Organisations accumulate phantom constraints — rules that were real once and calcified, or beliefs that were never tested, or someone's preference that got repeated until it sounded like policy. This skill separates genuine limits from assumed ones before any energy is spent working around or accepting them.
Your Process
Step 1: State the Constraint Write it exactly as it's been stated or assumed. Don't clean it up — the imprecision is often where the phantom lives.
Step 2: Source It Who said this constraint exists? When? Is the source a law or regulation, a signed contract, a technical impossibility, a leadership decision, a team preference, or unknown? The source determines the hardness ceiling.
Step 3: Consequence Test What actually happens if this constraint is violated? State the consequence concretely. If the answer is "I'm not sure" or "someone would be unhappy," the constraint is not hard. Distinguish real consequences (fine, contract breach, system failure) from assumed ones (pushback, awkwardness, political cost).
Step 4: Precedent Check Has anyone tried to change or violate this constraint before? What happened? No precedent often means no one has tested it — not that it can't be changed.
Step 5: Conditions Test Under what circumstances would this constraint not apply? A truly hard constraint has no exceptions. If you can find a scenario where it wouldn't hold, the constraint is softer than stated.
Step 6: Classify
- Hard: real source, concrete consequence, no exceptions, precedent confirms
- Soft: real but negotiable, consequence is political or preferential
- Assumed: no clear source, untested consequence
- Outdated: was once hard, circumstances have changed
Human Check-in
Before proceeding, use the AskUserQuestion tool:
- Question: "How do you want to run this?"
- Header: "Scope"
- Options:
- Full analysis — Complete all steps, reasoning shown throughout
- Key findings only — Bottom-line output, skip step-by-step detail
- Real vs assumed verdict only — Classify each constraint without full analysis
- Refine the framing — Adjust what we're analyzing before starting
Proceed based on their selection.
Output Format
Constraint as stated:
[Exact wording]
Source: [Law/contract/technical/decision/preference/unknown] — [specific origin]
Consequence if violated:
[Concrete statement] — Real / Assumed
Precedent: [What happened when tested, or "untested"]
Conditions where it wouldn't apply:
[If any]
Classification: Hard / Soft / Assumed / Outdated
Recommended action:
| Classification | Action |
|---|---|
| Hard | Accept — design around it |
| Soft | Negotiate explicitly |
| Assumed | Test it — ask the source directly |
| Outdated | Challenge it — propose removal |
Notes
The most dangerous class is Assumed, because it looks like Hard. Before accepting any constraint that cannot be sourced precisely, treat it as Assumed and test it — the cost of testing is almost always lower than the cost of a permanent workaround.
What's Next
After delivering this output, use AskUserQuestion to offer the next move:
- Question: "Constraints tested. What's next?"
- Header: "Next"
- Options:
/constraint-workaround-mapping— Route around the hard constraints/constraint-scope-reduction— Reduce scope to avoid constraints that can't be bypassed/decision-option-mapping— See what options remain given hard and soft constraints- Done — Wrap up and synthesise what we have so far