solana-security-standard

solana-security-standard

Skill by ara.so — Security Skills collection.

The Solana Security Standard provides 37 SOL-0XX rules distilled from $514M+ of real Solana exploits. It catches caller-controlled clock values, cross-market state asymmetry, missing Anchor constraints, wrapper/engine drift, and 33 other bug classes — firing as you code in AI tools, editors, CLI, and CI.

SOL-001 alone covers 2 confirmed bounty wins (caller-controlled now_slot in percolator). The standard works across Claude Code, Cursor, Windsurf, Copilot, Cline, Aider, the VS Code extension, the CLI, Semgrep, GitHub Actions, and any MCP client.

Installation

Claude Code (quickest)

mkdir -p .claude && \
  curl -sL https://raw.githubusercontent.com/Copenhagen0x/solana-security-standard/main/plugin-guidance.md \
       -o .claude/claude-security-guidance.md && \
  curl -sL https://raw.githubusercontent.com/Copenhagen0x/solana-security-standard/main/security-patterns.yaml \
       -o .claude/security-patterns.yaml

Then install the security-guidance plugin:

/plugin install security-guidance@claude-plugins-official
/reload-plugins

Or install the full MCP-powered plugin:

/plugin marketplace add Copenhagen0x/solana-security-standard
/plugin install solana-security-standard@solana-security-standard

CLI (any machine)

npx @jelleo/solana-security-standard scan ./programs

No install needed — runs via npx. Zero dependencies.

CI — GitHub Action

# .github/workflows/solana-security.yml
name: Solana Security Standard
on: [pull_request]
permissions:
  contents: read
  security-events: write
jobs:
  scan:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: Copenhagen0x/solana-security-standard@v1
        with:
          paths: ./programs
          fail-on-findings: true
          upload-sarif: true

VS Code / Cursor / Windsurf

Install from Open VSX:

• Cursor / Windsurf / VSCodium: Search "Solana Security Standard" in Extensions
• VS Code: Sideload .vsix from extensions/vscode/

Shows inline warnings as you type in Rust and TypeScript/JS files.

MCP (any MCP client)

Add to your MCP settings (.cline/mcp.json, .cursor/mcp.json, etc.):

{
  "mcpServers": {
    "solana-security-standard": {
      "command": "npx",
      "args": ["-y", "@jelleo/solana-security-mcp"]
    }
  }
}

Exposes scan_solana_code tool + list_solana_security_rules resource.

AI Agents (Codex, Copilot, Cursor, Windsurf, Cline, Aider)

Copy the rules file for your tool:

Tool	Copy into repo
Codex / AGENTS.md	integrations/codex/AGENTS.md
GitHub Copilot	integrations/copilot/.github/copilot-instructions.md
Cursor	integrations/cursor/.cursor/
Windsurf	integrations/windsurf/.windsurf/
Cline	integrations/cline/.clinerules
Aider	integrations/aider/

# Example for Cursor:
curl -sL https://raw.githubusercontent.com/Copenhagen0x/solana-security-standard/main/integrations/cursor/.cursor/rules -o .cursor/rules

Semgrep

semgrep --config https://raw.githubusercontent.com/Copenhagen0x/solana-security-standard/main/semgrep/solana-security-standard.yaml ./programs

CLI Commands

Scan programs

# Scan a directory
npx @jelleo/solana-security-standard scan ./programs

# Scan specific file
npx @jelleo/solana-security-standard scan ./programs/src/lib.rs

# JSON output (for CI)
npx @jelleo/solana-security-standard scan ./programs --format json

# SARIF output (GitHub Code Scanning)
npx @jelleo/solana-security-standard scan ./programs --format sarif > results.sarif

# Filter by tier (drop low-severity hygiene findings)
npx @jelleo/solana-security-standard scan ./programs --min-tier high

# Baseline mode (gate only on NEW findings)
npx @jelleo/solana-security-standard scan ./programs --baseline .sss-baseline.json

Generate baseline

# Create baseline of current findings
npx @jelleo/solana-security-standard scan ./programs --format json > .sss-baseline.json

# Future scans compare against it
npx @jelleo/solana-security-standard scan ./programs --baseline .sss-baseline.json

CLI exits non-zero on findings (gates CI).

Key Rules (37 total)

Critical On-Chain Rules

Rule	Bug Class	Example
SOL-001	Unauthenticated now_slot / clock spoofing	2 confirmed bounty wins
SOL-002	Cross-market state asymmetry	Counter inflation
SOL-003	Wrapper re-implements engine logic	Drift from canonical
SOL-004	Missing signer check	Unauthorized access
SOL-005	Missing owner check	Wrong program control
SOL-006	Account data type confusion	Misinterpreted structs
SOL-007	Uninitialized account	Reading zeroed data
SOL-008	Arithmetic overflow/underflow	Balance manipulation
SOL-009	PDA derivation mismatch	Wrong seeds → wrong account
SOL-010	Missing rent exemption	Account closure drain

Integrator / Client-Side Rules

Rule	Bug Class
SOL-029	Client-side arithmetic overflow (TypeScript/JS)
SOL-030	Missing transaction simulation
SOL-031	Hardcoded program IDs in client

Full list: https://github.com/Copenhagen0x/solana-security-standard/blob/main/claude-security-guidance.md

Real Code Examples

SOL-001: Unauthenticated now_slot (2 bounty wins)

Vulnerable (percolator-prog ACTIVATE):

pub fn activate(ctx: Context<Activate>, now_slot: u64) -> Result<()> {
    let state = &mut ctx.accounts.state;
    let clock = Clock::get()?;
    
    // Caller controls now_slot — can bypass time checks
    if now_slot < state.cooldown_end {
        return Err(ErrorCode::CooldownActive.into());
    }
    
    state.active = true;
    Ok(())
}

Fixed:

pub fn activate(ctx: Context<Activate>) -> Result<()> {
    let state = &mut ctx.accounts.state;
    let clock = Clock::get()?;
    
    // Use authenticated clock.slot
    if clock.slot < state.cooldown_end {
        return Err(ErrorCode::CooldownActive.into());
    }
    
    state.active = true;
    Ok(())
}

SOL-004: Missing signer check

Vulnerable:

#[derive(Accounts)]
pub struct Withdraw<'info> {
    #[account(mut)]
    pub authority: AccountInfo<'info>, // Not checked as signer!
    #[account(mut)]
    pub vault: Account<'info, Vault>,
}

pub fn withdraw(ctx: Context<Withdraw>, amount: u64) -> Result<()> {
    // Anyone can pass any authority account
    transfer_from_vault(&ctx.accounts.vault, amount)?;
    Ok(())
}

Fixed:

#[derive(Accounts)]
pub struct Withdraw<'info> {
    #[account(mut)]
    pub authority: Signer<'info>, // Enforces signature
    #[account(
        mut,
        has_one = authority // Links vault to this authority
    )]
    pub vault: Account<'info, Vault>,
}

SOL-009: PDA derivation mismatch

Vulnerable:

// Instruction handler uses wrong seeds
let (vault_pda, _bump) = Pubkey::find_program_address(
    &[b"vault", user.key().as_ref()], // Missing market_id seed!
    program_id
);

if vault_pda != *vault.key {
    return Err(ErrorCode::InvalidVault.into());
}

Fixed:

// Match the seeds used in account derivation
let (vault_pda, _bump) = Pubkey::find_program_address(
    &[b"vault", market_id.as_ref(), user.key().as_ref()],
    program_id
);

if vault_pda != *vault.key {
    return Err(ErrorCode::InvalidVault.into());
}

SOL-029: Client-side arithmetic overflow (TypeScript)

Vulnerable:

// JavaScript numbers lose precision beyond 2^53
const depositAmount = userBalance + newDeposit;
const tx = await program.methods
  .deposit(new BN(depositAmount)) // Precision loss!
  .accounts({ vault, user })
  .rpc();

Fixed:

import { BN } from "@coral-xyz/anchor";

// Use BN for all token math
const depositAmount = new BN(userBalance).add(new BN(newDeposit));
const tx = await program.methods
  .deposit(depositAmount)
  .accounts({ vault, user })
  .rpc();

Configuration

GitHub Action Options

- uses: Copenhagen0x/solana-security-standard@v1
  with:
    paths: ./programs              # Scan target (default: .)
    fail-on-findings: true         # Exit 1 on findings (default: true)
    upload-sarif: true             # GitHub Code Scanning (default: true)
    min-tier: high                 # Filter: critical, high, medium, low
    baseline: .sss-baseline.json   # Gate only on NEW findings

VS Code Extension Settings

{
  "solanaSecurityStandard.minTier": "medium",
  "solanaSecurityStandard.excludePatterns": [
    "**/tests/**",
    "**/target/**"
  ]
}

Semgrep Integration

# Custom config location
semgrep --config ./my-sss-rules.yaml ./programs

# Combine with other rulesets
semgrep --config auto --config ./semgrep/solana-security-standard.yaml ./programs

Common Patterns

Pre-commit Hook

# .git/hooks/pre-commit
#!/bin/bash
npx @jelleo/solana-security-standard scan programs/ --min-tier high
if [ $? -ne 0 ]; then
  echo "❌ Solana security scan failed — fix findings or commit with --no-verify"
  exit 1
fi

chmod +x .git/hooks/pre-commit

Baseline Workflow (gate on regressions only)

# 1. Generate baseline from current state
npx @jelleo/solana-security-standard scan ./programs --format json > .sss-baseline.json

# 2. Commit it
git add .sss-baseline.json
git commit -m "chore: add security baseline"

# 3. CI uses --baseline mode
npx @jelleo/solana-security-standard scan ./programs --baseline .sss-baseline.json

Only new findings fail the build.

Multi-repo Monorepo Scan

# Scan multiple program directories
npx @jelleo/solana-security-standard scan \
  ./packages/protocol/programs \
  ./packages/staking/programs \
  ./packages/governance/programs

Filter by Rule

CLI doesn't support per-rule filtering, but you can grep JSON output:

npx @jelleo/solana-security-standard scan ./programs --format json \
  | jq '.findings[] | select(.rule == "SOL-001")'

MCP Tools

When installed as MCP server:

scan_solana_code

// Agent calls this automatically when you say:
// "scan my Solana program for security issues"

{
  "code": "pub fn transfer(ctx: Context<Transfer>, amount: u64) -> Result<()> { ... }",
  "language": "rust"
}

// Returns findings with rule, severity, message, line numbers

list_solana_security_rules

// Agent calls when you ask:
// "what SOL-0XX rules apply to signer checks?"

// Returns all 37 rules with descriptions, tiers, examples

Troubleshooting

"Module not found" in CLI

# Clear npx cache and retry
rm -rf ~/.npm/_npx
npx @jelleo/solana-security-standard scan ./programs

VS Code extension not activating

1. Check output panel: View → Output → "Solana Security Standard"
2. Verify file is .rs, .ts, or .js
3. Reload window: Cmd/Ctrl+Shift+P → "Reload Window"

False positives

// SOL-004 flags this as missing signer check:
pub authority: AccountInfo<'info>,

// If it's intentional (e.g., read-only query), add comment:
pub authority: AccountInfo<'info>, // READ-ONLY: no sig required

// Or use Anchor's Signer<'info> to suppress:
pub authority: UncheckedAccount<'info>, // Explicitly unchecked

GitHub Action fails with SARIF error

# Ensure security-events: write permission:
permissions:
  contents: read
  security-events: write

Baseline not filtering findings

Baseline only gates new findings. If you want to suppress known findings:

# Regenerate baseline after fixing some issues
npx @jelleo/solana-security-standard scan ./programs --format json > .sss-baseline.json

Real Exploit Mappings

Every rule links to real disclosed exploits:

• SOL-001: Wormhole ($325M), percolator bounties
• SOL-008: Mango Markets ($114M)
• SOL-010: Cashio ($52M)
• SOL-017: Crema Finance ($8.8M)

See hacks/README.md for full database.

Badge for Your README

Show adoption:

[![Solana Security Standard](https://img.shields.io/badge/Solana%20Security%20Standard-SOL--0XX-a855f7?labelColor=6d28d9)](https://github.com/Copenhagen0x/solana-security-standard)

Resources

• Full rule catalog: https://github.com/Copenhagen0x/solana-security-standard/blob/main/claude-security-guidance.md
• Hacks database ($514M+): https://github.com/Copenhagen0x/solana-security-standard/blob/main/hacks/README.md
• Per-rule explainers: https://github.com/Copenhagen0x/solana-security-standard/tree/main/content
• Benchmark (rule accuracy): https://github.com/Copenhagen0x/solana-security-standard/blob/main/BENCHMARK.md