cognitive-mode

Name: cognitive-mode
Author: DNYoussef

32stars6forks

How to install cognitive-mode?

dnyoussef/context-cascade/cognitive-mode

$npx -y skills add DNYoussef/context-cascade --skill cognitive-mode

Installs into the current project.

Summary

Comprehensive cognitive mode management skill for the VERILINGUA x VERIX x DSPy x GlobalMOO integration. Enables automatic mode selection, frame configuration, VERIX epistemic notation, and GlobalMOO optimization. Use this skill when configuring AI behavior for specific task type

Files · 1

View on GitHub

SKILL.md

1	## SKILL-SPECIFIC GUIDANCE
2
3	### When to Use This Skill
4	- Configuring cognitive modes for different task types (research, coding, security)
5	- Optimizing prompt engineering through GlobalMOO optimization
6	- Ensuring epistemic consistency with VERIX notation
7	- Selecting appropriate VERILINGUA cognitive frames
8	- Running multi-objective optimization on prompt configurations
9	- Meta-loop recursive improvement on foundry skills
10
11	### When NOT to Use This Skill
12	- Simple one-off tasks that don't require specialized configuration
13	- Tasks where speed is paramount and optimization overhead is unacceptable
14	- When default balanced mode is sufficient
15	- Non-technical conversational interactions
16
17	### Success Criteria
18	- Appropriate mode selected for task domain and complexity
19	- VERIX epistemic notation applied correctly to claims
20	- Cognitive frames activated match task requirements
21	- GlobalMOO optimization produces Pareto-optimal configurations
22	- Cross-model consistency maintained (Claude + Gemini + Codex)
23
24	### Edge Cases & Limitations
25	- Mode selection may be uncertain for novel task types
26	- VERIX parsing may miss nuanced epistemic markers
27	- GlobalMOO optimization requires multiple iterations
28	- Some cognitive frames may conflict (e.g., minimal vs comprehensive)
29
30	### Critical Guardrails
31	- NEVER skip VERIX grounding for high-confidence claims
32	- ALWAYS validate mode selection for security-sensitive tasks
33	- NEVER use minimal mode for compliance/audit tasks
34	- ALWAYS include confidence levels for factual assertions
35	- NEVER modify holdout corpus (never_optimize: true)
36
37	### Evidence-Based Validation
38	- Mode selection validated against expected_metrics
39	- VERIX consistency checked via ConsistencyChecker
40	- Optimization results compared to Pareto frontier
41	- Cross-model evaluation via 3-model council
42
43	---
44
45	# Cognitive Mode Management
46
47	A comprehensive skill for managing cognitive modes, VERILINGUA frames, VERIX epistemic notation, and GlobalMOO optimization in the Context Cascade plugin system.
48
49	## Overview
50
51	This skill integrates four major systems:
52
53	1. VERILINGUA: 7 cognitive frames from diverse linguistic traditions
54	2. VERIX: Epistemic notation for claim validation
55	3. DSPy: Two-layer prompt optimization (Level 2 caching + Level 1 evolution)
56	4. GlobalMOO: Multi-objective optimization for configuration tuning
57
58	## Slash Commands
59
60	### /mode - Mode Selection
61
62	Select and configure cognitive modes:
63
64	```bash
65	/mode # List available modes
66	/mode <name> # Select mode by name
67	/mode auto "<task>" # Auto-select based on task
68	/mode info <name> # Show mode details
69	/mode recommend "<task>" # Get top-3 recommendations
70	```
71
72	Available Modes:
73	- `strict`: Maximum epistemic consistency (research, legal, medical)
74	- `balanced`: Good tradeoff for general use (default)
75	- `efficient`: Optimized for token efficiency (high-volume APIs)
76	- `robust`: Edge case handling (security, adversarial)
77	- `minimal`: Lightweight with no frames (simple Q&A)
78
79	### /eval - Evaluation
80
81	Evaluate tasks against cognitive architecture metrics:
82
83	```bash
84	/eval "<task>" "<response>" # Evaluate response
85	/eval --corpus <path> # Run corpus evaluation
86	/eval --metrics # Show metric definitions
87	/eval --graders # List available graders
88	```
89
90	Metrics:
91	- `task_accuracy`: Correctness (0.0 - 1.0)
92	- `token_efficiency`: Tokens vs target (0.0 - 1.0)
93	- `edge_robustness`: Adversarial handling (0.0 - 1.0)
94	- `epistemic_consistency`: VERIX compliance (0.0 - 1.0)
95
96	### /optimize - GlobalMOO Optimization
97
98	Run multi-objective optimization:
99
100	```bash
101	/optimize # Show optimization status
102	/optimize start # Start optimization run
103	/optimize suggest # Get configuration suggestions
104	/optimize report # Get optimization report
105	/optimize phase <A\|B\|C> # Run specific cascade phase
106	```
107
108	Three-MOO Cascade:
109	- Phase A: Framework structure optimization
110	- Phase B: Edge case discovery
111	- Phase C: Production frontier refinement
112
113	### /pareto - Pareto Frontier
114
115	Explore the Pareto frontier:
116
117	```bash
118	/pareto # Display frontier
119	/pareto filter <metric> # Filter by metric
120	/pareto export # Export as JSON
121	/pareto distill # Distill into named modes
122	/par

People who liked this also viewed

6 picks

Type

cognitive-mode

32stars6forks

View on GitHub

How to install cognitive-mode?

dnyoussef/context-cascade/cognitive-mode

$npx -y skills add DNYoussef/context-cascade --skill cognitive-mode

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	## SKILL-SPECIFIC GUIDANCE
2
3	### When to Use This Skill
4	- Configuring cognitive modes for different task types (research, coding, security)
5	- Optimizing prompt engineering through GlobalMOO optimization
6	- Ensuring epistemic consistency with VERIX notation
7	- Selecting appropriate VERILINGUA cognitive frames
8	- Running multi-objective optimization on prompt configurations
9	- Meta-loop recursive improvement on foundry skills
10
11	### When NOT to Use This Skill
12	- Simple one-off tasks that don't require specialized configuration
13	- Tasks where speed is paramount and optimization overhead is unacceptable
14	- When default balanced mode is sufficient
15	- Non-technical conversational interactions
16
17	### Success Criteria
18	- Appropriate mode selected for task domain and complexity
19	- VERIX epistemic notation applied correctly to claims
20	- Cognitive frames activated match task requirements
21	- GlobalMOO optimization produces Pareto-optimal configurations
22	- Cross-model consistency maintained (Claude + Gemini + Codex)
23
24	### Edge Cases & Limitations
25	- Mode selection may be uncertain for novel task types
26	- VERIX parsing may miss nuanced epistemic markers
27	- GlobalMOO optimization requires multiple iterations
28	- Some cognitive frames may conflict (e.g., minimal vs comprehensive)
29
30	### Critical Guardrails
31	- NEVER skip VERIX grounding for high-confidence claims
32	- ALWAYS validate mode selection for security-sensitive tasks
33	- NEVER use minimal mode for compliance/audit tasks
34	- ALWAYS include confidence levels for factual assertions
35	- NEVER modify holdout corpus (never_optimize: true)
36
37	### Evidence-Based Validation
38	- Mode selection validated against expected_metrics
39	- VERIX consistency checked via ConsistencyChecker
40	- Optimization results compared to Pareto frontier
41	- Cross-model evaluation via 3-model council
42
43	---
44
45	# Cognitive Mode Management
46
47	A comprehensive skill for managing cognitive modes, VERILINGUA frames, VERIX epistemic notation, and GlobalMOO optimization in the Context Cascade plugin system.
48
49	## Overview
50
51	This skill integrates four major systems:
52
53	1. VERILINGUA: 7 cognitive frames from diverse linguistic traditions
54	2. VERIX: Epistemic notation for claim validation
55	3. DSPy: Two-layer prompt optimization (Level 2 caching + Level 1 evolution)
56	4. GlobalMOO: Multi-objective optimization for configuration tuning
57
58	## Slash Commands
59
60	### /mode - Mode Selection
61
62	Select and configure cognitive modes:
63
64	```bash
65	/mode # List available modes
66	/mode <name> # Select mode by name
67	/mode auto "<task>" # Auto-select based on task
68	/mode info <name> # Show mode details
69	/mode recommend "<task>" # Get top-3 recommendations
70	```
71
72	Available Modes:
73	- `strict`: Maximum epistemic consistency (research, legal, medical)
74	- `balanced`: Good tradeoff for general use (default)
75	- `efficient`: Optimized for token efficiency (high-volume APIs)
76	- `robust`: Edge case handling (security, adversarial)
77	- `minimal`: Lightweight with no frames (simple Q&A)
78
79	### /eval - Evaluation
80
81	Evaluate tasks against cognitive architecture metrics:
82
83	```bash
84	/eval "<task>" "<response>" # Evaluate response
85	/eval --corpus <path> # Run corpus evaluation
86	/eval --metrics # Show metric definitions
87	/eval --graders # List available graders
88	```
89
90	Metrics:
91	- `task_accuracy`: Correctness (0.0 - 1.0)
92	- `token_efficiency`: Tokens vs target (0.0 - 1.0)
93	- `edge_robustness`: Adversarial handling (0.0 - 1.0)
94	- `epistemic_consistency`: VERIX compliance (0.0 - 1.0)
95
96	### /optimize - GlobalMOO Optimization
97
98	Run multi-objective optimization:
99
100	```bash
101	/optimize # Show optimization status
102	/optimize start # Start optimization run
103	/optimize suggest # Get configuration suggestions
104	/optimize report # Get optimization report
105	/optimize phase <A\|B\|C> # Run specific cascade phase
106	```
107
108	Three-MOO Cascade:
109	- Phase A: Framework structure optimization
110	- Phase B: Edge case discovery
111	- Phase C: Production frontier refinement
112
113	### /pareto - Pareto Frontier
114
115	Explore the Pareto frontier:
116
117	```bash
118	/pareto # Display frontier
119	/pareto filter <metric> # Filter by metric
120	/pareto export # Export as JSON
121	/pareto distill # Distill into named modes
122	/par

People who liked this also viewed

6 picks

Type