Guardrails before you turn Breakflare trading agent live

Breakflare Plus includes a Trading agent that defaults to paper mode. That is intentional. Live fills should wait until you have written four numbers and proven a fifth control: score floor, size cap, stop, max hold, and a kill switch you have actually pressed in paper. Blank defaults with a plan to “tighten later” is how overnight regret happens. Core research on the board still does not place trades or publish price targets; guardrails are what keep the optional agent from inventing risk you never approved.

Key Takeaways

• Write score floor, size cap, stop, and max hold before any live discussion; kill switch is the fifth control.
• Copy the same Breakout Score™ and confidence floors you use by hand (score floor rules).
• Do not enable live with empty defaults. Paper until the fill log is explainable.
• Rehearse the kill switch on a schedule (kill switch drill).
• Plus is $399/mo or $3,990/yr with a 14-day trial; agent stays paper until you say otherwise.

Why do score floors come before hope?

A score floor is the minimum Breakout Score™ (and usually confidence) required before the agent may consider an entry. Without it, the agent will happily explore thin rows that you would never star on the 51-200 board.

Floors should match your manual promote rules from the daily watchlist workflow. If you require Climb confirmation and reject high hype / no climb, encode that. If quiet climbs stay research-only (quiet climb), do not give the agent a looser Climb policy.

Hope is not a floor. “See more action” is how zero floors get rationalized on live.

How should size caps survive a bad day?

Size caps answer: how much notional can one name, and the book, take when the thesis is wrong immediately? Mid-caps in ranks 51-200 can gap. Caps that only feel comfortable on green days are not caps.

Practical habits:

• Cap per position and aggregate exposure.
• Reduce size when confidence is merely adequate, not strong.
• Avoid stacking highly correlated pack names (for example several Solana mids) at full size on the same day; see Solana pack discipline.
• Keep paper size rules identical to the live rules you intend to use later.

Size is not an attention question. Hype score does not tell you how large to be. The scoring stack informs eligibility; sizing is your risk policy.

How do stops and max hold express different risks?

Stops are price exits. Max hold is time risk. Attention can stay elevated while your thesis window expires. Read the dedicated max hold and stop post for interaction with the daily UTC snapshot. Do not disable stops because the entry-day composite looked pretty.

What belongs in a kill switch drill?

A kill switch you have never pressed is decoration. In paper, rehearse until the motion is boring: where it lives, what it halts, what still logs, and how you resume. Schedule the drill. Hiding the control “so you do not fat-finger it” often means you will not find it under stress. Details and after-action review live in kill switch drill.

Treat the kill switch as part of onboarding, not as an emergency-only myth. The first live day is the wrong day to learn the UI path.

What is the pre-live checklist?

1. Floors written (Breakout Score™, confidence; optional Climb rules).
2. Size caps written (per name and book).
3. Stops written (and visible next to floors).
4. Max hold written.
5. Kill switch rehearsed in paper this week.
6. Paper fill log reviewed: every fill explainable from rules (paper mode).
7. Exchange keys, if ever connected, stay trade-only and minimal (Binance trade-only keys).
8. Live remains off until the above is boring.

Also keep Bitcoin dominance agent rules and low-confidence skips in the same binder. Regime and coverage are guardrails too.

How does this relate to the research board?

Guardrails do not replace the board habit. You still sort Breakout Score™, read hype, momentum, Climb, and confidence, and refuse single-feed promotes. The agent is optional automation of gates you already trust by hand. If the hand process is fuzzy, automate nothing.

Graduation sequencing is covered in paper then live. Credits and waitlist context sit in trading agent waitlist and credits.

Plus members who want live ranks plus paper agent can start a 14-day Plus trial. Soft CTA once is enough; the work is writing the numbers.

How do guardrails interact with attention regimes?

Attention regimes change how often floors clear, not whether controls exist. On quiet boards, floors may produce few paper fills; that is success, not failure. On pack-bid days, floors may clear for several correlated names; size caps and max hold then do more work than floors alone.

Write regime modifiers in advance where you can: tighter aggregate size when Bitcoin dominance spikes, higher confidence floors on thin coverage days, shorter max hold during violent meme rotations. The risk-off mid-cap attention and dominance posts give language for those modifiers. Do not invent modifiers mid-drawdown.

Guardrails are also communication tools. If you ever hand off the agent configuration, the five controls should be readable without tribal knowledge. Ambiguous “use judgment” clauses are how live keys get connected before paper logs are clean.

What does a clean paper graduation log contain?

Before live, you should be able to open a recent paper window and explain every fill with: floor pass, size, stop, max hold remaining, and why kill switch was not needed. If explanations require storytelling, floors are too soft or discretion is leaking into the agent.

Keep the log beside your attention research journal so research promotes and agent fills can be compared. Divergence is useful data: either the agent is looser than your desk, or your desk is starring names it would never want automated. Align them before keys.

What are the non-negotiable live blockers?

Do not enable live if any of these are true:

• Score floors are zero or undefined
• Size cap is “figure it out later”
• Stop rule is missing or routinely widened in paper
• Max hold is undefined
• Kill switch path is unrehearsed
• Paper fill log cannot be explained from rules
• Keys are broader than trade-only needs

These blockers are deliberately boring. Boring is the point. Attention research can stay exciting on the board while agent risk stays dull. If you need excitement from the agent, you are using the wrong surface. Use the board and your journal for narrative energy; use the agent for constrained execution experiments in paper first.

How do credits and paper mode affect guardrail discipline?

Paper mode removes exchange P&L pain, which can quietly erode guardrail seriousness. Treat paper fills as if size caps still represent real notional you care about. If you inflate paper size to “see what happens,” you will graduate with untested emotional responses.

Credits and agent usage metering (where applicable) are operational details; they are not permission to skip floors. Read trading agent credits for product metering, then return to the five controls. Metering answers “can the agent run?” Guardrails answer “should this fill exist?”

Frequently asked questions

What is a score floor?

The minimum composite (and usually confidence) required before the agent may enter. Match your manual promote rules.

Can I skip stops in paper?

No. Paper is where you prove the full control set. Skipping stops trains the wrong reflex.

Where is the kill switch?

Know the path in your agent UI and rehearse it. If you cannot find it quickly in paper, you are not ready for live.

Does a kill switch cancel paper activity too?

Treat rehearsal as a full halt test. Confirm in your environment what stops versus what continues to log, then document it.

Does Breakflare core research trade for me?

No. Core research is an attention tracker. Trading agent is optional, paper by default, and bound by your written rules.

Related reading

• Paper mode first: simulator until gates match.
• Score floor rules: copying desk floors into the agent.
• Kill switch drill: rehearsal cadence.
• Max hold and stops: time risk vs price risk.