workflow-planning
Skill69 starsChanged 7 months agoApache-2.0View on GitHub
Plan a new calibration workflow by discussing experiment sequences, success/failure criteria, and extracted parameters with the user before any files are created. Use when the user asks to plan, design, or create a new calibration sequence, or to build a multi-step workflow for a qubit or device. Requires explicit user confirmation before writing workflow files.
3 other projects use a version of this.
Adopted by
Most recently taken up by
- unional/skillstheir own version, first seen Nov 6, 2026
- IgorBabikov/Flowcairntheir own version, first seen Nov 1, 2026
- microsoft/agora-workbenchtheir own version, first seen Oct 13, 2026
Other versions3 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 3
- microsoft/agora-workbench★ 10same name
- IgorBabikov/Flowcairn★ 4same name
- unional/skills★ 1same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository5 other files it gives its agents.
- analysis-scriptsSkill
- experiment-executionSkill
- vlm-configurationSkill
- workflow-executionSkill
- writing-experiment-scriptsSkill
The file
We split this skill into the rules it gives an agent
10 rules. About 0 are common, 6 are distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈6
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 8
Create calibration workflows through discussion with the user.
- Very specific: line 13
Read memory/workflow-preferences.md if it exists
- Distinctive: line 14
Look for past conventions, preferred experiment sequences, success criteria patterns
- Distinctive: line 49
Wait for explicit confirmation like:
- Distinctive: line 55
If the user says "don't ask, just do it" for future workflows, note this in memory.
- Distinctive: line 59
If the user corrects your decisions:
- Distinctive: line 84
After confirmation, create the workflow files.
- Very specific: line 101
Use a descriptive ID like calibrate_q0_20240315.
- Very specific: line 103
IMPORTANT: Always use relative paths: when creating workflow files.
- Very specific: line 238
Read the current workflow.json
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Workflow Planning
Create calibration workflows through discussion with the user. A workflow defines a sequence of operations with dependencies.
Before Planning
Always check memory first:
- Read
memory/workflow-preferences.mdif it exists - Look for past conventions, preferred experiment sequences, success criteria patterns
- Apply any user preferences to your proposal
Planning Process
Step 1: Understand the Goal
Ask the user:
- What is the calibration objective?
- Which qubits/targets are involved?
- Any specific constraints or requirements?
Step 2: Propose the Sequence
Based on the goal, propose a sequence of nodes. For each node, discuss:
- Experiment type - Which experiment to run
- Success criteria - How to know it worked
- Failure criteria - How to know it failed
- What to extract - Parameters to carry forward
Example discussion:
For Node 1 (Resonator Spectroscopy), I suggest:
- Success: Clear dip with depth > 3 dB
- Failure: No dip visible or multiple dips
- Extract: resonator_frequency
Does this match your expectations? Any adjustments?Step 3: Wait for Confirmation
Do NOT create the workflow until the user confirms.
Wait for explicit confirmation like:
- "go ahead"
- "create it"
- "looks good, proceed"
- "confirmed"
If the user says "don't ask, just do it" for future workflows, note this in memory.
Step 4: Remember Corrections
If the user corrects your decisions:
- Apply the correction to the current workflow
- Update memory with the preference for future workflows
Example corrections to remember:
- "Always use SNR > 10, not SNR > 5"
- "For T1 measurements, extract t1_time in microseconds"
- "Start resonator sweeps at 5.5 GHz for this device"
Write to memory/workflow-preferences.md:
## Success Criteria Preferences
- Resonator spectroscopy: depth > 3 dB, SNR > 10
- Qubit spectroscopy: SNR > 10 (not 5)
## Parameter Conventions
- T1 times in microseconds
- Frequencies in GHz
## Device-Specific
- Resonator range: 5.5 - 6.5 GHzCreating the Workflow
After confirmation, create the workflow files.
Storage Structure
data/workflows/
└── {workflow_id}/
├── workflow.json # DAG structure
└── plan.md # Detailed planning documentStep 1: Create Folder
mkdir -p data/workflows/{workflow_id}Use a descriptive ID like calibrate_q0_20240315.
IMPORTANT: Always use relative paths when creating workflow files. Use data/workflows/{workflow_id}/workflow.json, NOT absolute paths like /home/.../data/workflows/.... Absolute paths will fail due to virtual filesystem mode.
Step 2: Write workflow.json
{
"id": "calibrate_q0",
"name": "Calibrate Qubit Q0",
"objective": "Full calibration sequence for Q0",
"status": "created",
"nodes": [
{
"id": "node_1",
"name": "Resonator Spectroscopy",
"dependencies": []
},
{
"id": "node_2",
"name": "Qubit Spectroscopy",
"dependencies": ["node_1"]
}
]
}Required fields:
id: Must match folder namename: Human-readable namenodes: Array withid,name,dependencies
Status values: created, running, completed, failed
Step 3: Write plan.md
Include all discussed details:
# Calibration Plan: Calibrate Qubit Q0
## Objective
Full calibration sequence for qubit Q0.
## Node 1: Resonator Spectroscopy
**Experiment type**: ResonatorSpectroscopy
**Parameters**:
- Start frequency: 5.5 GHz
- Stop frequency: 6.5 GHz
- Number of points: 101
**Success criteria**:
- Clear dip visible
- Depth > 3 dB
- SNR > 10
**Failure criteria**:
- No dip visible
- Multiple dips (mode crossing)
**Extract**: `resonator_frequency`
---
## Node 2: Qubit Spectroscopy
**Experiment type**: QubitSpectroscopyFrequency
**Parameters**:
- Center: Use expected qubit frequency
- Span: 200 MHz
**Success criteria**:
- Single clear peak
- SNR > 10
**Failure criteria**:
- No peak visible
- Multiple peaks
**Extract**: `qubit_frequency`Step 4: Validate (REQUIRED)
You MUST call validate after creating the workflow files:
workflow(action="validate", workflow_id="calibrate_q0")The validation returns detailed checks:
valid: Whether the workflow is validchecks: List of individual validation checks with pass/fail statuserrors: Critical issues that must be fixedwarnings: Non-critical issues to considerworkflow_info: Summary of the workflow
If validation fails:
- Review the
errorslist - Fix each error in workflow.json
- Re-run validate until
valid: true
Do NOT consider the workflow creation complete until validation passes.
Dependencies
Nodes execute in dependency order:
node_1 ──→ node_2 ──→ node_3
│
└──→ node_4In JSON:
{
"nodes": [
{"id": "node_1", "name": "First", "dependencies": []},
{"id": "node_2", "name": "Second", "dependencies": ["node_1"]},
{"id": "node_3", "name": "Third A", "dependencies": ["node_2"]},
{"id": "node_4", "name": "Third B", "dependencies": ["node_2"]}
]
}Listing Workflows
workflow(action="list")Modifying a Workflow
- Read the current
workflow.json - Edit using the Edit tool
- Re-validate (REQUIRED):
workflow(action="validate", workflow_id="your_workflow_id") - Fix any errors and re-validate until
valid: true
You MUST validate after ANY modification to workflow.json.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

