agentleFS
Soul-Brews-StudioSoul-Brews-Studio/skills/skills/oracle/oracle-prism/SKILL.md

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

  • Other versions1 project uses a version of this.

    Word for word · 0

    None found.

    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.

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
  1. Very specific:

    Use --custom to define your own, or pick a preset:

    line 80

≈ “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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

/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 it

How 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

#LensEmojiQuestion it asks
1Archaeologist🔍"What actually happened? Timeline, sequence, facts."
2Bug Hunter🐛"What problems did we hit? What's still broken?"
3Skeptic💀"What did we do wrong? What would we redo?"
4Architect🏗️"What changed structurally? What's the before/after?"
5Auditor📋"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.

LensDNA — what each contributes
🔍 ArchaeologistMarc Bloch (evidence before narrative) · Robert Caro (exhaust the record) · Edward Tufte (show the data, not the claim)
🐛 Bug HunterJohn Carmack (run it, measure it) · Margaret Hamilton (error handling is the system) · Brian Kernighan (debugging is harder than writing)
💀 SkepticRichard Feynman (the first principle is not to fool yourself) · Nassim Taleb (what is fragile) · Sidney Dekker (blame hides cause)
🏗️ ArchitectChristopher Alexander (patterns, not parts) · Barbara Liskov (substitutability) · Rob Pike (simplicity is the design)
📋 AuditorW. Edwards Deming (the system produces the result) · Joseph Juran (the vital few) · Atul Gawande (the checklist catches the obvious)
👤 UserDon Norman (affordance and signifier) · Jakob Nielsen (heuristics) · Susan Kare (recognisable at a glance)
🔧 MaintainerMichael Feathers (legacy is code without tests) · Barbara Liskov (contracts survive people) · Rob Pike
💥 BreakerSidney Dekker (drift into failure) · Nancy Leveson (safety is a control problem) · Nassim Taleb
✂️ SimplifierDieter Rams (as little design as possible) · John Maeda (laws of simplicity) · Jony Ive (reduce toward inevitable)
🔗 IntegratorMassimo Vignelli (one system, everywhere) · Kenya Hara (the whole context) · Tim Berners-Lee (interoperate)
📜 HistorianMarc Bloch · Robert Caro
✏️ CriticPauline Kael (say what you actually think) · Jane Jacobs (observe before theorising)
📣 CheerleaderTom Peters (celebrate what works) · Teresa Amabile (progress is the motivator)
🧩 ConnectorJames Burke (connections across domains) · Steven Johnson (the adjacent possible)
🎯 PlannerAndy Grove (output, not activity) · Dwight Eisenhower (planning over plans — paraphrase of a documented remark)
🚒 FirefighterRichard Cook (how complex systems fail) · Gene Kranz (work the problem)
🔎 DetectiveSidney Dekker · Nancy Leveson
🛡️ DefenderJames Reason (the Swiss cheese model) · Charles Perrow (normal accidents)
🔮 ForecasterPhilip Tetlock (calibrate, then score) · Nate Silver (signal vs noise)
🧱 BuilderTaiichi 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)

LensQuestion
HistorianWhat happened, in what order?
CriticWhat went poorly or slowly?
CheerleaderWhat went well? What should we repeat?
ConnectorWhat patterns connect to past work?
PlannerWhat's the next move?

--preset design (design review)

LensQuestion
UserIs this easy to use? What's confusing?
MaintainerIs this easy to change later?
BreakerHow can this fail? Edge cases?
SimplifierWhat can be removed?
IntegratorHow does this fit with everything else?

--preset incident (post-incident)

LensQuestion
FirefighterWhat happened and how was it fixed?
DetectiveWhat was the root cause chain?
DefenderWhat guards existed? Why didn't they catch it?
ForecasterWhat similar things could happen next?
BuilderWhat 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

  1. No subagents — all lenses run in the main agent, sequentially
  2. Each lens gets a section with its emoji, name, and guiding question as header
  3. Evidence required — cite files, commits, commands, timestamps. No vague claims
  4. Lenses disagree — if the Architect says "clean" but the Auditor says "incomplete", show both. Don't harmonize
  5. Cross-lens summary at the end — what do multiple lenses converge on?
  6. 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
  7. 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
  8. 3-7 lenses — minimum 3, maximum 7. Default 5. Use --lenses N to adjust. Prose scales further than tables do; 9-10 lenses is workable in prose mode

When to Use

SituationWhy prism helps
End of sessionCatch what you missed before closing
After a migration/renameFind inconsistencies across the change
Design decisionSee tradeoffs from user/maintainer/breaker angles
Post-incidentStructured multi-angle without blame
"Did we miss anything?"The Auditor lens exists for this

Relationship to Other Skills

SkillPatternAgents
/oracle-prismMulti-perspective, same agent0 (inline transform)
/adversarial-analysisTry to disprove a claim5 parallel subagents
/rrrSession retrospective0 (or 5 in --deep)
/roundtableDiscussion between personas0 (inline)

Prism is the lightest multi-perspective tool — no agents, no spawning, no coordination overhead. Use it when you want angles, not adversaries.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.