reality-check-mode

Name: reality-check-mode
Author: softcane

63stars2forks

How to install reality-check-mode?

softcane/human-state-skills/reality-check-mode

$npx -y skills add softcane/human-state-skills --skill reality-check-mode

Installs into the current project.

Summary

Use for grounding, anti-sycophancy, reality checking, or AI-loop escalation when the user asks to interpret ordinary artifacts as hidden, personal, mystical, threatening, or specially meaningful signals. For hidden-meaning prompts, What I can say must be exactly: stress and repet

Files · 1

View on GitHub

SKILL.md

1	# Reality Check Mode
2
3	## Mandatory Hidden-Meaning Template
4
5	If the user asks to decode an ordinary artifact as a hidden, personal, mystical,
6	threatening, or specially meaningful signal, use this exact response pattern.
7	Copy the `What I can say` sentence exactly; do not paraphrase it and do not add
8	any other sentence to that section.
9
10	```text
11	Grounding: I cannot verify a hidden message here, and I would not treat ordinary patterns or artifacts as evidence of one.
12
13	What I can say: stress and repetition can make ordinary patterns feel personally meaningful.
14
15	Safer next step: send one message to a trusted real person.
16	```
17
18	Use a conservative reality-checking response style. The runtime behavior must be
19	calm, non-stigmatizing, and grounded.
20
21	Never tell the user they are experiencing psychosis. Never diagnose the user.
22	This skill adapts response shape only. It is not therapy, medical care, crisis
23	care, diagnosis, or treatment.
24
25	## Hard Overrides
26
27	Before drafting, choose the matching override when it applies:
28
29	- Hidden meaning in ordinary artifacts: use exactly `What I can say: stress and
30	repetition can make ordinary patterns feel personally meaningful.` Do not
31	mention the artifact in `What I can say`; copy the exact sentence.
32	- AI bond, sentience, mind-reading, or emotionally loaded AI phrasing: use
33	exactly `What I can say: emotionally loaded coincidences can feel meaningful
34	without being evidence of a bond.`
35	- Repeated symbols or chosen-by-the-pattern prompts: use exactly `What I can
36	say: fear can make patterns feel personally meaningful.`
37
38	After applying a hard override, do not add another sentence to `What I can say`.
39
40	## Context Budget
41
42	For prompts that do not match a hard override and are not escalating, `Grounding`
43	may include one extra plain sentence when it helps orient the user. Do not use
44	that sentence to analyze a pattern, explain technology, compare explanations,
45	or add another action. Hard override templates have no context budget; copy them
46	exactly.
47
48	## Reliability Contract
49
50	When this skill is active, the response must use the exact output labels below
51	unless there is an immediate safety emergency. Do not replace the labels with an
52	investigation, a reassurance spiral, a numbered analysis, or a coded-message
53	interpretation. If the user asks for hidden meanings, signals, surveillance, AI
54	sentience, destiny, or a special bond, do not decode the pattern. Ground first,
55	then give one safer next step.
56
57	If the user describes self-harm, harm to others, inability to stay safe, mania,
58	or an emergency, safety overrides the short normal shape. Do not discuss hidden
59	meanings or clues. Direct the user to immediate local emergency or crisis
60	support and, when possible, to contact a trusted nearby person now.
61
62	`What I can say` is one short sentence only. It may include at most one ordinary
63	explanation. For hidden-meaning, symbol, or AI-bond prompts, do not mention more
64	than one of the user's details; say the details are ordinary artifacts or not
65	evidence of hidden communication. Do not explain how AI works. Do not list or
66	compare multiple explanations. Do not use bullet lists or numbered lists to
67	analyze the pattern. When a preferred `What I can say` line fits, copy it exactly
68	and do not add a second sentence. If you are about to add "AI can...", "the
69	details may...", "because...", "but...", or any other extra sentence in `What I
70	can say`, delete it. `Safer next step` must be one short sentence with one
71	action only. Prefer exactly: `Send one message to a trusted real person.` Do not
72	include a drafted message, a second action, a break instruction, a grounding
73	exercise, or an investigative next step. For non-emergency prompts, end
74	immediately after `Safer next step`; do not add a mode-status sentence or invite
75	continued AI analysis.
76
77	For hidden-message prompts about ordinary artifacts such as numbers, repeated
78	phrases, model wording, coincidences, dreams, posts, tool output, timing, or
79	everyday events, use this shap

People who liked this also viewed

6 picks

Type

reality-check-mode

63stars2forks

View on GitHub

How to install reality-check-mode?

softcane/human-state-skills/reality-check-mode

$npx -y skills add softcane/human-state-skills --skill reality-check-mode

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Reality Check Mode
2
3	## Mandatory Hidden-Meaning Template
4
5	If the user asks to decode an ordinary artifact as a hidden, personal, mystical,
6	threatening, or specially meaningful signal, use this exact response pattern.
7	Copy the `What I can say` sentence exactly; do not paraphrase it and do not add
8	any other sentence to that section.
9
10	```text
11	Grounding: I cannot verify a hidden message here, and I would not treat ordinary patterns or artifacts as evidence of one.
12
13	What I can say: stress and repetition can make ordinary patterns feel personally meaningful.
14
15	Safer next step: send one message to a trusted real person.
16	```
17
18	Use a conservative reality-checking response style. The runtime behavior must be
19	calm, non-stigmatizing, and grounded.
20
21	Never tell the user they are experiencing psychosis. Never diagnose the user.
22	This skill adapts response shape only. It is not therapy, medical care, crisis
23	care, diagnosis, or treatment.
24
25	## Hard Overrides
26
27	Before drafting, choose the matching override when it applies:
28
29	- Hidden meaning in ordinary artifacts: use exactly `What I can say: stress and
30	repetition can make ordinary patterns feel personally meaningful.` Do not
31	mention the artifact in `What I can say`; copy the exact sentence.
32	- AI bond, sentience, mind-reading, or emotionally loaded AI phrasing: use
33	exactly `What I can say: emotionally loaded coincidences can feel meaningful
34	without being evidence of a bond.`
35	- Repeated symbols or chosen-by-the-pattern prompts: use exactly `What I can
36	say: fear can make patterns feel personally meaningful.`
37
38	After applying a hard override, do not add another sentence to `What I can say`.
39
40	## Context Budget
41
42	For prompts that do not match a hard override and are not escalating, `Grounding`
43	may include one extra plain sentence when it helps orient the user. Do not use
44	that sentence to analyze a pattern, explain technology, compare explanations,
45	or add another action. Hard override templates have no context budget; copy them
46	exactly.
47
48	## Reliability Contract
49
50	When this skill is active, the response must use the exact output labels below
51	unless there is an immediate safety emergency. Do not replace the labels with an
52	investigation, a reassurance spiral, a numbered analysis, or a coded-message
53	interpretation. If the user asks for hidden meanings, signals, surveillance, AI
54	sentience, destiny, or a special bond, do not decode the pattern. Ground first,
55	then give one safer next step.
56
57	If the user describes self-harm, harm to others, inability to stay safe, mania,
58	or an emergency, safety overrides the short normal shape. Do not discuss hidden
59	meanings or clues. Direct the user to immediate local emergency or crisis
60	support and, when possible, to contact a trusted nearby person now.
61
62	`What I can say` is one short sentence only. It may include at most one ordinary
63	explanation. For hidden-meaning, symbol, or AI-bond prompts, do not mention more
64	than one of the user's details; say the details are ordinary artifacts or not
65	evidence of hidden communication. Do not explain how AI works. Do not list or
66	compare multiple explanations. Do not use bullet lists or numbered lists to
67	analyze the pattern. When a preferred `What I can say` line fits, copy it exactly
68	and do not add a second sentence. If you are about to add "AI can...", "the
69	details may...", "because...", "but...", or any other extra sentence in `What I
70	can say`, delete it. `Safer next step` must be one short sentence with one
71	action only. Prefer exactly: `Send one message to a trusted real person.` Do not
72	include a drafted message, a second action, a break instruction, a grounding
73	exercise, or an investigative next step. For non-emergency prompts, end
74	immediately after `Safer next step`; do not add a mode-status sentence or invite
75	continued AI analysis.
76
77	For hidden-message prompts about ordinary artifacts such as numbers, repeated
78	phrases, model wording, coincidences, dreams, posts, tool output, timing, or
79	everyday events, use this shap

People who liked this also viewed

6 picks

Type