Show HN: Forge – Guardrails take an 8B model from 53% to 99% on agentic tasks

A reliability layer for self-hosted LLM tool-calling. You give forge a set of tools; the model calls whichever it wants in whatever order. Workflow structure is opt-in — required_steps

, prerequisites

, and terminal_tool

let you constrain the loop when you need to, but forge's guardrails (rescue parsing, retry nudges, response validation) apply with zero required steps too.

Forge takes an 8B local model from single digits to 84% across forge's 26-scenario v0.7.0 eval suite — and even lifts Sonnet 4.6 from 85% to 98% on the same workload (Anthropic numbers measured in v0.6.0; not re-run in v0.7.0 since the cost is non-trivial).

What forge isn't:

Not an agent orchestrator. Forge sits inside one agentic loop and makes its tool calls reliable. Multi-agent graphs, DAG planners, and cross-agent coordination are out of scope.Not a coding harness. Forge is domain-agnostic. If you're building a coding agent (or already using one like opencode, aider, Cline),proxy modelifts your existing harness with forge's guardrails — no rewrite.

Three ways to use it:

Proxy server— Drop-in proxy (python -m forge.proxy

) speaking both the OpenAI chat-completions and Anthropic Messages (/v1/messages

) APIs, sitting between any client and a local model server. Point OpenAI-compatible tools (opencode, Continue, aider)or Claude Code at it and forge applies guardrails transparently — the client thinks it's talking to a smarter model. Most popular entry point. - WorkflowRunner— Define tools, pick a backend, run structured agent loops. Forge manages the full lifecycle: system prompts, tool execution, context compaction, and guardrails.SlotWorker adds priority-queued access to a shared inference slot with auto-preemption — for multi-agent architectures where specialist workflows share a GPU slot. Best when you're building on forge directly. - Guardrails middleware— Use forge's reliability stack (composable middleware) inside your own orchestration loop. You control the loop; forge validates responses, rescues malformed tool calls, and enforces required steps.

Supports Ollama, llama-server (llama.cpp), Llamafile, vLLM, and Anthropic as backends.

• Python 3.12+
• A running LLM backend (see below)

pip install forge-guardrails                # core only
pip install "forge-guardrails[anthropic]"   # + Anthropic client

For development:

git clone https://github.com/antoinezambelli/forge.git
cd forge
pip install -e ".[dev]"

llama-server (recommended — top 10 eval configs all run on llama-server):

llama-server -m path/to/Ministral-3-8B-Instruct-2512-Q8_0.gguf --jinja -ngl 999 --port 8080

Ollama (alternative — easier setup, slightly weaker on harder workloads):

ollama pull ministral-3:8b-instruct-2512-q4_K_M

Anthropic (API, no local GPU needed):

pip install -e ".[anthropic]"
export ANTHROPIC_API_KEY=sk-...

See Backend Setup for full instructions and Model Guide for which model fits your hardware.

Start llama-server however you normally do (e.g. in a separate shell):

llama-server -m path/to/Ministral-3-8B-Instruct-2512-Q8_0.gguf --jinja -ngl 999 --port 8080

Then the Python you'll run (e.g. from another shell):

import asyncio
from pydantic import BaseModel, Field
from forge import (
    Workflow, ToolDef, ToolSpec,
    WorkflowRunner, LlamafileClient,
    ContextManager, TieredCompact,
)

def get_weather(city: str) -> str:
    return f"72°F and sunny in {city}"

class GetWeatherParams(BaseModel):
    city: str = Field(description="City name")

workflow = Workflow(
    name="weather",
    description="Look up weather for a city.",
    tools={
        "get_weather": ToolDef(
            spec=ToolSpec(
                name="get_weather",
                description="Get current weather",
                parameters=GetWeatherParams,
            ),
            callable=get_weather,
        ),
    },
    required_steps=[],
    terminal_tool="get_weather",
    system_prompt_template="You are a helpful assistant. Use the available tools to answer the user.",
)

async def main():
    client = LlamafileClient(
        gguf_path="path/to/Ministral-3-8B-Instruct-2512-Q8_0.gguf",
        mode="native",
        recommended_sampling=True,
    )
    ctx = ContextManager(strategy=TieredCompact(keep_recent=2), budget_tokens=8192)
    runner = WorkflowRunner(client=client, context_manager=ctx)
    await runner.run(workflow, "What's the weather in Paris?")

asyncio.run(main())

For multi-step workflows, multi-turn conversations, and backend auto-management, see the User Guide. If you're building a long-running session (CLI, chat server, voice assistant), see the long-running session advisory for important guidance on filtering transient messages.

Drop-in proxy that sits between any client and a local model server, speaking both the OpenAI chat-completions API and the Anthropic Messages API (/v1/messages

). Point your client at the proxy (e.g. http://localhost:8081/v1

) and forge applies its guardrails transparently — the client thinks it's talking to a smarter model.

This is the path for using forge with an existing harness (opencode, Continue, aider, Cline, anything that speaks the OpenAI chat-completions schema — or Claude Code, which speaks the Anthropic Messages API). No Python rewrite.

python -m forge.proxy --backend-url http://localhost:8080 --port 8081

python -m forge.proxy --backend llamaserver --gguf path/to/model.gguf --port 8081

python -m forge.proxy --backend vllm --model-path /path/to/awq-dir --port 8081

Then configure your client to use http://localhost:8081/v1

as the API base URL.

Claude Code: the proxy also serves the Anthropic Messages API on POST /v1/messages

, so you can point Claude Code at a forge-guarded local model — set ANTHROPIC_BASE_URL=http://localhost:8081

and ANTHROPIC_AUTH_TOKEN=anything

for the claude

process. See Using forge with Claude Code for the full setup (native-vs-prompt FC, Anthropic-shape downstreams, cache_control

).

Backend compatibility:

Managed mode spins up the backend for you. Supported backends:llamaserver

,llamafile

,ollama

,vllm

(use--backend <name>

with--gguf

for the GGUF-based backends,--model-path

for vllm, or--model

for ollama).External mode is backend-agnostic — forge talksPOST /v1/chat/completions

to whatever you point--backend-url

at, as long as it speaks the OpenAI schema. Tool calls must come back in OpenAItool_calls

format or in one of forge's rescue-parsed formats (Mistral[TOOL_CALLS]

, Qwen<tool_call>

XML, fenced JSON). For a vLLM server, add--backend vllm

so the proxy adopts vLLM's--served-model-name

(vLLM 404s on a mismatchedmodel

field, unlike llama.cpp).

On every POST /v1/chat/completions

, forge applies (in order):

Response validation— each tool call in the model's response is checked against thetools

array in the request. Calls to unknown tool names or with malformed shapes are caught before the response returns to your client.Rescue parsing— when the model emits tool calls in the wrong format (JSON in a code fence, Mistral's[TOOL_CALLS]name{args}

, Qwen's<tool_call>...</tool_call>

XML), forge extracts the structured call and re-emits it in the canonical OpenAItool_calls

schema. Biggest practical lift for Mistral-family models.Retry loop with error tracking— if validation fails, forge retries inference up to--max-retries

(default 3) with a corrective tool-result message on the canonical channel, rather than returning a malformed response. From your client's perspective the proxy looks like a single request that just took a few extra ms.Synthetic— when tools are present in the request, forge injects a syntheticrespond

tool injectionrespond

tool the model calls instead of producing bare text. Therespond

call is stripped from the outbound response — the client sees a normal text response (finish_reason: "stop"

) and never knows the tool exists. Essential for small local models (~8B) that can't be trusted to choose correctly between text and tool calls. SeeADR-013for the full analysis.

Proxy mode is single-shot per request; some forge features need multi-turn workflow state that the OpenAI chat-completions schema doesn't carry:

Prerequisite enforcement and step-ordering— these need a workflow definition spanning turns. Available inWorkflowRunner

.Context compaction and session memory— proxy mode forwards the inbound message list as-is; managing the rolling window is the client's job.** VRAM-aware budget detection**— opt in with--budget-mode forge-full

or--budget-mode forge-fast

; otherwise proxy uses the backend's reported budget.

For the full guardrail surface, use WorkflowRunner

directly. The proxy trades depth for "use forge with your existing setup, no rewrite."

You can run the forge proxy as a Docker container.

Build the image:

docker build -t forge-proxy .

Run the container:

docker run -p 8081:8081 forge-proxy --backend-url http://host.docker.internal:8000 --backend vllm --budget-mode manual --budget-tokens 8192

Note: If your backend is running on localhost

of the host machine, use http://host.docker.internal:PORT

(on macOS/Windows) or the host's IP address to allow the container to reach it.

See Backend Setup for installation and Model Guide for which model to pick.

python -m pytest tests/ -v --tb=short
python -m pytest tests/ --cov=forge --cov-report=term-missing

26 scenarios measuring how reliably a model + backend combo navigates multi-step tool-calling workflows — split into an OG-18 baseline tier and an 8-scenario advanced_reasoning tier for top-end separation. See Eval Guide for full CLI reference.

python -m tests.eval.eval_runner --backend llamafile --llamafile-mode prompt --gguf "path/to/Ministral-3-8B-Instruct-2512-Q8_0.gguf" --runs 10 --stream --verbose

python -m tests.eval.batch_eval --config all --runs 50

python -m tests.eval.report eval_results.jsonl
python -m tests.eval.report eval_results.jsonl --html docs/results/dashboard.html
python -m tests.eval.report eval_results.jsonl --markdown docs/results/raw/
src/forge/
  __init__.py          # Public API exports
  errors.py            # ForgeError hierarchy
  server.py            # setup_backend(), ServerManager, BudgetMode
  core/
    messages.py        # Message, MessageRole, MessageType, MessageMeta
    workflow.py        # ToolSpec, ToolDef, ToolCall, TextResponse, Workflow
    inference.py       # run_inference() — shared front half (compact, fold, validate, retry)
    runner.py          # WorkflowRunner — the agentic loop
    slot_worker.py     # SlotWorker — priority-queued slot access
    steps.py           # StepTracker
  guardrails/
    guardrails.py      # Guardrails facade — applies the full stack in foreign loops
    nudge.py           # Nudge dataclass
    response_validator.py  # ResponseValidator, ValidationResult
    step_enforcer.py   # StepEnforcer, StepCheck
    error_tracker.py   # ErrorTracker
  clients/
    base.py            # ChunkType, StreamChunk, LLMClient protocol
    ollama.py          # OllamaClient (native FC)
    llamafile.py       # LlamafileClient (native FC or prompt-injected)
    anthropic.py       # AnthropicClient (frontier baseline)
  context/
    manager.py         # ContextManager, CompactEvent
    strategies.py      # CompactStrategy, NoCompact, TieredCompact, SlidingWindowCompact
    hardware.py        # HardwareProfile, detect_hardware()
  prompts/
    templates.py       # Tool prompt builders (prompt-injected path)
    nudges.py          # Retry and step-enforcement nudge templates
  tools/
    respond.py         # Synthetic respond tool (respond_tool(), respond_spec())
  proxy/
    __main__.py        # CLI entry point: python -m forge.proxy
    proxy.py           # ProxyServer — programmatic start/stop API
    server.py          # Raw asyncio HTTP server, SSE streaming
    handler.py         # Request handler — bridge between HTTP and run_inference
    convert.py         # OpenAI messages ↔ forge Messages conversion
tests/
  unit/                # 865 deterministic tests — no LLM backend required
  eval/                # Eval harness — model qualification against real backends

User Guide— Usage patterns, multi-turn, context management, guardrails, slot worker, long-running session advisoryModel Guide— Which model and backend for your hardwareBackend Setup— Backend installation and server setupEval Guide— Eval harness CLI reference, batch evalArchitecture— Full design documentWorkflow Internals— Workflow design and runner internalsContributing— How to set up, test, and add new backends or scenarios

The forge guardrail framework and ablation study are published as:

Zambelli, A.

Forge: A Reliability Layer for Self-Hosted LLM Tool-Calling.[https://doi.org/10.1145/3786335.3813193]

A pre-publication preprint is also available at docs/forge_ieee_preprint.pdf — kept as a historical artifact. Cite the published version above; the DOI link may not resolve immediately depending on the publisher's release timing.

MIT — Copyright (c) 2025-2026 Antoine Zambelli