generate launch config

$0.50 USDC: generate deterministic vLLM or llama.cpp launch templates from an exact StackBench recommendation without claiming a launch occurred.

liveverifiedunknown~5s typical

mcpprobedcomputesagestackbench

$0.0020

per call

plus $0.001 platform fee per run

Open MCP server; Cracked bills routing only

Provider Upstream docs

$Set up https://cracked-ai-pearl.vercel.app/SKILL.md, then use Cracked to run mcp-com-computesage-stackbench /generate_launch_config for me.

Input

Field	Type	Description
runtime*	string (vllm \| llama.cpp)
model*	string
hardware_system*	string
prompt_tokens*	integer
quantization	string
engine_version	string
decode_context_depth	integer
batch_size	integer
ubatch_size	integer
concurrency	integer	default 1
flash_attention	string
candidate_kv_formats	string
minimum_prefill_tok_s	number
minimum_decode_tok_s	number
maximum_vram_bytes	integer
maximum_power_watts	number
objective	string (balanced \| maximize_decode \| maximize_prefill \| minimize_vram \| minimize_power \| minimize_capex)	default "balanced"
constraint_policy	string (point_estimate \| conservative)	default "point_estimate"
evidence_policy	string (verified_only \| include_public \| all)	default "include_public"

Call it

curl

curl https://cracked-ai-pearl.vercel.app/v1/run \
  -H "Authorization: Bearer ck_live_..." -H "content-type: application/json" \
  -d '{"provider":"mcp-com-computesage-stackbench","endpoint":"/generate_launch_config","input":{"runtime":"vllm","model":"...","hardware_system":"...","prompt_tokens":1,"concurrency":1,"objective":"balanced","constraint_policy":"point_estimate","evidence_policy":"include_public"}}'

cli

npx cracked-ai run -p mcp-com-computesage-stackbench -e /generate_launch_config -i '{"runtime":"vllm","model":"...","hardware_system":"...","prompt_tokens":1,"concurrency":1,"objective":"balanced","constraint_policy":"point_estimate","evidence_policy":"include_public"}'

mcp

run_tool({ provider: "mcp-com-computesage-stackbench", endpoint: "/generate_launch_config", input: {"runtime":"vllm","model":"...","hardware_system":"...","prompt_tokens":1,"concurrency":1,"objective":"balanced","constraint_policy":"point_estimate","evidence_policy":"include_public"} })

Try it

Runs against your signed-in workspace balance. Sign in if you have not.