reflexion

dug-21 的头像作者 dug-21

记录通过get-pattern检索的模式是否有用的反馈。对训练推荐系统至关重要。

📖 中文导读:记录通过get-pattern检索的模式是否有用的反馈。对训练推荐系统至关重要。 以下正文为该技能的原始 SKILL.md 内容(英文),上方安装命令可直接使用。

name: "reflexion" description: "Record feedback on pattern effectiveness. Stores episodes that train the recommendation system and enable pattern discovery via learner."

Reflexion - Evaluate Pattern Effectiveness

What This Skill Does

Records feedback on patterns and approaches used during work. This feedback:

  1. Trains the recommendation system for better pattern suggestions
  2. Provides data for learner skill to auto-discover new patterns
  3. Tracks what works and what doesn't over time

Use this AFTER completing work to record what helped and what didn't.


Quick Reference

# Store feedback
agentdb reflexion store "session-id" "task description" reward success "critique"

# Retrieve similar experiences
agentdb reflexion retrieve "search query" --k 5 --only-successes

# Get critique summary
agentdb reflexion critique-summary "topic" true

Primary Method: Store Feedback

agentdb reflexion store \
  "dp-004" \
  "Used domain-adapter pattern for new HTTP source" \
  1.0 \
  true \
  "Pattern was complete - followed Source trait steps exactly, tests passed first try"

Parameters (positional)

Position Parameter Description
1 session-id Feature ID (e.g., dp-004, air-011)
2 task Description of what you did
3 reward Success score 0-1
4 success true or false
5 critique Specific feedback (required)
6 input Optional: task input
7 output Optional: task output
8 latency-ms Optional: execution time
9 tokens Optional: tokens used

Examples

Pattern Worked Well

agentdb reflexion store \
  "dp-004" \
  "Used domain-adapter pattern for new HTTP source" \
  1.0 \
  true \
  "Pattern was complete - followed Source trait steps exactly, tests passed first try"

Pattern Partially Worked

agentdb reflexion store \
  "dp-004" \
  "Used add-stream pattern but needed adjustment" \
  0.6 \
  true \
  "Pattern missing retention field requirement added in v2.0 - should update pattern via save-pattern"

Pattern Failed

agentdb reflexion store \
  "dp-004" \
  "Pattern mqtt-routing failed for multi-topic subscription" \
  0.2 \
  false \
  "Pattern assumes single topic per source - needs update for multi-topic. Used workaround with topic array."

No Pattern Found

agentdb reflexion store \
  "dp-004" \
  "Implemented TimescaleDB continuous aggregate - no existing pattern" \
  0.85 \
  true \
  "No pattern existed. Created new approach using hypertable + continuous_aggregate. Should save as new pattern."

Retrieve Similar Experiences

# Find successful similar work
agentdb reflexion retrieve "HTTP source implementation" \
  --k 5 \
  --only-successes \
  --min-reward 0.7

# Find failures to learn from
agentdb reflexion retrieve "MQTT configuration" \
  --k 5 \
  --only-failures

# Get synthesized summary
agentdb reflexion retrieve "parquet storage" \
  --k 10 \
  --synthesize-context

Retrieve Parameters

Parameter Description
--k Number of results
--only-successes Only successful episodes
--only-failures Only failed episodes
--min-reward Minimum reward threshold
--synthesize-context Generate summary

Get Critique Summary

Aggregate lessons from critiques:

# Get critique summary for failures
agentdb reflexion critique-summary "mqtt" true

# Get all critiques for a topic
agentdb reflexion critique-summary "architecture" false

Reward Scale

Score Meaning When to Use
1.0 Perfect Pattern/approach worked exactly as expected
0.8 Good Minor adjustments needed
0.6 Partial Significant modifications required
0.4 Weak Marginally helpful, major workarounds
0.2 Failed Didn't work, caused issues
0.0 Harmful Actively wrong, wasted time

Session ID Convention

Use consistent session IDs for aggregation:

Session ID Use For
{feature-id} Feature work (e.g., dp-004, air-011)
{feature-id}-{phase} Specific phase (e.g., dp-004-spec)
maintenance Bug fixes, refactoring
exploration Research, spikes, experiments

Critique Best Practices

Good critiques (specific, actionable):

"Pattern was complete - followed steps exactly and deployment succeeded"
"Missing retention field that's now required in v2.0 schema"
"TimescaleDB connection pattern assumed localhost but we use Docker networking"
"Architecture pattern outdated - ADR-005 superseded the approach"

Poor critiques (vague, unusable):

"It worked"              # Too vague
"Failed"                 # No actionable info
"Good pattern"           # Doesn't explain what made it good

The Pattern Workflow

1. BEFORE work:  get-pattern  → Search for relevant patterns
2. DURING work:  Apply patterns, note gaps and discoveries
3. AFTER work:   reflexion    → Record what helped (THIS SKILL)
                 save-pattern → Store NEW discoveries (if any)
                 learner      → Auto-discover patterns from episodes (periodic)

After Recording Feedback

If your critique identifies a pattern that needs updating:

# 1. Record the feedback (this skill)
agentdb reflexion store \
  "dp-004" \
  "Used add-stream pattern" \
  0.6 \
  true \
  "Pattern missing required retention field"

# 2. Update the pattern (save-pattern skill)
agentdb skill create \
  "add-stream-v2" \
  "Add Data Stream (v2.0): Now requires retention field. Steps: 1) Create config.yaml, 2) Add retention field (required), 3) Run sync..." \
  "tags: streams, config, updated"

  • get-pattern - Search patterns BEFORE work
  • save-pattern - Store NEW patterns after discovering reusable approaches
  • learner - Auto-discover patterns from reflexion episodes

What NOT to Use This For

Don't Record Use Instead
New patterns you discovered save-pattern
Swarm coordination state claude-flow memory tools
Transient task/agent memory claude-flow memory tools
Architecture decisions save-pattern

Reflexion is for FEEDBACK on work done, not storing new knowledge.