash
-uvx gac init
-uvx gac
+uv tool install gac
+gac init
+gac
 ```
 
 ---
 
 ## Key Features
 
-### **Supported Providers**
+### 🌐 **Supported Providers**
 
-- **Anthropic** • **Cerebras** • **Chutes.ai** • **Gemini** • **Groq**
-- **LM Studio** • **Ollama** • **OpenAI** • **OpenRouter**
+- **Anthropic** • **Cerebras** • **Chutes.ai** • **Fireworks** • **Gemini**
+- **Groq** • **LM Studio** • **Ollama** • **OpenAI** • **OpenRouter**
 - **Streamlake** • **Synthetic.new** • **Z.AI** • **Z.AI Coding**
 
-### **Smart LLM Analysis**
+### 🧠 **Smart LLM Analysis**
 
 - **Understands intent**: Analyzes code structure, logic, and patterns to understand the "why" behind your changes, not just what changed
 - **Semantic awareness**: Recognizes refactoring, bug fixes, features, and breaking changes to generate contextually appropriate messages
 - **Intelligent filtering**: Prioritizes meaningful changes while ignoring generated files, dependencies, and artifacts
 
-### **Multiple Message Formats**
+### 📝 **Multiple Message Formats**
 
 - **One-liner** (-o flag): Single-line commit message following conventional commit format
 - **Standard** (default): Summary with bullet points explaining implementation details
 - **Verbose** (-v flag): Comprehensive explanations including motivation, technical approach, and impact analysis
 
-### **Developer Experience**
+### 💻 **Developer Experience**
 
 - **Interactive feedback**: Regenerate messages with specific requests like `r "make it shorter"` or `r "focus on the bug fix"`
 - **One-command workflows**: Complete workflows with flags like `gac -ayp` (stage all, auto-confirm, push)
 - **Git integration**: Respects pre-commit and lefthook hooks, running them before expensive LLM operations
 
-### **Built-in Security**
+### 🛡️ **Built-in Security**
 
 - **Automatic secret detection**: Scans for API keys, passwords, and tokens before committing
 - **Interactive protection**: Prompts before committing potentially sensitive data with clear remediation options

{gac-1.9.5 → gac-1.10.1}/README.md

@@ -27,58 +27,52 @@ Instead of generic messages like `"update stuff"`, `"fix bug"`, or `"add feature
 
 ## Quick Start
 
-### Install & Use
+### Use without installing
 
 ```bash
-# 1. Install globally
-uv tool install gac
-
-# 2. Configure your LLM provider
-gac init
-
-# 3. Use on your staged changes
-git add .
-gac
+uvx gac init  # Configure your LLM provider
+uvx gac  # Generate and commit with LLM
 ```
 
 That's it! Review the generated message and confirm with `y`.
 
-### Use without installing
+### Install gac globally
 
 ```bash
-uvx gac init
-uvx gac
+uv tool install gac
+gac init
+gac
 ```
 
 ---
 
 ## Key Features
 
-### **Supported Providers**
+### 🌐 **Supported Providers**
 
-- **Anthropic** • **Cerebras** • **Chutes.ai** • **Gemini** • **Groq**
-- **LM Studio** • **Ollama** • **OpenAI** • **OpenRouter**
+- **Anthropic** • **Cerebras** • **Chutes.ai** • **Fireworks** • **Gemini**
+- **Groq** • **LM Studio** • **Ollama** • **OpenAI** • **OpenRouter**
 - **Streamlake** • **Synthetic.new** • **Z.AI** • **Z.AI Coding**
 
-### **Smart LLM Analysis**
+### 🧠 **Smart LLM Analysis**
 
 - **Understands intent**: Analyzes code structure, logic, and patterns to understand the "why" behind your changes, not just what changed
 - **Semantic awareness**: Recognizes refactoring, bug fixes, features, and breaking changes to generate contextually appropriate messages
 - **Intelligent filtering**: Prioritizes meaningful changes while ignoring generated files, dependencies, and artifacts
 
-### **Multiple Message Formats**
+### 📝 **Multiple Message Formats**
 
 - **One-liner** (-o flag): Single-line commit message following conventional commit format
 - **Standard** (default): Summary with bullet points explaining implementation details
 - **Verbose** (-v flag): Comprehensive explanations including motivation, technical approach, and impact analysis
 
-### **Developer Experience**
+### 💻 **Developer Experience**
 
 - **Interactive feedback**: Regenerate messages with specific requests like `r "make it shorter"` or `r "focus on the bug fix"`
 - **One-command workflows**: Complete workflows with flags like `gac -ayp` (stage all, auto-confirm, push)
 - **Git integration**: Respects pre-commit and lefthook hooks, running them before expensive LLM operations
 
-### **Built-in Security**
+### 🛡️ **Built-in Security**
 
 - **Automatic secret detection**: Scans for API keys, passwords, and tokens before committing
 - **Interactive protection**: Prompts before committing potentially sensitive data with clear remediation options

{gac-1.9.5 → gac-1.10.1}/pyproject.toml

@@ -61,7 +61,6 @@ dev = [
     
     # Linting and formatting
     "ruff",
-    "pre-commit",
     
     # Release tools
     "build",
@@ -150,7 +149,6 @@ dependencies = [
     
     # Linting and formatting
     "ruff",
-    "pre-commit",
 ]
 
 [tool.hatch.envs.default.scripts]
@@ -232,6 +230,5 @@ dev = [
     "pytest>=8.4.2",
     "pytest-asyncio>=1.2.0",
     "pytest-cov>=7.0.0",
-    "pre-commit",
     "mypy>=1.8.0",
 ]

{gac-1.9.5 → gac-1.10.1}/src/gac/__version__.py

@@ -1,3 +1,3 @@
 """Version information for gac package."""
 
-__version__ = "1.9.5"
+__version__ = "1.10.1"

{gac-1.9.5 → gac-1.10.1}/src/gac/ai.py

@@ -13,6 +13,7 @@ from gac.providers import (
     call_anthropic_api,
     call_cerebras_api,
     call_chutes_api,
+    call_fireworks_api,
     call_gemini_api,
     call_groq_api,
     call_lmstudio_api,
@@ -30,7 +31,7 @@ logger = logging.getLogger(__name__)
 
 def generate_commit_message(
     model: str,
-    prompt: str | tuple[str, str],
+    prompt: str | tuple[str, str] | list[dict[str, str]],
     temperature: float = EnvDefaults.TEMPERATURE,
     max_tokens: int = EnvDefaults.MAX_OUTPUT_TOKENS,
     max_retries: int = EnvDefaults.MAX_RETRIES,
@@ -59,18 +60,28 @@ def generate_commit_message(
         'docs: Update README with installation instructions'
     """
     # Handle both old (string) and new (tuple) prompt formats
-    if isinstance(prompt, tuple):
+    if isinstance(prompt, list):
+        messages = [{**msg} for msg in prompt]
+    elif isinstance(prompt, tuple):
         system_prompt, user_prompt = prompt
+        messages = [
+            {"role": "system", "content": system_prompt or ""},
+            {"role": "user", "content": user_prompt},
+        ]
     else:
-        # Backward compatibility: treat string as user prompt with no system prompt
-        system_prompt = ""
-        user_prompt = prompt
+        # Backward compatibility: treat string as user prompt with empty system prompt
+        user_prompt = str(prompt)
+        messages = [
+            {"role": "system", "content": ""},
+            {"role": "user", "content": user_prompt},
+        ]
 
     # Provider functions mapping
     provider_funcs = {
         "anthropic": call_anthropic_api,
         "cerebras": call_cerebras_api,
         "chutes": call_chutes_api,
+        "fireworks": call_fireworks_api,
         "gemini": call_gemini_api,
         "groq": call_groq_api,
         "lm-studio": call_lmstudio_api,
@@ -88,8 +99,7 @@ def generate_commit_message(
         return generate_with_retries(
             provider_funcs=provider_funcs,
             model=model,
-            system_prompt=system_prompt,
-            user_prompt=user_prompt,
+            messages=messages,
             temperature=temperature,
             max_tokens=max_tokens,
             max_retries=max_retries,

{gac-1.9.5 → gac-1.10.1}/src/gac/ai_utils.py

@@ -78,8 +78,7 @@ def _classify_error(error_str: str) -> str:
 def generate_with_retries(
     provider_funcs: dict,
     model: str,
-    system_prompt: str,
-    user_prompt: str,
+    messages: list[dict[str, str]],
     temperature: float,
     max_tokens: int,
     max_retries: int,
@@ -97,6 +96,7 @@ def generate_with_retries(
         "anthropic",
         "cerebras",
         "chutes",
+        "fireworks",
         "gemini",
         "groq",
         "lm-studio",
@@ -111,10 +111,8 @@ def generate_with_retries(
     if provider not in supported_providers:
         raise AIError.model_error(f"Unsupported provider: {provider}. Supported providers: {supported_providers}")
 
-    messages = [
-        {"role": "system", "content": system_prompt},
-        {"role": "user", "content": user_prompt},
-    ]
+    if not messages:
+        raise AIError.model_error("No messages provided for AI generation")
 
     # Set up spinner
     if quiet:

{gac-1.9.5 → gac-1.10.1}/src/gac/init_cli.py

@@ -35,6 +35,7 @@ def init() -> None:
         ("Anthropic", "claude-haiku-4-5"),
         ("Cerebras", "qwen-3-coder-480b"),
         ("Chutes", "zai-org/GLM-4.6-FP8"),
+        ("Fireworks", "accounts/fireworks/models/gpt-oss-20b"),
         ("Gemini", "gemini-2.5-flash"),
         ("Groq", "meta-llama/llama-4-maverick-17b-128e-instruct"),
         ("LM Studio", "gemma3"),

{gac-1.9.5 → gac-1.10.1}/src/gac/main.py

@@ -202,39 +202,49 @@ def main(
             )
         )
 
+    conversation_messages: list[dict[str, str]] = []
+    if system_prompt:
+        conversation_messages.append({"role": "system", "content": system_prompt})
+    conversation_messages.append({"role": "user", "content": user_prompt})
+
     try:
-        # Count tokens for both prompts
-        prompt_tokens = count_tokens(system_prompt, model) + count_tokens(user_prompt, model)
-
-        warning_limit_val = config.get("warning_limit_tokens", EnvDefaults.WARNING_LIMIT_TOKENS)
-        assert warning_limit_val is not None
-        warning_limit = int(warning_limit_val)
-        if warning_limit and prompt_tokens > warning_limit:
-            console.print(
-                f"[yellow]⚠️  WARNING: Prompt contains {prompt_tokens} tokens, which exceeds the warning limit of "
-                f"{warning_limit} tokens.[/yellow]"
+        first_iteration = True
+
+        while True:
+            prompt_tokens = count_tokens(conversation_messages, model)
+
+            if first_iteration:
+                warning_limit_val = config.get("warning_limit_tokens", EnvDefaults.WARNING_LIMIT_TOKENS)
+                assert warning_limit_val is not None
+                warning_limit = int(warning_limit_val)
+                if warning_limit and prompt_tokens > warning_limit:
+                    console.print(
+                        f"[yellow]⚠️  WARNING: Prompt contains {prompt_tokens} tokens, which exceeds the warning limit of "
+                        f"{warning_limit} tokens.[/yellow]"
+                    )
+                    if require_confirmation:
+                        proceed = click.confirm("Do you want to continue anyway?", default=True)
+                        if not proceed:
+                            console.print("[yellow]Aborted due to token limit.[/yellow]")
+                            sys.exit(0)
+
+            first_iteration = False
+
+            raw_commit_message = generate_commit_message(
+                model=model,
+                prompt=conversation_messages,
+                temperature=temperature,
+                max_tokens=max_output_tokens,
+                max_retries=max_retries,
+                quiet=quiet,
             )
-            if require_confirmation:
-                proceed = click.confirm("Do you want to continue anyway?", default=True)
-                if not proceed:
-                    console.print("[yellow]Aborted due to token limit.[/yellow]")
-                    sys.exit(0)
+            commit_message = clean_commit_message(raw_commit_message)
 
-        commit_message = generate_commit_message(
-            model=model,
-            prompt=(system_prompt, user_prompt),
-            temperature=temperature,
-            max_tokens=max_output_tokens,
-            max_retries=max_retries,
-            quiet=quiet,
-        )
-        commit_message = clean_commit_message(commit_message)
+            logger.info("Generated commit message:")
+            logger.info(commit_message)
 
-        logger.info("Generated commit message:")
-        logger.info(commit_message)
+            conversation_messages.append({"role": "assistant", "content": commit_message})
 
-        # Reroll loop
-        while True:
             console.print("[bold green]Generated commit message:[/bold green]")
             console.print(Panel(commit_message, title="Commit Message", border_style="cyan"))
 
@@ -252,76 +262,38 @@ def main(
                     response = click.prompt(
                         "Proceed with commit above? [y/n/r <feedback>]", type=str, show_default=False
                     ).strip()
+                    response_lower = response.lower()
 
-                    if response.lower() in ["y", "yes"]:
-                        break  # Exit both loops and proceed with commit
-                    elif response.lower() in ["n", "no"]:
+                    if response_lower in ["y", "yes"]:
+                        break
+                    if response_lower in ["n", "no"]:
                         console.print("[yellow]Prompt not accepted. Exiting...[/yellow]")
                         sys.exit(0)
-                    elif response.lower() == "r" or response.lower().startswith("r ") or response.lower() == "reroll":
-                        # Parse the reroll command for optional feedback
-                        if response.lower() == "r" or response.lower() == "reroll":
-                            # Simple reroll without feedback
-                            reroll_feedback = ""
+                    if response_lower == "r" or response_lower == "reroll" or response_lower.startswith("r "):
+                        if response_lower == "r" or response_lower == "reroll":
+                            feedback_message = (
+                                "Please provide an alternative commit message using the same repository context."
+                            )
                             console.print("[cyan]Regenerating commit message...[/cyan]")
                         else:
-                            # Extract feedback from "r <feedback>"
-                            reroll_feedback = response[2:].strip()  # Remove "r " prefix
-                            console.print(f"[cyan]Regenerating commit message with feedback: {reroll_feedback}[/cyan]")
-
-                        # Combine hints if reroll feedback provided
-                        combined_hint = hint
-                        if reroll_feedback:
-                            # Create conversational prompt with previous attempt and feedback
-                            conversational_hint = f"Previous attempt: '{commit_message}'. User feedback: {reroll_feedback}. Please revise accordingly."
-
-                            if hint:
-                                combined_hint = f"{hint}. {conversational_hint}"
-                            else:
-                                combined_hint = conversational_hint
-
-                            # Regenerate prompt with conversational feedback
-                            reroll_system_prompt, reroll_user_prompt = build_prompt(
-                                status=status,
-                                processed_diff=processed_diff,
-                                diff_stat=diff_stat,
-                                one_liner=one_liner,
-                                hint=combined_hint,
-                                infer_scope=infer_scope,
-                                verbose=verbose,
+                            reroll_feedback = response[2:].strip()
+                            feedback_message = (
+                                f"Please revise the commit message based on this feedback: {reroll_feedback}"
                             )
-                        else:
-                            # No hint given, just reroll with same prompts
-                            reroll_system_prompt, reroll_user_prompt = system_prompt, user_prompt
+                            console.print(f"[cyan]Regenerating commit message with feedback: {reroll_feedback}[/cyan]")
 
-                        # Update prompts and recalculate token count for the next generation
-                        system_prompt = reroll_system_prompt
-                        user_prompt = reroll_user_prompt
-                        prompt_tokens = count_tokens(system_prompt, model) + count_tokens(user_prompt, model)
+                        conversation_messages.append({"role": "user", "content": feedback_message})
 
                         console.print()  # Add blank line for readability
+                        break
+
+                    console.print(
+                        "[red]Invalid response. Please enter y (yes), n (no), r (reroll), or r <feedback>.[/red]"
+                    )
 
-                        # Generate new message
-                        commit_message = generate_commit_message(
-                            model=model,
-                            prompt=(system_prompt, user_prompt),
-                            temperature=temperature,
-                            max_tokens=max_output_tokens,
-                            max_retries=max_retries,
-                            quiet=quiet,
-                        )
-                        commit_message = clean_commit_message(commit_message)
-                        break  # Exit inner loop, continue outer loop
-                    else:
-                        console.print(
-                            "[red]Invalid response. Please enter y (yes), n (no), r (reroll), or r <feedback>.[/red]"
-                        )
-
-                # If we got here with 'y', break the outer loop
-                if response.lower() in ["y", "yes"]:
+                if response_lower in ["y", "yes"]:
                     break
             else:
-                # No confirmation required, exit loop
                 break
 
         if dry_run:

{gac-1.9.5 → gac-1.10.1}/src/gac/providers/__init__.py

@@ -3,6 +3,7 @@
 from .anthropic import call_anthropic_api
 from .cerebras import call_cerebras_api
 from .chutes import call_chutes_api
+from .fireworks import call_fireworks_api
 from .gemini import call_gemini_api
 from .groq import call_groq_api
 from .lmstudio import call_lmstudio_api
@@ -17,6 +18,7 @@ __all__ = [
     "call_anthropic_api",
     "call_cerebras_api",
     "call_chutes_api",
+    "call_fireworks_api",
     "call_gemini_api",
     "call_groq_api",
     "call_lmstudio_api",

gac-1.10.1/src/gac/providers/fireworks.py

@@ -0,0 +1,38 @@
+"""Fireworks AI API provider for gac."""
+
+import os
+
+import httpx
+
+from gac.errors import AIError
+
+
+def call_fireworks_api(model: str, messages: list[dict], temperature: float, max_tokens: int) -> str:
+    """Call Fireworks AI API directly."""
+    api_key = os.getenv("FIREWORKS_API_KEY")
+    if not api_key:
+        raise AIError.authentication_error("FIREWORKS_API_KEY not found in environment variables")
+
+    url = "https://api.fireworks.ai/inference/v1/chat/completions"
+    headers = {"Authorization": f"Bearer {api_key}", "Content-Type": "application/json"}
+
+    data = {"model": model, "messages": messages, "temperature": temperature, "max_tokens": max_tokens}
+
+    try:
+        response = httpx.post(url, headers=headers, json=data, timeout=120)
+        response.raise_for_status()
+        response_data = response.json()
+        content = response_data["choices"][0]["message"]["content"]
+        if content is None:
+            raise AIError.model_error("Fireworks AI API returned null content")
+        if content == "":
+            raise AIError.model_error("Fireworks AI API returned empty content")
+        return content
+    except httpx.HTTPStatusError as e:
+        if e.response.status_code == 429:
+            raise AIError.rate_limit_error(f"Fireworks AI API rate limit exceeded: {e.response.text}") from e
+        raise AIError.model_error(f"Fireworks AI API error: {e.response.status_code} - {e.response.text}") from e
+    except httpx.TimeoutException as e:
+        raise AIError.timeout_error(f"Fireworks AI API request timed out: {str(e)}") from e
+    except Exception as e:
+        raise AIError.model_error(f"Error calling Fireworks AI API: {str(e)}") from e