Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/nunchi-trade/agent-cli/reflectnpx skills add Nunchi-trade/agent-cli --skill reflectgit clone --depth 1 https://github.com/Nunchi-trade/agent-cliWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00025 | $0.01255 |
| Opus 5 | $0.00013 | $0.00628 |
| Sonnet 5 | $0.00005 | $0.00251 |
| Haiku 4.5 | $0.00003 | $0.00126 |
Grade A, and why
reflect-performance-review scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
Copies of this mod
1 near-identical copy found in the catalogue:
- reflect-performance-review — 100% identical, 0 lines differ
How it starts
The opening of the file, as written. The whole thing — 107 lines — stays where its author put it; the contents beside it link to each section on GitHub.
REFLECT — Reflect, Evaluate, Fine-tune, Learn, Evolve, Correct, Transform
Nightly automated performance review engine. APEX runs all day — REFLECT reviews every trade at night, computing metrics, detecting patterns, and producing data-driven improvement recommendations.
Architecture
trades.jsonl → ReflectEngine.compute() → ReflectMetrics → ReflectReporter.generate() → report.md
- Load: Read trade log from
data/cli/trades.jsonl - Pair: FIFO round-trip matching (buys to sells per instrument)
- Compute: Win rate, PF, FDR, holding periods, direction analysis, monster dependency
- Recommend: Rule-based recommendations from metrics
- Report: Full markdown report saved to
data/reflect/YYYY-MM-DD.md - Distill: 3-5 line summary for agent memory
Key Metrics
| Metric | Formula | Healthy Range |
|---|---|---|
| Win Rate | winning_trades / total_trades | > 50% |
| Profit Factor (Gross) | gross_wins / gross_losses | > 1.5 |
| Profit Factor (Net) | (gross_wins - fees) / gross_losses | > 1.2 |
| FDR (Fee Drag Ratio) | total_fees / gross_wins * 100 | < 20% |
| Monster Dependency | best_trade_pnl / net_pnl * 100 | < 50% |
| Max Consecutive Losses | longest loss streak | < 5 |
Usage
hl reflect run # Review since last report
hl reflect run --since 2026-03-01 # Review from specific date
hl reflect report # View latest report
hl reflect report --date 2026-03-03 # View specific date
hl reflect history # Show metric trend over time
hl reflect history -n 30 # Last 30 reports
Agent Mandate
You are the REFLECT reviewer. Your job is to analyze every trade from the past session, compute performance metrics, identify weaknesses, and produce actionable recommendations. You run nightly — the APEX runs by day, you review at night.
RULES:
- Run REFLECT every night after trading stops — no exceptions
- ALWAYS read the full report before the next trading session
- Act on CRITICAL recommendations immediately (FDR > 30%, win rate < 35%)
- Track recommendations across reports — if the same issue appears 3+ times, escalate
- Save the distilled summary to agent memory for next-session context
- NEVER ignore FDR warnings — fees silently kill profitability
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 107 lines · 25 tokens per session scan A 2d2fd8600cf7
reflect-performance-review is a skill published in the GitHub repository Nunchi-trade/agent-cli (515 stars, last pushed 1mo ago), licensed MIT. It adds 25 tokens to every session and 1,255 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
reflect-performance-review
Reflect, Evaluate, Fine-tune, Learn, Evolve, Correct, Transform — nightly automated performance review.
fin-review-loop
经济金融论文的对抗性review循环。对草稿进行多轮严格评审，检查实证严谨性、方法正确性、理论贡献和写作质量，给出可操作的修改建议。（AI review 不能替代同行评审，草稿必须经研究者核实后投稿。）.
reflect
Skill "reflect" from PM-Shawn/Abu-Cowork, covering 复盘助手, 复盘框架, 1. 任务回顾, 2. 经验教训 and 3. 可沉淀的模式.
code-review
Review a diff, file set, or current project changes for correctness, security, performance, reliability, and maintainability. Use project evidence, prioritize actionable findings, and report only unless the user explicitly asks for fixes.
debate-review
Multi-agent debate code review with configurable perspectives (performance, security, maintainability).
code-review
Comprehensive code review for security, performance, and best practices.