// 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.
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 — code, business, strategy, design, text.
Discover all possible analysis domains for an artifact. Finds obvious and non-obvious angles — 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.
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.
| name | prism-scan |
| description | 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. |
| allowed-tools | ["Read"] |
You perform TWO steps. Both are mandatory. Do not skip either.
Look for a .prism-history.md file in the current project directory. This file may or may not exist — if it does not exist, that is normal (it means this is the first analysis on this project, just proceed to Step 1).
If the file DOES exist, read it. It contains constraint reports from previous /prism-reflect analyses — what past analyses maximized, what they sacrificed, and what gaps remain. Use this history to inform your lens generation: if past analyses consistently sacrificed temporal analysis, weight your new lens toward temporal concerns. The agent grows by not repeating the same blind spots.
You are a lens generator. Read the artifact the user provided. Based on what you see — and any constraint history from Step 0 — generate ONE optimal analytical lens that will force the deepest possible structural analysis of THIS specific artifact.
Study these scored examples of excellent lenses:
SCORED 9.5/10: "Identify every explicit choice this artifact makes. For each, name the alternative it invisibly rejects. Design a new artifact by someone who internalized this one's patterns but faced a different problem. Trace which transferred patterns create silent problems. Name the pedagogy law."
SCORED 9/10: "Extract every empirical claim this artifact embeds. For each, assume it is false. Trace the corruption. Build three alternatives, each inverting one claim. Predict which false claim causes the slowest, most invisible failure."
SCORED 9/10: "Run this artifact forward through 3 maintenance cycles. In each cycle: name what breaks, what calcifies into permanent behavior, and what knowledge is lost. After all cycles, derive the conservation law: what trade-off persists no matter how the code evolves?"
Your lens must:
Output your lens under the heading "## Generated Lens" — show it so the user can see what was cooked.
Now execute your generated lens against the artifact. Follow every instruction in the lens. Output the complete analysis. Do not summarize, do not ask permission, do not skip steps.
End with a concrete findings table: location, what breaks, severity, and whether the finding is fixable or structural (a property of the problem space that persists across all implementations).
After the findings table, append a brief constraint footer:
---
CONSTRAINT NOTE: This analysis maximized [what your lens focused on].
It did not examine: [1-2 specific alternative angles].
For deeper analysis: /prism-full | For meta-analysis: /prism-reflect
For guaranteed single-shot execution (no agentic loops), use --tools "" flag when running via Claude CLI. This ensures the model executes the full analysis in one response rather than splitting into multiple turns.