Skill

CLMM Position Manager Skill

> Addon to solana-dev-skill (core) and > sendai/skills (DeFi protocol calls). This skill does not > re-implement pool/swap integration — it adds the position-lifecycle + risk layer on top: > impermanent-loss accounting, range-exit probability, rebalance expected-value, and > delta-neutral hedging across Orca Whirlpools, Raydium CLMM, and Meteora DLMM.

When to load this skill

Load when the user is managing an existing concentrated-liquidity (CLMM/DLMM) position, not just opening one. Trigger phrases: "am I out of range", "impermanent loss", "should I rebalance", "is my LP actually profitable", "hedge my LP", "delta-neutral LP", "fees vs IL", "range exit".

For opening positions / raw swaps / pool discovery, defer to sendai (DeFi) and the protocol SDKs directly — do not duplicate that here.

Routing — load only the module you need (progressive disclosure)

| If the task is... | Load | | ------------------------------------------------------------- | ----------------------------- | | Read a position's current state (range, ticks/bins, holdings) | clmm-positions.md | | "Am I in range / when will I exit" + probability over horizon | range-monitoring.md | | "What's my real IL" / fees-vs-IL net P&L | impermanent-loss.md | | "Should I rebalance" (cost vs benefit gate) | rebalancing.md | | "Hedge my directional exposure" / delta-neutral LP | delta-neutral-hedge.md | | SDK package names, versions, docs (2026 stack) | resources.md |

Each module is self-contained and ~one screen. Never load all six at once.

Protocol coverage (unified abstraction)

This skill normalizes three CLMM designs behind one position model so the math is written once:

| Protocol | Liquidity model | TS package (2026) | Position primitive | | ---------------- | --------------------- | ------------------------------ | ------------------------- | | Orca Whirlpools | Tick-range (CLMM) | @orca-so/whirlpools v8 + -core v3 | tick lower/upper | | Raydium CLMM | Tick-range (CLMM) | @raydium-io/raydium-sdk-v2 | tick lower/upper | | Meteora DLMM | Discrete bins (DLMM) | @meteora-ag/dlmm v1.9.x | binId lower/upper |

> Normalization rule. Convert every venue to a common {priceLower, priceUpper, priceCurrent, > liquidityL, amountX, amountY, feesX, feesY} struct before doing any math. DLMM bins are mapped to > an equivalent price band so the same IL / range-exit / hedge formulas apply. See clmm-positions.md.

Hard rules

1. Read before you write. Never suggest a rebalance or hedge tx without first fetching live

position + pool state. Stale state produces wrong IL and wrong hedge size.

1. Never auto-execute. This skill computes and recommends. Signing is always user-gated.

It is a decision-support layer, not an autonomous trader.

1. Quote, don't churn. Every rebalance suggestion must pass the EV gate in rebalancing.md

(gas + crystallized IL + swap cost < expected incremental fee capture). No EV pass → no suggestion.

1. Math is venue-agnostic; SDK calls are venue-specific. Keep them separated.
2. Two-strike rule. If a live read fails twice on the same pool, STOP and surface the raw error.

Core agent & commands

• Agent: position-manager (opus) — full lifecycle reasoning.
• Commands: /check-positions, /rebalance-plan, /hedge-plan.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: Makabeez
• Source: Makabeez/position-manager-skill
• License: MIT

Install and usage instructions live in the source repository linked above.