<!-- SPDX-License-Identifier: MIT-0 Copyright 2026 InvestorClaw Contributors

This SKILL.md is MIT-0-licensed. The InvestorClaw service it connects to is Apache 2.0. See the InvestorClaw repository for that license. -->

InvestorClaw — Skill (openclaw runtime)

Powered by InvestorClaw (Apache 2.0). This skill file is MIT-0-licensed; the underlying service is Apache 2.0.

What this is

InvestorClaw is a containerized portfolio-analysis service exposed to openclaw as two MCP-HTTP servers:

• investorclaw — portfolio analysis tools at http://localhost:18090/mcp
• mnemos — memory + knowledge graph at http://localhost:5002/mcp

Both run inside a Docker compose stack on the user's machine (docker compose up -d is the entire service install). openclaw connects to them as native MCP servers via its mcp.servers config block — no plugin manifest, no dist/index.js, no npm install, no skill bootstrap files.

If openclaw runs in a container itself, the two MCP URLs reach the host's loopback through the compose bridge network or host.docker.internal, depending on how the openclaw container is launched. See INSTALL.md.

Tool surface

When the service is running, openclaw's tool catalog gains:

Portfolio analysis (investorclaw.*)

• investorclaw.portfolio_ask — natural-language portfolio question

routed through the deterministic engine

• investorclaw.portfolio_holdings — current snapshot of positions,

values, weights

• investorclaw.portfolio_performance — Sharpe, volatility, top/bottom

performers, max drawdown

• investorclaw.portfolio_bonds — bond analytics (YTM, duration, FRED

yield curve)

• investorclaw.portfolio_analyst — analyst ratings per holding
• investorclaw.portfolio_news — news correlation for held positions
• investorclaw.portfolio_lookup — ticker / account lookup
• investorclaw.portfolio_optimize — Sharpe / min-vol optimization
• investorclaw.portfolio_rebalance — current vs target with tax impact
• investorclaw.portfolio_scenario — what-if scenarios on holdings
• investorclaw.portfolio_cashflow — projected cashflow from bonds
• investorclaw.portfolio_peer — peer comparison vs benchmark
• investorclaw.portfolio_setup — auto-discover portfolio files in

/data/portfolios/

• investorclaw.portfolio_refresh — refresh market data without

re-uploading files

• investorclaw.portfolio_guardrails — view educational-only guardrails

Memory (mnemos.*)

• mnemos.search_memories — full-text + semantic search across

remembered observations

• mnemos.create_memory — record an observation about the user's

preferences, prior questions, or current investing context

• mnemos.list_memories — browse by category / date

How users interact with it

Users ask portfolio questions in openclaw chat. The LLM sees the MCP tools in its function-calling schema and routes the question to the right tool automatically. Examples:

• "What's in my portfolio?" → investorclaw.portfolio_holdings
• "How am I doing this year?" → investorclaw.portfolio_performance
• "What did I tell you about BABA last month?" → mnemos.search_memories

The first call after a cold cache may take 30–60 seconds while the deterministic pipeline builds the signed envelope; subsequent calls reuse the cache.

A typical flow:

1. User asks: "What changed since last review?"
2. openclaw's LLM calls mnemos.search_memories for prior portfolio

context.

1. LLM calls investorclaw.portfolio_holdings for the current snapshot.
2. LLM compares the two and synthesizes a narrative.
3. LLM calls mnemos.create_memory to record salient observations from

the review.

Recommended narrative model

openclaw's chat completion goes through whichever provider is configured in models.providers.<name> of ~/.openclaw/openclaw.json. Anthropic on openclaw — paid path only since 2026-04-04: routing OAuth- subscription tokens to a claws-agent violates Anthropic's ToS per their Apr 3 announcement. To use Anthropic models you need either (a) the discounted "extra usage bundle" add-on for your subscription, or (b) a direct Anthropic API key. Even with paid credits, Anthropic isn't cost-competitive with Together for InvestorClaw narrative work; we don't deploy Anthropic on our own fleet for openclaw.

Recommended:

• Default narrative — Together AI google/gemma-4-31B-it — serverless

tier, ~100 tok/s, ~$0.0008 / 1 K tokens, fleet default. This is what the InvestorClaw container expects via INVESTORCLAW_NARRATIVE_MODEL.

• Higher-quality alternative — Together AI MiniMaxAI/MiniMax-M2 —

larger context, but moved off Together's serverless tier 2026-05; requires a paid dedicated endpoint.

• Local-only / offline — Ollama gemma4:e4b on host — zero cloud

cost, GPU-bound, no key required.

Set TOGETHER_API_KEY in the InvestorClaw container's portfolios/keys.env (or via portfolio_keys_set) so the engine can synthesize narratives directly. openclaw's own model config is a separate concern.

After delivering analysis, the LLM should record only non-obvious observations the user might want next time — not every detail, just the ones that would be hard to recover from re-reading the data.

Important behaviors

• The investorclaw tools are deterministic at the data layer. Each

response includes a structured ic_result envelope plus a narrative text body. Trust the structured envelope — it is the source of truth. The narrative is decoration. If a portfolio file format isn't recognized, the tool returns a structured error with detected columns; surface that error and direct the user to the dashboard's column-mapping wizard at http://localhost:18092/portfolios/map.

• Educational only — never investment advice. All outputs include

the disclaimer envelope. Echo it when summarizing for the user. Do not recommend buying, selling, or holding specific securities.

• The MCP servers run on loopback by default. localhost:18090 and

localhost:5002. If the user deploys remotely (Tailscale VM, cloud host), the URLs change but the tool surface is identical.

• openclaw's own LLM provider config is separate. openclaw routes

its chat completions through models.providers.<name> in ~/.openclaw/openclaw.json (Together, OpenAI, Ollama, etc.). That is unrelated to InvestorClaw's optional narrative tier, which is configured inside the InvestorClaw dashboard at http://localhost:18092/.

v4.0 vs v2.x — what's different on openclaw

v4.0 eliminates the v2.x openclaw install friction:

• No openclaw.plugin.json manifest (there is no plugin)
• No dist/index.js (there is no plugin shim to compile)
• No install step inside an openclaw container
• No workspace bootstrap files (BOOTSTRAP.md / IDENTITY.md /

USER.md) to seed

• No schema-validation daemon to fight when writing provider config

for the plugin

The integration is just two MCP server URLs. openclaw's existing native MCP support handles the rest.

What this skill does NOT do

• Does not manage money or execute trades
• Does not give investment advice
• Does not access user accounts or move funds
• Educational outputs only

Install

OpenClaw / ZeroClaw / Hermes (ClawHub):

clawhub install investorclaw

Claude Code / Claude Desktop:

/plugin marketplace add argonautsystems/InvestorClaw
/plugin install investorclaw

See INSTALL.md in this directory for manual install steps.

Reporting issues

This SKILL.md describes how openclaw connects to the InvestorClaw service. If a tool returns an unexpected result, the issue is in the service (Apache 2.0 — mnemos-os/mnemos-ic-runtime and argonautsystems/InvestorClaw), not in this file. If openclaw fails to register the MCP servers, see INSTALL.md in this directory — in particular the note about always using the validated openclaw mcp set / openclaw config patch CLI rather than editing openclaw.json by hand.