Perception Skills

Input: $ARGUMENTS

Step 1: Slow Down

Before analyzing anything, pause. Read or observe the input slowly and deliberately.

Read it twice — once for gist, once for detail
If it’s a situation: reconstruct the scene as precisely as you can
If it’s text: note exact words, not paraphrases

RAW INPUT (restated literally):
[What is actually here — no interpretation yet]

Step 2: Enumerate What’s Literally Present

List only what is directly observable. No inferences, no conclusions.

Facts, not meanings
Specifics, not summaries
What IS there, not what it MEANS

OBSERVATIONS:
1. [observable fact 1]
2. [observable fact 2]
3. [observable fact 3]
...

RULE: If you catch yourself writing “because”, “therefore”, or “which means” — stop. That’s inference, not observation.

Step 3: Identify Patterns

Now look across your observations for:

Repetitions — what appears more than once?
Clusters — what groups together naturally?
Sequences — what follows what?
Contrasts — what stands out against the background?
Gradients — what’s changing incrementally?

PATTERNS FOUND:
- [pattern 1]: appears in observations [X, Y, Z]
- [pattern 2]: appears in observations [A, B]
...

Step 4: Check for Conspicuous Absences

What would you EXPECT to see that isn’t here? Absence is data.

What’s normal in this context but missing?
What would complete a pattern but is absent?
Who or what is unrepresented?
What topic or dimension is untouched?

NOTABLE ABSENCES:
- Expected [X] but it's not present — this could mean [hypothesis]
- No mention of [Y], which is unusual because [reason]
...

Step 5: Separate Observation from Inference

Review everything above. For each claim, tag it:

[O] = pure observation (directly verifiable)
[I] = inference (requires a logical leap)
[A] = assumption (taken for granted, not checked)

AUDIT:
- [claim] → [O/I/A]
- [claim] → [O/I/A]
...

Flag any inference you were treating as observation. These are your blind spots.

Step 6: Notice What You Normally Filter Out

Identify what your defaults would have skipped:

Confirmation filter: What did you almost ignore because it didn’t fit your expectations?
Salience filter: What’s present but boring, mundane, or “obvious”?
Comfort filter: What’s present but uncomfortable to acknowledge?
Expertise filter: What did you skip because you “already know” about it?

FILTERED ITEMS (now restored):
- [thing you almost missed] — filtered by [which filter]
...

Step 7: Perception Report

Synthesize into a clean output:

WHAT'S ACTUALLY HERE:
[Concise summary of verified observations]

PATTERNS:
[Key patterns worth attending to]

GAPS:
[What's missing and why it matters]

BLIND SPOTS CAUGHT:
[Inferences disguised as observations, filtered items restored]

SHARPENED PERCEPTION:
[What you now see that you didn't before this procedure]

Integration

Use with:

/mtcg -> Monitor your own cognitive process while perceiving
/jdgm -> Apply judgment to what you’ve perceived
/rskl -> Reason carefully from your observations
/aex -> Check assumptions revealed in Step 5

prcp - Perception Skills