oracle-prism
Skill0 starsChanged 2 months agoMITView on GitHub
Multi-perspective analysis — one agent refracts through N named lenses in sequence, no subagents. Use when user says "prism", "oracle-prism", "multi-perspective", "look at this from different angles", "what am I missing", "did we miss anything", "critique this from N angles", or asks for a design/retro/incident review through several viewpoints. Presets - retro, design, incident, designer. Do NOT trigger for adversarial verification of a single claim (use /adversarial-analysis) or for spawning parallel agents (use /sonnet).
1 other project uses a version of this, 1 of them edited.
Adopted by
Most recently taken up by
- Soul-Brews-Studio/arra-oracle-skills-cliedited, first seen Apr 13, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 1
- Soul-Brews-Studio/arra-oracle-skills-cli★ 12178% same
Wrote their own version · 0
None found.
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 repository4 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- oracle-facetSkill
- oracle-roundtableSkill
- plaeng-rangSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 80
Use --custom to define your own, or pick a preset:
≈ “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
/oracle-prism — Multi-Perspective Analysis
"แสงเดียวผ่านปริซึม แตกเป็นหลายสี — เรื่องเดียวกัน มองจากหลายมุม เห็นต่างกัน"
One agent, five perspectives, no subagents. Transform inline — shift between lenses to dig through work, decisions, or problems from angles that a single viewpoint would miss.
Usage
/oracle-prism # Analyze current session (default)
/oracle-prism "the rename migration" # Analyze a specific topic
/oracle-prism --lenses 3 # Use 3 lenses instead of 5
/oracle-prism --custom "Security Auditor,Performance Engineer,UX Designer"
/oracle-prism --tables # inventory mode — tables instead of prose
/oracle-prism --dna # each lens names the real practitioners behind itHow It Works
No subagents. The main agent transforms between perspectives sequentially, producing one section per lens. Each lens sees the same facts but asks different questions.
This is NOT adversarial analysis (which tries to disprove). This is prismatic analysis — same light, different colors.
Default 5 Lenses
| # | Lens | Emoji | Question it asks |
| 1 | Archaeologist | 🔍 | "What actually happened? Timeline, sequence, facts." |
| 2 | Bug Hunter | 🐛 | "What problems did we hit? What's still broken?" |
| 3 | Skeptic | 💀 | "What did we do wrong? What would we redo?" |
| 4 | Architect | 🏗️ | "What changed structurally? What's the before/after?" |
| 5 | Auditor | 📋 | "What's left undone? What's inconsistent?" |
DNA — the real people behind each lens
--dna makes each lens declare its inheritance before it speaks: the named,
documented practitioners whose method it is running. Boy Pinyo's distillation
pattern — a persona that enumerates which real humans contributed which traits
is more honest than a persona that is an unattributed blob.
Rule 6 governs this absolutely. The lens speaks as the archetype, never
as the person. Write **💀 Skeptic** — DNA: Feynman (don't fool yourself) · …
and then reason. Never As Feynman, I…. Never a quote you cannot attribute. If
a documented position is being paraphrased, say paraphrase. For one named
master applied in depth, that is /oracle-facet, not this.
| Lens | DNA — what each contributes |
| 🔍 Archaeologist | Marc Bloch (evidence before narrative) · Robert Caro (exhaust the record) · Edward Tufte (show the data, not the claim) |
| 🐛 Bug Hunter | John Carmack (run it, measure it) · Margaret Hamilton (error handling is the system) · Brian Kernighan (debugging is harder than writing) |
| 💀 Skeptic | Richard Feynman (the first principle is not to fool yourself) · Nassim Taleb (what is fragile) · Sidney Dekker (blame hides cause) |
| 🏗️ Architect | Christopher Alexander (patterns, not parts) · Barbara Liskov (substitutability) · Rob Pike (simplicity is the design) |
| 📋 Auditor | W. Edwards Deming (the system produces the result) · Joseph Juran (the vital few) · Atul Gawande (the checklist catches the obvious) |
| 👤 User | Don Norman (affordance and signifier) · Jakob Nielsen (heuristics) · Susan Kare (recognisable at a glance) |
| 🔧 Maintainer | Michael Feathers (legacy is code without tests) · Barbara Liskov (contracts survive people) · Rob Pike |
| 💥 Breaker | Sidney Dekker (drift into failure) · Nancy Leveson (safety is a control problem) · Nassim Taleb |
| ✂️ Simplifier | Dieter Rams (as little design as possible) · John Maeda (laws of simplicity) · Jony Ive (reduce toward inevitable) |
| 🔗 Integrator | Massimo Vignelli (one system, everywhere) · Kenya Hara (the whole context) · Tim Berners-Lee (interoperate) |
| 📜 Historian | Marc Bloch · Robert Caro |
| ✏️ Critic | Pauline Kael (say what you actually think) · Jane Jacobs (observe before theorising) |
| 📣 Cheerleader | Tom Peters (celebrate what works) · Teresa Amabile (progress is the motivator) |
| 🧩 Connector | James Burke (connections across domains) · Steven Johnson (the adjacent possible) |
| 🎯 Planner | Andy Grove (output, not activity) · Dwight Eisenhower (planning over plans — paraphrase of a documented remark) |
| 🚒 Firefighter | Richard Cook (how complex systems fail) · Gene Kranz (work the problem) |
| 🔎 Detective | Sidney Dekker · Nancy Leveson |
| 🛡️ Defender | James Reason (the Swiss cheese model) · Charles Perrow (normal accidents) |
| 🔮 Forecaster | Philip Tetlock (calibrate, then score) · Nate Silver (signal vs noise) |
| 🧱 Builder | Taiichi Ohno (stop the line, fix the cause) · W. Edwards Deming |
Named influences describe the angle taken — not endorsement, not expertise transferred, and not a claim that anyone reviewed this.
Alternate Lens Sets
Use --custom to define your own, or pick a preset:
--preset retro (session retrospective)
| Lens | Question |
| Historian | What happened, in what order? |
| Critic | What went poorly or slowly? |
| Cheerleader | What went well? What should we repeat? |
| Connector | What patterns connect to past work? |
| Planner | What's the next move? |
--preset design (design review)
| Lens | Question |
| User | Is this easy to use? What's confusing? |
| Maintainer | Is this easy to change later? |
| Breaker | How can this fail? Edge cases? |
| Simplifier | What can be removed? |
| Integrator | How does this fit with everything else? |
--preset incident (post-incident)
| Lens | Question |
| Firefighter | What happened and how was it fixed? |
| Detective | What was the root cause chain? |
| Defender | What guards existed? Why didn't they catch it? |
| Forecaster | What similar things could happen next? |
| Builder | What systemic fix prevents recurrence? |
Output Format
Prose by default. One tight paragraph per lens. Bold the pivot word inside the sentence, and close each lens with a bold verdict — the one line someone would quote back. This is what lets a prism carry 7–10 lenses without becoming a spreadsheet.
### 🔍 Lens 1: Archaeologist — "What happened?"
*DNA: Bloch (evidence before narrative) · Caro (exhaust the record)* ← only with --dna
The rename landed in three passes, not one — `a1b2c3` moved the files, `d4e5f6`
fixed the imports it broke, `g7h8i9` caught the two it missed. The gap between
them is where the **stale references** survived. **Archaeologist verdict**: the
migration is done, its cleanup is not.
### 🐛 Lens 2: Bug Hunter — "What broke?"
[one paragraph, same shape]
### 💀 Lens 3: Skeptic — "What went wrong?"
[…]
---
**Cross-lens summary:** [2-3 sentences on what multiple lenses converge on]Reach for a table only when the content is genuinely inventory — a timeline, a status list, a before/after, a per-file comparison. A judgement ("is this confusing?", "is the voice consistent?") is prose; a count is a table. Mixing is fine: table the Auditor, keep the Skeptic in prose.
--tables forces the inventory style throughout, for when the whole subject is
a list.
Rules
- No subagents — all lenses run in the main agent, sequentially
- Each lens gets a section with its emoji, name, and guiding question as header
- Evidence required — cite files, commits, commands, timestamps. No vague claims
- Lenses disagree — if the Architect says "clean" but the Auditor says "incomplete", show both. Don't harmonize
- Cross-lens summary at the end — what do multiple lenses converge on?
- Prose over tables — one paragraph per lens, bold pivot, bold verdict. Tables only for genuine inventory (timelines, status lists, before/after), or throughout with
--tables - Session context by default — if no topic given, analyze the current session's work
7b. DNA is inheritance, not impersonation — with
--dna, name the practitioners in a one-line italic under the header, then reason as the archetype. Rule 6: never first person as a real person, never a fabricated quote - 3-7 lenses — minimum 3, maximum 7. Default 5. Use
--lenses Nto adjust. Prose scales further than tables do; 9-10 lenses is workable in prose mode
When to Use
| Situation | Why prism helps |
| End of session | Catch what you missed before closing |
| After a migration/rename | Find inconsistencies across the change |
| Design decision | See tradeoffs from user/maintainer/breaker angles |
| Post-incident | Structured multi-angle without blame |
| "Did we miss anything?" | The Auditor lens exists for this |
Relationship to Other Skills
| Skill | Pattern | Agents |
/oracle-prism | Multi-perspective, same agent | 0 (inline transform) |
/adversarial-analysis | Try to disprove a claim | 5 parallel subagents |
/rrr | Session retrospective | 0 (or 5 in --deep) |
/roundtable | Discussion between personas | 0 (inline) |
Prism is the lightest multi-perspective tool — no agents, no spawning, no coordination overhead. Use it when you want angles, not adversaries.
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.

