burnrate track

Log the actual cost of an LLM call after execution. Call this after every LLM request to build calibration data that improves burnrate_estimate accuracy over ti

liveverifiedunknown~5s typical

mcpprobedplith

$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-ai-plith-plith /burnrate_track for me.

Input

Field	Type	Description
provider*	string	LLM provider identifier. Supported: anthropic, openai, google, mistral, cohere, deepseek,
model*	string	Model identifier as returned by the provider. Examples: claude-sonnet-4-6, gpt-4o, gemini-
input_tokens*	number	Actual prompt tokens used. Must be >= 0.
output_tokens*	number	Actual completion tokens used. Must be >= 0.
task_id	string	Optional task ID for cross-referencing spend with DedupQ deduplication results. Use the sa
cache_read_tokens	number	Optional. Cache-read tokens.

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-ai-plith-plith","endpoint":"/burnrate_track","input":{"provider":"...","model":"...","input_tokens":1,"output_tokens":1}}'

cli

npx cracked-ai run -p mcp-ai-plith-plith -e /burnrate_track -i '{"provider":"...","model":"...","input_tokens":1,"output_tokens":1}'

mcp

run_tool({ provider: "mcp-ai-plith-plith", endpoint: "/burnrate_track", input: {"provider":"...","model":"...","input_tokens":1,"output_tokens":1} })

Try it

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