القائمة
Q-Value Augmented Vector Retrieval for learned memory ranking. Tracks which memories are most useful over time and prioritizes them in retrieval.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
Self-improving agent architecture using ChromaDB for continuous learning, self-evaluation, and improvement storage. Agents maintain separate memory collections for learned patterns, performance metrics, and self-assessments without modifying their static .md configuration.
Guide for manually creating custom skills from scratch with templates and validation. Use when designing NEW skills with specific requirements. For auto-generating skills from existing git repos, use /generate-skills instead.
Display QAVR (Q-Value Augmented Vector Retrieval) system status including mode, interaction counts, top memories by Q-value, and configuration.
Enables continuous self-improvement through learning from failures, user corrections, and capability gaps. Integrates with QAVR for learned memory ranking.
Stress-test solutions using the STRIKE framework. Systematically attack proposals to find weaknesses before deployment.
Exhaustive solution space exploration. Use when you need ALL viable options (not just the best), solution space is unknown, or you can't afford to miss alternatives.
استنادا إلى تصنيف SOC المهني
| name | qavr-memory |
| description | Q-Value Augmented Vector Retrieval for learned memory ranking. Tracks which memories are most useful over time and prioritizes them in retrieval. |
| metadata | {"clawdbot":{"requires":{"bins":["python3"]}}} |
| user-invocable | false |
| disable-model-invocation | false |
Memory system that learns which information is most useful over time.
Standard vector retrieval returns results by semantic similarity alone. QAVR adds learned utility scoring based on actual usage outcomes:
Final Score = (1 - α) × Semantic Similarity + α × Q-Value
Where:
After each interaction:
# Positive outcome (task succeeded, user satisfied)
q_new = q_old + learning_rate * (reward - q_old)
reward = 1.0 for success, 0.0 for failure
# Temporal decay (unused memories fade)
q_decayed = q_old * decay_factor # e.g., 0.99 per day
{
"memories": {
"memory_id_1": {
"q_value": 0.75,
"access_count": 12,
"last_accessed": "2026-01-26",
"success_count": 9,
"failure_count": 3
}
},
"contexts": {
"debugging": {"interactions": 82, "mode": "cold"},
"coding": {"interactions": 156, "mode": "warm"}
},
"config": {
"learning_rate": 0.1,
"decay_factor": 0.99,
"warm_threshold": 100
}
}
def qavr_query(query_text, collection, n_results=5):
# Get semantic results
results = collection.query(
query_texts=[query_text],
n_results=n_results * 2 # Over-fetch for re-ranking
)
# Apply Q-value re-ranking if warm
if context_is_warm():
results = rerank_by_qvalue(results, alpha=0.3)
return results[:n_results]
QAVR learns from implicit signals:
| Signal | Interpretation | Reward |
|---|---|---|
| Memory used in successful task | Highly useful | +1.0 |
| Memory retrieved but not used | Somewhat relevant | +0.1 |
| Memory retrieved, task failed | Possibly misleading | -0.2 |
| Memory not retrieved for days | Decaying relevance | decay |
{
"qavr": {
"enabled": true,
"learning_rate": 0.1,
"decay_factor": 0.99,
"warm_threshold": 100,
"alpha_warm": 0.3,
"contexts": ["debugging", "coding", "research"]
}
}
Check QAVR status: