CometixCode
Plugin / AgentAn unofficial Rust 1:1 reimplementation of the Claude Code terminal UI, ported file-for-file onto an iocraft TUI framework.
- terminal-ui
- rust
- reimplementation
- unofficial
Anthropic's official Claude Cowork plugin for financial advisors — eight ready-to-run compliance-aware workflows over the firm's own connectors.
A plugin from Anthropic that gives financial advisors ready-to-run workflows for Claude Cowork drawing live data from the tools their firm already uses via third-party connectors. The plugin is a set of skills written as plain instructions (Markdown and JSON); it holds no client data — when an advisor runs a skill, Claude pulls what it needs from the connected systems, and every action that writes back to an external system (CRM updates, client drafts) pauses for explicit advisor approval. Eight skills cover the advisory loop: /onboarding (learns the firm, priorities and tech stack), /pre-meeting (client review prep doc with snapshot, relationship signals, portfolio review and agenda), /post-meeting (turns a transcript into a CRM note), /compliance (pre-checks client-facing material against SEC Marketing Rule, antifraud, books-and-records and Reg BI, producing a findings table, suggested disclosures and a clean redraft), /prospect-intake, /portfolio-rebalance-review, /alts-brief and /estate-and-tax-brief. Nine connectors are live — Addepar, BlackRock, Envestnet Tamarac, iCapital, MoneyGuide, Orion Advisor Solutions, Wealthbox, Wealth.com and Zocks — with Schwab marked coming soon; every skill falls back to paste/upload when a system isn't connected.
US-registered advisory firms that want Claude over their own portfolio, planning and CRM data without anything leaving the approval loop; also the reference design to fork for other vertical Cowork plugins.
claude plugin marketplace add anthropics/claude-for-financial-advisors claude plugin install claude-for-financial-advisors
Apache-2.0. Anthropic labels it a reference implementation: not actively maintained or monitored, not accepting contributions. Some guardrails (advisor approval before writes) are enforced by the model following the instructions, not by the runtime — the README says so explicitly. Third-party content read by subagents should be treated as untrusted. For financial-professional use only; educational and operational support, not investment, legal or tax advice.
Repository
anthropics/claude-for-financial-advisorsCharted
An unofficial Rust 1:1 reimplementation of the Claude Code terminal UI, ported file-for-file onto an iocraft TUI framework.
Hook-level enforcement for the natural-language rules in AGENTS.md and CLAUDE.md — every edit is judged and violations are repaired in the same turn.
A Claude Code plugin that replaces the lossy compaction summary with Jev-scored pruning: stale tool calls are dropped, and everything kept stays verbatim.