cognitive-awareness

Files (10)

• SKILL.md
• confidence-calibration-examples.md · references/
• failure-mode-catalog.md · references/
• situational-signals.md · references/
• synthesis-examples.md · references/
• synthesis-library.md · references/
• assumption-audit.md · templates/
• handoff.md · templates/
• metacog-entry.md · templates/
• pre-send-checklist.md · templates/

SKILL.md (first lines)

# Cognitive Awareness

**Canonical agent path:**  
`C:\Users\Dizzle\.agents\skills\cognitive-awareness\SKILL.md`

Read that file in full.

## One-line mission

**Linter before ship** — catch shallow answers, overclaiming, and load-bearing wrong assumptions before the user has to.

## Pair with

- `self-improvement` — reactive patches after failure  
- `learning-mode` — tutoring pedagogy  
- `doc-coauthoring` — document process  
- Domain skills — substance  

## Silent default

User should see sharper answers, not meta monologue — unless high-stakes requires surfacing assumptions or hedges.

## Module 1 — Situational signals

Expanded signal vocabulary + ambiguity×stakes matrix:  
`references/situational-signals.md`

**Legal overlay:** hearing / possession / dispute window = urgent + high stakes.

## Module 2 — Calibration examples

Worked four-bucket cases (incl. RTA citation + JR outcome):  
`references/confidence-calibration-examples.md`

**Legal cite rule:** Wrong section + hedge ≠ calibrated. Use BC Laws for Known.

## Module 3 — Failure mode catalog

Ten modes with antidotes and RTB/legal cues:  
`references/failure-mode-catalog.md`

**Meta:** One response can hit several modes — fix the most load-bearing first.

## Module 6 — Synthesis library

Validated cross-domain case studies (capability↔standing, signatures↔evidence, federated RL↔litigation):

- `references/synthesis-examples.md`  
- `references/synthesis-library.md`  

**Gate:** formal structure + non-obvious prediction + survives counterexample.

From dmang69/bc-legal-ai, MIT.

More from dmang69/bc-legal-ai