agentleFS
tusharagg1tusharagg1/aifolimizer/.claude/skills/top-trades-today/SKILL.md

top-trades-today

Skill2 starsChanged 5 months agoMITView on GitHub

Surface the top N actionable trade ideas for today across holdings + watchlist, each with entry, stop, target, R:R and conviction. Use when the user asks "top trades today", "what should I trade today", "best entries right now", "top stock picks", or wants a ranked, decision-ready trade shortlist. Composes get_trade_ideas + crowding + macro guards. One-shot.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 2 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Before ranking, load prior decisions so verdicts stay consistent across sessions:

    line 14
  2. Very specific:

    Call get_profile FIRST (accounts, capital - never hardcode).

    line 21
  3. Distinctive:

    Check macro snapshot for events TODAY (US Eastern): FOMC, CPI, NFP, PCE, GDP advance, VIX > 25.

    line 28
  4. Very specific:

    Use the idea's own current_price as entry reference.

    line 48

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Top Trades Today (ranked, decision-ready)

Goal

Short ranked list of names worth trading TODAY - held or watchlist - each with explicit entry / stop / target / R:R / conviction. Built on the same scoring engine as the dashboard signals (get_trade_ideas), then gated by crowding + macro catalysts. Designed to be pushed to Telegram each morning.

Stage 0 - Decision Memory (load FIRST)

Before ranking, load prior decisions so verdicts stay consistent across sessions:

  • mcp__aifolimizer__get_cross_ticker_lessons with max_lessons=3 - portfolio-level win/loss patterns
  • For any name you promote to the list, also load mcp__aifolimizer__get_ticker_decision_history (ticker=…, max_decisions=5) and mcp__aifolimizer__get_ticker_reflection (symbol=…, n=3).

Reconciliation rule: if a prior decision exists and your new read flips it, state explicitly WHY it changed (new data / catalyst / price move). Never silently contradict a logged decision - that drift is exactly what this prevents.

How to run

Call get_profile FIRST (accounts, capital - never hardcode). Then in parallel:

  1. mcp__aifolimizer__get_trade_ideas with top_n=5, include_watchlist=true, min_risk_reward=1.5
  2. mcp__aifolimizer__get_macro_snapshot (regime + scheduled catalysts)
  3. mcp__aifolimizer__get_positioning_signals with the symbols returned by step 1 (crowding guard)

Catalyst gate (FIRST - before ranking)

Check macro snapshot for events TODAY (US Eastern): FOMC, CPI, NFP, PCE, GDP advance, VIX > 25. If a major catalyst hits today, prefix the headline ⚠️ CATALYST: <event> @ <time> and add: "Halve intraday size or wait until after the print." Do not suppress the list - annotate it.

Crowding guard (per CLAUDE.md)

For each idea, cross-check get_positioning_signals:

  • crowding_label == consensus (score ≥ 70) AND action is BUY/ADD → downgrade conviction one notch, tag consensus-crowded - late-entry risk. Do NOT promote a crowded name to the top slot on a BUY.
  • crowding_label == contrarian (score ≤ 30) AND fundamentals/score support → tag contrarian setup (favourable).
  • TSX (.TO) names with null crowding → treat as neutral, don't over-weight the absence.

Output structure (≤ 350 words, Telegram-friendly)

Headline

Top N trades · [Regime label] · [catalyst flag or "no catalyst today"]

Ranked ideas (one block per idea, best first)

N. TICKER (held / watch) · ACTION · conviction
   entry ~$X · stop $Y · target $Z · R:R N.N
   why: <one line - the strongest reason>
   flag: <consensus-crowded / contrarian / earnings in Nd / none>

Use the idea's own current_price as entry reference; if entry_timing == acceptable, say "entry acceptable now"; the engine already excludes wait-for-pullback names.

Skipped / thin

One line: how many names scored, how many were actionable, and why the rest dropped (e.g. "12 scored, 4 actionable; rest HOLD/WATCH or R:R < 1.5").

After output - log decisions

For each ranked idea (BUY/ADD/SELL/TRIM), call mcp__aifolimizer__log_recommendation with action, conviction (HIGH/MED/LOW), target_pct + stop_pct (% from entry; entry captured live at call time), 1-line thesis (the "why" line), skill="top-trades-today". Include crowding flag in thesis. Feeds forward win-rate / track-record loop.

Rules

  • Direct. No hedging. If get_trade_ideas returns zero ideas, say so plainly: "No actionable setups today - all names HOLD/WATCH or below R:R floor." Do not invent trades.
  • Entry/stop/target come straight from get_trade_ideas - do NOT recompute or round differently.
  • Respect account context from get_profile for tax framing only; ranking is account-agnostic.
  • One-shot - do not chain into another skill unless the user asks. For a deep dive on one name, suggest /stock-analysis TICKER or /adversarial-research TICKER.

Gotchas

  • get_trade_ideas already filters wait_pullback and risk_reward < min_risk_reward - don't re-add names it dropped.
  • SELL/TRIM ideas can appear (held names) alongside BUY/ADD - label the action clearly; a SELL is a trade too.
  • Crypto + ETFs may surface with thin reasons - keep them but note "index/crypto - different dynamics."
  • risk_reward may be null for TRIM (target/RR intentionally omitted by the engine) - show stop only, omit R:R, don't treat null as an error.
  • Watchlist names have weight 0 - that's expected (not held), not a data gap.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.