LC

    local codegen workflow

    1

    Generate production code with a local LLM on the GPU you already own. Spec-driven codegen via LM Studio/Ollama/vLLM, with auto model-loading, a 13-bug review checklist, and a working genfile script — all model-agnostic.

    Free

    2 installsSecurity scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+22 more
    LC

    local codegen workflow

    Example session with this skill installed

    I need a Python module for an LRU cache. It should use the standard library only, support a max_size, and have get/put methods. Save it to lru_cache.py using the local Qwen model.

    • Read your context and instructions
    • Compiled the local codegen workflow

    I have generated the lru_cache.py module using the local qwen2.5-coder-14b-instruct model. The model was pre-loaded at a 32k context window to ensure VRAM stability.

    Generated LRU Cache Module

    from collections import OrderedDict

class LRUCache:
    """
    A simple LRU Cache using OrderedDict.
    """
    def __init__(self, max_size: int = 128):
        if max_size <= 0:
            raise ValueError("max_size must be positive")
        self.cache = OrderedDict()
        self.max_size = max_size

    def get(self, key):
        if key not in self.cache:
            return None
        self.cache.move_to_end(key)
        return self.cache[key]

    def put(self, key, value):
        if key in self.cache:
            self.cache.move_to_end(key)
        self.cache[key] = value
        if len(self.cache) > self.max_size:
            self.cache.popitem(last=False)

if __name__ == "__main__":
    # Smoke test
    lru = LRUCache(2)
    lru.put(1, "a")
    lru.put(2, "b")
    print(lru.get(1))  # a
    lru.put(3, "c")    # evicts 2
    print(lru.get(2))  # None

    Next steps

    1. Run pytest on the generated module to verify eviction logic.
    2. Review the OrderedDict implementation against your performance requirements.
    3. Commit with the message feat: add lru_cache (AI-gen, reviewed).

    Connects securely to your tools. The creator never sees your data.

    About this skill

    The problem

    Paying per-token API costs for repetitive boilerplate and large-scale code generation is expensive. Local LLMs often fail during generation due to OOM errors because they are loaded with default context windows that exceed available VRAM.

    What it does

    • Automates the spec-driven generation of production code using local LLM servers.
    • Manages VRAM-safe model loading via LM Studio v1 REST API to prevent OOM "terminated" errors.
    • Enforces a senior-level review checklist to catch common local model fumbles like connection leaks and empty-list SQL crashes.
    • Implements a 2-strike rule to prevent time-wasting loops on low-quality generations.
    • Generates both the implementation module and its corresponding test suite from a single source-of-truth spec.

    Frameworks & tools

    Python 3.11, httpx, LM Studio, Ollama, vLLM, Qwen2.5-Coder, DeepSeek-Coder.

    Why this beats prompting it yourself

    Standard prompting often results in code blocks wrapped in conversational prose that break automation scripts. This skill includes a specialized extraction layer and a model-management handshake that ensures the model is actually loaded at a stable context length before the first token is even generated.

    Use cases

    • Generating large volumes of boilerplate for new microservices without incurring API fees.
    • Prototyping internal tools where data privacy prevents sending code to third-party APIs.
    • Setting up a repeatable generate-review-test-commit pipeline for junior-level tasks.

    Known limitations

    Not recommended for surgical edits to existing complex files. Requires a GPU with at least 8GB VRAM for 7B models or 16GB for 14B models.

    How to use

    Skills on Agensi run inside your AI agent. Copy the prompt, paste it into your agent, and it connects to Agensi and installs this skill for you. If a payment is needed, your agent shows it to you first.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill passes our trust checks below.

    Security scanned

    Passed our 8-point scan before listing

    2 installs

    Downloaded by developers to date

    Free forever

    No account required to browse

    Trust & safety

    Security scanned

    Verified clean 1 month ago

    • Free to download with an account

    Requires

    Python

    Listed1 month ago

    What's inside

    Frequently Asked Questions