Edgar Change Interpreter Claude skill

A short skill that makes Claude compare SEC 10-K, 10-Q and 8-K filings for material changes and silent risks, with quoted evidence and fixed output sections.

Our verdict Recommended with caveats

A sensible, tightly specified prompt, but it is small, little used (7 stars) and its MCP fetcher could not be reviewed.

What Edgar Change Interpreter does

Edgar Change Interpreter is a compact system-prompt skill for reading SEC filings through changes rather than summaries. You give Claude the text of a current filing, and optionally the prior one, and it returns a structured report. The rules in SKILL.md require every material change to carry an evidence quote of 25 words or fewer, separate fact from interpretation and alternative explanations, give a confidence rating of High, Medium or Low, and list open questions when evidence is missing. If no prior filing is supplied, the skill must not invent comparisons and treats the filing as a single snapshot. For 8-Ks it adds an item summary section. It states that it gives no investment advice, price targets or trade recommendations.

The output must follow a template file exactly, and there are three example files for a 10-K, 10-Q and 8-K, a user-input template and an optional JSON schema for a provenance manifest. The README also describes an optional MCP server in an mcp/edgar-server folder that resolves tickers and downloads filings from the SEC using a user-agent you configure. That folder was not in the copy we reviewed, so we could not check it. The SKILL.md itself is about 3 KB, with no scripts.

Good for

  • Comparing year-over-year risk factor changes in 10-K filings
  • Reviewing a 10-Q or 8-K for what was added or dropped
  • Analysts who want quoted evidence and explicit uncertainty

Not for

  • Investment recommendations or valuation
  • Large-scale automated filing screening

What's inside depth 2/5

A 3 KB SKILL.md with an output template, three examples and an optional schema, no scripts or tests in the reviewed files.

Skills
1
Reference docs
0
Scripts
0
Tests / evals
0
Templates
6
Avg SKILL.md
3 KB

Counted from the repository file tree. How we score depth.

Dependencies and costs

  • Optional edgar MCP server (Node.js 20+ per the README)
  • SEC user-agent name and email for the MCP server

Security review

Review before installing

A Markdown skill that interprets changes between SEC EDGAR filings, with an optional local MCP server that downloads filings from sec.gov and caches them on disk. The only credentials-like settings are your name and email for the SEC User-Agent. Setup copies files into ~/.claude/skills, registers the MCP server, and the README appendix suggests the nvm curl|bash installer.

Worth knowing

  • Makes network requestsThe optional MCP server fetches from data.sec.gov and www.sec.gov with a User-Agent built from SEC_USER_AGENT_NAME/EMAIL (rate-limited, cached on disk under .cache); the edgar.fetch_filing_text tool accepts any http(s) URL the agent supplies.mcp/edgar-server/src/edgar.ts:203
  • Installs packages or toolsThe MCP server needs npm install/build of @modelcontextprotocol/sdk ^1.0.4, dotenv, he, zod (locked versions carry known advisories per the scan).mcp/edgar-server/package.json:11
  • Changes files outside your projectREADME tells the user to copy the skill into ~/.claude/skills and register the server with claude mcp add-json edgar ....README.md:17
  • Installs with curl | shAppendix for installing Node.js suggests curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.40.1/install.sh | bash (third-party nvm installer, version-pinned).README.md:239

What we checked

The repository (28 files) on Oct 29, 2026. Not a full review: it relies on code outside the repository: npm dependencies of the MCP server (@modelcontextprotocol/sdk, dotenv, he, zod, ts-node, typescript) and the nvm installer script.

NVIDIA SkillSpector static scan: 32 pattern hits (Supply Chain 22, Privilege Escalation 3, Agent Snooping 3, Rogue Agent 2, Tool Misuse 1, Data Exfiltration 1). Most SkillSpector hits are known-CVE dependency versions in mcp/edgar-server/package-lock.json (not exercised by the skill text); the .env hits are dotenv loading of SEC user-agent settings; the 'data exfiltration' hit is the intended fetch from data.sec.gov; the curl|bash hit is the nvm appendix.

A review, not a certification. Skills change after we check them — read the files before installing.

License

MIT — commercial use allowed.

Can it run unattended?

Not really. It analyzes one filing or pair at a time on request and does not define a loop or verification step.

Alternatives to Edgar Change Interpreter

Claude for Financial Services

anthropics/financial-services

Anthropic's reference agents, skills and data connectors for banking, research, fund admin and KYC, usable as Claude plugins or Managed Agents.

With caveats Check first Depth 3/5

Wind AliceMarket

Wind-Alice/AliceMarket

Chinese-language finance skills for A-share and global equity analysis, plus Wind data and agent entry points that need a Wind API key.

With caveats Check first Depth 2/5

FAQ

What is Edgar Change Interpreter?

It is a Claude skill from CMD+RVL that identifies material changes, silent risks and interpretation traps in SEC filings. It compares periods when you give it a prior filing and quotes evidence for each change.

How do I install Edgar Change Interpreter?

Copy skills/edgar-change-interpreter into ~/.claude/skills and restart Claude. The README also describes an optional MCP server to fetch filings automatically, which needs Node.js 20 or newer.

Does it give investment advice?

No. The skill states it gives no investment advice, price targets or trade recommendations, and it asks Claude to state uncertainty and lower confidence where evidence is missing.

Edgar Change Interpreter vs anthropics/financial-services?

This skill does one narrow job on SEC filings. Anthropic's financial-services repo is a broader set of finance plugins. Choose this one only for filing change analysis.

Sources: GitHub repository (stars, license, file tree, last push Sep 19, 2026). Review based on the README and SKILL.md files as published.