| name | prcp - Perception Skills |
| description | Systematic procedure for improving what you notice. Slows down observation, separates what's present from what's interpreted, and surfaces what you normally filter out. |
| output | {"format":"prose"} |
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
