MCP Tools Catalog

The QCOS MCP server exposes 71 tools to your AI assistants. Rather than typing complex CLI commands or writing raw Python scripts, you can converse with your assistant in natural language, and the agent will call the appropriate tools behind the scenes to run the tasks.

1. Core Tool Categories

Below is the index of the most frequently used tools mapped by functional category:

Category	Stable Tool Name	Description
Solvers & Status	`qcos_get_v1_runtime_backends` `qcos_get_v1_backends` `qcos_get_health`	List all available backends with queue depth and status, and probe API health.
Estimation & Planning	`qcos_post_v1_runtime_estimate` `qcos_autopilot_plan` `qcos_recommend_qubit_network`	Estimate cost/approval state and plan noise-aware qubit routing before a run.
Execution	`qcos_post_v1_runtime_jobs` `qcos_get_v1_runtime_jobs_job_id` `qcos_get_v1_jobs_job_id_result` `qcos_delete_v1_runtime_jobs_job_id` `qcos_post_v1_runtime_jobs_job_id_approve`	Submit circuits (202), poll execution states, retrieve counts, cancel, and approve priced runs.
Algorithmic Templates	`qcos_run_grover` `qcos_run_shor` `qcos_run_qaoa` `qcos_run_qft`	Run standard quantum algorithms using optimized, pre-built template parameters.
Interactive Analytics	`qcos_visualize_bloch` `qcos_visualize_histogram` `qcos_visualize_circuit`	Render Bloch spheres, counts histograms, and circuit diagrams.
Error Correction	`qcos_qec_pipeline` `qcos_qec_decoder_decode` `qcos_qec_benchmark`	Run the QEC pipeline, decode syndromes, and benchmark code performance.
Evidence & Audit	`qcos_get_v1_runtime_jobs_job_id_execution_evidence` `qcos_search_jobs` `qcos_get_job_timeline`	Retrieve sealed execution evidence, search jobs, and get audit timelines.
Security & Randomness	`qcos_qrng_uuid` `qcos_qrng_bytes` `qcos_pqc_generate_keypair` `qcos_pqc_sign`	Quantum randomness and post-quantum (ML-DSA) signature primitives.

2. Step-by-Step Agent Execution Flows

When you interact with an agent, it will automatically orchestrate multiple tool calls. Below are common agent invocation sequences:

Flow A: Multi-Device Planning and Execution

User Prompt: “Find the healthiest backend with at least 8 qubits and simulate my circuit first before running it.”
Agent Execution Sequence:
1. qcos_get_v1_runtime_backends: Agent fetches all available solvers.
2. qcos_calibration_analyze: Agent checks the measured device state of the candidates to find the healthiest one.
3. qcos_autopilot_plan: Agent asks Autopilot for a run/remap/defer/block decision on the chosen backend.
4. qcos_post_v1_runtime_jobs: Agent executes on simulator_aer to test behavior.
5. qcos_get_v1_jobs_job_id_result: Agent reads counts and confirms correct execution.

Flow B: Cost Auditing and Cloud QPU Submission

User Prompt: “Estimate the cost to run this QASM with 50,000 shots on ibm_brisbane. If it’s under 10 credits, submit it.”
Agent Execution Sequence:
1. qcos_post_v1_runtime_estimate: Agent computes the credit reservation and approval state.
2. Audit Verification: Agent prints the estimate and checks user budgets.
3. qcos_post_v1_runtime_jobs: If within budget, agent submits the circuit to ibm_brisbane.
4. qcos_post_v1_runtime_jobs_job_id_approve: The human approves the priced run with the returned token.
5. qcos_get_v1_runtime_jobs_job_id: Agent tracks job completion status.

3. Example Natural Language Prompts

Try pasting these prompts directly into your MCP-enabled chat client:

Backend Diagnostics:
“List all QCOS backends and show their active queue lengths. Are any of them experiencing calibration drift?”
Workflow Simulation:
“Take this QASM circuit, check if its syntax is valid, run it locally on the simulator for 2048 shots, and display the counts in a formatted markdown table.”
Algorithmic Solutions:
“I need to search an unstructured database of 16 items. Run Grover’s algorithm to search for the element ‘1011’ and print the final counts and Bloch sphere coordinates.”
Cost Auditing:
“Compare running a 20-qubit circuit on two backends. What is the difference in estimated credit reservations, and would either need approval?”

​MCP Tools Catalog

​1. Core Tool Categories

​2. Step-by-Step Agent Execution Flows

​Flow A: Multi-Device Planning and Execution

​Flow B: Cost Auditing and Cloud QPU Submission

​3. Example Natural Language Prompts