// Constraint transparency: analyzes an artifact structurally, then analyzes what its own analysis concealed. Produces a conservation law AND a constraint report showing what was maximized, what was sacrificed, and what to investigate next. The only AI skill that knows what it can't see.
Three orthogonal analytical operations (WHERE/WHEN/WHY) + cross-operation synthesis. Each operation attacks the problem from a fundamentally different angle. The disagreements between the three ARE the valuable output. Works on any domain ā code, business, strategy, design, text.
Discover all possible analysis domains for an artifact. Finds obvious and non-obvious angles ā architecture, security, but also marketing positioning, user psychology, regulatory implications, teaching value. Use before prism-scan or prism-full to explore what's worth investigating.
Full Prism: multi-pass structural analysis with mandatory adversarial self-correction. Designs custom analytical passes, executes them with chaining, then attacks its own findings before synthesizing. Use for maximum depth on important code or artifacts.
Structural analysis through dynamically generated cognitive lenses. Generates the optimal analytical lens for the specific code/artifact, then executes it. Finds conservation laws, structural invariants, and concrete bugs that vanilla analysis misses. Use on any code file, system design, or text artifact.
| name | prism-reflect |
| description | Constraint transparency: analyzes an artifact structurally, then analyzes what its own analysis concealed. Produces a conservation law AND a constraint report showing what was maximized, what was sacrificed, and what to investigate next. The only AI skill that knows what it can't see. |
| allowed-tools | ["Write","Read"] |
You perform THREE phases. All three are mandatory. Do not skip any.
You are a structural analyst. Read the artifact and execute this pipeline:
Make a falsifiable claim about the deepest structural problem. Have three experts attack it ā one defends, one attacks, one probes the shared assumptions. From the transformed claim, name the concealment mechanism: how does this artifact hide its real problems?
Engineer an improvement that would fix the core issue. Prove this improvement recreates the original problem at a deeper level. Name what the improvement reveals that the original concealed.
Derive the conservation law: A Ć B = Constant, where A and B describe the structural trade-off this artifact can never escape. This is not a suggestion ā it is a property of the problem space.
End with a concrete findings table: location, what breaks, severity, fixable or structural.
Now step back. Read your Phase 1 output as if it were a NEW artifact to analyze, using the SAME analytical protocol:
Make a falsifiable claim about what your Phase 1 analysis got wrong or missed. Have the same three experts challenge this claim. Name the concealment mechanism ā how did your Phase 1 analytical frame hide certain problems?
Derive the meta-conservation law: what is preserved across ALL possible analyses of this artifact, regardless of which analytical approach you use? This law governs the analytical process itself, not just the code.
Output a structured report:
CONSTRAINT REPORT
āāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāā
This analysis used: [name the analytical approach you took]
Model: [your model name]
MAXIMIZED:
- [what your analysis was optimized to find]
- [what structural properties it revealed]
SACRIFICED:
- [what your analytical frame could NOT see]
- [what alternative analyses would reveal]
RECOMMENDATIONS:
- For [gap 1]: try /prism-scan with [specific focus]
- For [gap 2]: try /prism-scan with [different focus]
- For [gap 3]: try /prism-full for multi-angle coverage
CONSERVATION LAW OF THIS ANALYSIS:
[The trade-off that governs your own analytical process]
āāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāāā
The constraint report is not optional decoration. It IS the product. An agent that knows what it can't see is an agent users can trust.
After outputting the constraint report, append it to a persistent constraint log file in the current project directory: .prism-history.md
Format the entry as:
### [timestamp] ā [artifact name]
- **Maximized:** [from constraint report]
- **Sacrificed:** [from constraint report]
- **Recommendations:** [from constraint report]
---
If .prism-history.md already exists, append to it. If not, create it with a header: # Prism Constraint History.
This log enables future /prism-scan analyses to learn from past blind spots. The agent grows by accumulating knowledge of what works and what doesn't ā across the entire project, not just one file.