The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill performs automated version checks and updates using remote resources from established services.
Fetches CLI version metadata from the NPM registry and skill release metadata from the GitHub API.
Executes updates using 'npm install -g minara@latest' for the CLI and 'git pull' for the skill source code from the author's official repositories.
[COMMAND_EXECUTION]: The skill relies on executing the minara CLI to perform all crypto and market operations.
Invokes commands for account management, balance checks, trading, and AI-driven analysis.
Modifies local system files including ~/.minara/ for configuration and various workspace files (e.g., ~/.claude/CLAUDE.md, ~/.openclaw/workspace/AGENTS.md) to integrate routing rules into the agent's environment.
[PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection due to its interaction with untrusted external data.
Ingestion points: External market data, price information, and prediction market content are retrieved and processed via the ask, research, and chat commands (SKILL.md, references/chat.md).
Boundary markers: The skill documentation enforces a strict separation between analysis and execution turns ('Analysis -> Trade boundary'), preventing immediate execution of trades based on research output.
Capability inventory: The skill possesses extensive fund-moving capabilities, including token swaps, transfers, withdrawals, and perpetual futures trading (references/swap.md, references/transfer.md, references/withdraw.md).
Sanitization: While no specific input sanitization scripts are provided, the skill mandates a human-in-the-loop confirmation step for all transaction-related commands, requiring explicit user approval before any funds are moved.

minara