brucelanlan/augur
Developer toolsThis page was translated automatically. Technical details may differ from the original. View the original
🦉 Augur — A local-first investment research memory system: 18 investment masters independently analyze + weighted consensus, evidence with temporal semantics, re-examinable after each run, CLI / Dashboard / MCP. A local-first research memory system with 18 investor personas.
Overview
⚠️ The master's views are standardized simulations based on publicly available investment concepts and do not represent my personal position; All scores, Kelly positions, and valuations depend on data quality and model assumptions. The screenshots below are all real AAPL runs (real-time market + SEC EDGAR) as of 2026-09-17, with no staged data. The terminal screenshots only capture the main output portion. A single command runs the entire research process. Before execution, check whether the required capabilities, domain names, and local resources are all declared in the checklist for each step; if any one is missed, execution will not start; Each execution is stored as RunBundle, which can be exported and reviewed. earnings-prep: 18 masters consensus, disagreements without evidence are separately marked as "evidence gap" and compared with the previous run. If there is no date in the financial report calendar, just write "No record" without guessing. filing-delta: Automatically retrieves the most recent two XBRL data sets of 10-K and lists substantial items with ≥5% change; no need to manually fill in accession numbers. There are also debt-covenant-review (leverage and interest coverage ratio) and insider-cluster-review (Form 4 insider cluster). See the Skill User Guide for details. augur research-report AAPL generates reports based on the most recent run: many-short votes, where the disagreement occurs (citing the evidence ID for each item and stating what facts can eliminate the disagreement), where the master consensus is, disagreements lacking evidence, as well as run IDs, code versions, step states, and evidence coverage. Compare the two most recent 10-K risk factors chapters to list additional, removal, and critical risks. Based on rule matching, minor wording changes may appear as both "New" and "Removal." Use augur backtest to generate historical records, remove each master one by one, and see whether the consensus IC improves or worsens; Scores are fixed and the master without much information is listed separately. Forward returns overlap daily, confidence intervals are optimistic, so they can only be used as screening, not as weights. augur eval factors provide the IC and t-tests for each factor.
Overview (machine translated)
🇨🇳 Chinese | 🇺🇸 English
🦉 Augur
**Not just another AI stock recommendation tool. It’s a local research memory system that helps you remember “when you knew something, what changed, and who was actually disagreeing with what.” **
⚠️ > ** For study and research purposes only and does not constitute investment advice. ** The master’s views are standardized simulations based on publicly available investment concepts and do not represent my personal position; All scores, Kelly positions, and valuations depend on data quality and model assumptions.
Contents: Why It’s Different · Screenshots of new v11 features · 5-minute start · Upgrading from v10 · 18 Masters · Dashboard · Access Claude / Agent · CLI Quick Search · Feature Maturity · Data and Privacy · Development · Documentation and Roadmap
🤔 Why Augur is different
Most AI financial tools give you an answer. Augur gives you re-examinable research products.
| Common AI Tools | Augur | |
|---|---|---|
| What to give you | A passage of “On one hand… On the other hand…” | Independent judgments of 18 masters + Where they are in agreement, where they fight |
| Where does the data come from | Unclear | Each piece of evidence brings the source, business timeline, available timepoint, and acquisition point |
| Missing data | Silence fill in 0 | Explicitly annotate missing, the master who relies on this field abstain and explain the reason |
| Looking back after three months | I don’t remember what I said last time | Each time I run an immutable RunBundle, you can export, compare, and review |
| Where is the data | Cloud | Local ~/.augur (can be modified with AUGUR_DATA_DIR), no telemetry |
✨ Overview of new v11 features
The screenshots below are all real AAPL runs (real-time market + SEC EDGAR) as of 2026-09-17, with no staged data. The terminal screenshots only capture the main output portion.
1. Executable Research Skill: augur skill run
A single command runs the entire research process. Before execution, check whether the required capabilities, domain names, and local resources are all declared in the checklist for each step; if any one is missed, execution will not start; Each execution is stored as RunBundle, which can be exported and reviewed.
earnings-prep: 18 masters consensus, points of divergence point by item, citing evidence IDs; disagreements without evidence support are separately marked as “evidence gap” and compared with the previous run. If there is no date in the financial report calendar, just write “No record” without guessing.
filing-delta: Automatically retrieves the two most recent XBRL data sets for 10-K and lists substantial items with ≥5% change, without manually filling in accession numbers. There are also debt-covenant-review (leverage and interest coverage ratio) and insider-cluster-review (Form 4 insider cluster).
This page was translated automatically. Technical details may differ from the original.
Install
This server does not publish a one-line install command.
Open the repository installation guide New to MCP? Start with the mcpskill.org guide →Configuration
{
"mcpServers": {
"augur": { "command": "augur-mcp", "env": { "AUGUR_EDGAR_CONTACT_EMAIL": "[email protected]" } }
}
}