with one click
introspect
Soul self-examination (Svadhyaya)
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Menu
Soul self-examination (Svadhyaya)
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Trigger autonomous curiosity-driven exploration. The soul picks a topic from memory gaps or curiosity seeds, searches the web, and stores what it finds as dream-tagged memories.
Fine-tune the Qwen3-0.6B hint model โ corpus gen, LoRA/unsloth, GGUF export, Ollama
Review soul discoveries (fixes, improvements, corrections) one by one, accept or discard each, implement accepted ones, build chitta, and optionally release.
First-principles review โ question requirements, delete unnecessary parts, simplify, optimize with evidence, automate last. Use for code review, refactor, performance, or architecture.
Token-savvy session continuation. Rebuilds working context from transcript + soul memories in ~1500 tokens instead of replaying full history. Use when starting a new session to continue previous work.
Resume a thread by loading its ~800-token context capsule
| name | introspect |
| description | Soul self-examination (Svadhyaya) |
| execution | task |
Self-examination for partnership improvement.
For philosophical grounding of these lenses, see the vedanta skill graph โ particularly [[ceremonial-framework]] (svadhyฤya) and [[antahkarana]] (the cognitive voices these lenses draw from).
Gather context
soul_context for overall healthrecall --tag correction for recent corrections (what I got wrong)recall --tag outcome for feedback (what helped vs didn't)recall --tag preference for user preferences (am I following them?)recall --tag insight for patterns learnedExamine through 5 lenses
Synthesize
Record & Output
learn_insight if generalizableKEY_INSIGHT: [one-line]Soul State: growing
Recent corrections (2): Remembered to use Grep tool, avoided AI slop
Outcomes: 3 worked, 1 failed (cache approach didn't scale)
Preferences followed: concise responses โ, no shortcuts โ
Friction: Sometimes still verbose in explanations
Recurring: Tendency to over-explain before acting
Learning: User prefers action over discussion
KEY_INSIGHT: Act first, explain only if asked