We The People

"People shouldn't be afraid of their government. Governments should be afraid of their people." - V

A bill and legislation lookup skill for AI agents.

Ask your AI agent about a bill, a senator, a campaign donor, or a lobbying disclosure. We The People returns the answer with sources cited. Built for civic engagement, journalism, advocacy, and anyone who wants to know what the U.S. government and their elected representatives are actually doing.

Works with Cowork, OpenClaw, Hermes Agent, Claude Code, Codex, and any other agent platform that supports the agentskills.io standard. One install, all platforms.

---

What it does

Give a command to look up a bill:

"Do a We The People search on HR22"

Ask questions like:

"What does H.R. 22 actually do?"
"Who are the top donors to Senator Elizabeth Warren?"
"Has Chip Roy traded any stocks in defense companies?"
"Who's lobbying for the SAFE Banking Act?"
"Has this immigration bill been introduced before?"
"What did Bernie Sanders say about the Inflation Reduction Act on the Senate floor?"

We The People pulls the answer from authoritative sources — Congress.gov, the FEC, OpenSecrets, Unusual Whales, Senate lobbying disclosures, C-SPAN, and more — and gives you the answer with links you can verify yourself.

It's a lookup tool. You ask, it answers, you decide what to do with the information.

---

Installation

You need three things to get started:

1. A free API key from https://api.data.gov/signup/ (one-minute email signup — no payment, no waiting list)
2. The skill files dropped into your platform's skills folder
3. The two env vars set with your API key

Pick the section below that matches your platform.

Cowork

1. Install the skill files:
   • Open Cowork settings → Skills → Add skill.
   • Either upload the we-the-people folder, or paste this repo URL: https://github.com/SamaritanOC/We-The-People.
2. Add the API key:
   • Open Cowork settings → Environment variables (or Secrets).
   • Add two entries with the same value:

     FEC_API_KEY=your_key_here
     CONGRESS_API_KEY=your_key_here
     

3. Restart your Cowork session. The skill is ready to use.

OpenClaw

1. Open OpenClaw Mission Control (the dashboard).
2. Go to the Skills section.
3. Click Add Skill (or Install Skill).
4. Either upload the we-the-people folder, or paste this repo URL: https://github.com/SamaritanOC/We-The-People
5. Once installed, open the Settings or Environment Variables section in Mission Control.
6. Add these two variables, both set to your api.data.gov key:

   FEC_API_KEY=your_key_here
   CONGRESS_API_KEY=your_key_here
   

7. Save and restart your OpenClaw session.

Claude Code

1. Place the skill folder where Claude Code can see it. Either inside your project (recommended) or in a global skills location:

   # Inside a project
   mkdir -p ./skills && cp -r we-the-people ./skills/
   
   # Or globally for all projects
   mkdir -p ~/.claude/skills && cp -r we-the-people ~/.claude/skills/

2. Run the interactive setup script:

   cd ./skills/we-the-people
   chmod +x setup.sh scripts/*.py
   ./setup.sh

3. Make the env vars available. Either:

   • Add source ./skills/we-the-people/.env to your shell profile, OR
   • Add the env vars to your project's .env and reference it in your Claude Code config

4. Reference the skill in your conversation. Claude Code will read SKILL.md and follow it.

Codex

1. Drop the skill folder into your Codex skills directory:

   cp -r we-the-people ~/.codex/skills/

2. Run the interactive setup script:

   cd ~/.codex/skills/we-the-people
   chmod +x setup.sh scripts/*.py
   ./setup.sh

3. Make the env vars available to Codex. Add to your shell profile:

   source ~/.codex/skills/we-the-people/.env

4. Start a new Codex session.

Hermes Agent

Hermes Agent (by Nous Research) supports installing skills directly from a GitHub repo, with no manual file copying.

1. Install Hermes Agent if you haven't already — one-line install at https://hermes-agent.nousresearch.com.

2. Install the skill. From inside a Hermes chat session:

   /skills install SamaritanOC/We-The-People
   

   Or from your terminal:

   hermes skills install SamaritanOC/We-The-People

3. First time the skill loads, Hermes will prompt you for your api.data.gov key. Get a free one at https://api.data.gov/signup/ if you don't have one. Hermes stores it securely and passes it to the skill automatically — no manual .env editing needed.

4. Use the skill by asking naturally in any Hermes chat (the agent loads it when it's relevant), or invoke explicitly:

   /we-the-people who sponsored H.R. 22?
   

Hermes works on the desktop CLI plus messaging platforms (Telegram, Discord, Slack, WhatsApp, Signal, Email), so you can query We The People from anywhere your Hermes is connected.

Other Anthropic-compatible platforms

The skill format is the standard Anthropic skills layout: SKILL.md at the root with frontmatter, supporting reference files in references/, helper scripts in scripts/. Drop the folder wherever your platform reads skills from, set the two environment variables (FEC_API_KEY and CONGRESS_API_KEY), and start a session.

Just running the helpers manually

If you're not using an agent platform and just want the API helpers:

cd we-the-people
chmod +x setup.sh scripts/*.py
./setup.sh
source .env

python3 scripts/congress_lookup.py bill --congress 119 --type hr --number 22
python3 scripts/fec_lookup.py candidate --name "Roy, Chip" --state TX

Verify it's working

Ask the agent something like:

"Look up H.R. 22 in the 119th Congress and tell me who sponsored it."

The agent should pull the bill metadata, name the sponsor, and cite Congress.gov as the source.

For troubleshooting, see SETUP.md.

---

Example interactions

You: "Pull up H.R. 22 and tell me who sponsored it."

Agent: Pulls bill metadata from the Congress.gov API, names the sponsor (Chip Roy, R-TX-21), summarizes what the bill does in plain language, links to the full text. Around 30 seconds.

---

You: "Top donors to Roy in 2024."

Agent: Hits the FEC API for Roy's principal campaign committee, returns the top 20 contributing employers with totals, links to OpenSecrets for the donor profile. Optionally: "Save it as a spreadsheet." → produces an .xlsx file you can open in Excel.

---

You: "Has Roy traded any stocks while H.R. 22 was active?"

Agent: Checks Unusual Whales and the House disclosure portal, summarizes any trades in the relevant window, flags any companies that benefit from the bill.

---

Where the data comes from

Question	Source
Federal bill text, sponsors, votes	Congress.gov
Federal bill summaries, plain-language	GovTrack
State bills	LegiScan, Ballotpedia, Open States
Campaign finance (federal)	OpenSecrets, FEC.gov
Campaign finance (state)	FollowTheMoney
Stock trades by legislators	Unusual Whales, House/Senate disclosure portals
Lobbying	Senate LDA, OpenSecrets Lobbying
Floor speeches, hearings	C-SPAN, Congressional Record
News coverage	AP, Reuters, Politico, The Hill, ProPublica
Polling	Gallup, Pew, YouGov, FiveThirtyEight

Full source list with URLs in references/sources.md.

---

What you need

• An AI agent platform that supports skills (Cowork, OpenClaw, Hermes Agent, Claude Code, Codex, or compatible)
• A free api.data.gov API key (covers both Congress.gov and FEC)
• Python 3 and bash (for the helper scripts and setup)

That's it. Everything is free. Everything is public-record. No paid services, no data brokers.

---

Security

The skill treats all fetched web content as untrusted data and will not follow instructions it finds on websites. This protects against prompt injection attacks, where a malicious page attempts to manipulate the agent — telling it to change its answer, ignore your question, recommend a different bill, or send data somewhere. If a page tries any of that, the skill stops and tells you.

The skill only fetches from a curated list of authoritative sources: Congress.gov, the FEC, OpenSecrets, Unusual Whales, Senate lobbying disclosures, and other official or widely-recognized outlets listed in references/sources.md. If you ask it to fetch from a domain outside that list, it will confirm with you before proceeding.

If the skill detects a manipulation attempt while processing a page, it stops immediately, tells you the URL and a plain-language description of what looked suspicious, and offers you three options: skip that source and continue with others, block the site for the rest of this session, or blacklist it permanently.

Permanently blacklisted domains are stored in blacklist.txt at the skill root. You can also edit that file manually at any time to pre-block sites you know you don't want the skill to access.

---

What's in this repo

we-the-people/
├── SKILL.md                          ← agent reads this first
├── README.md                         ← you are here
├── SETUP.md                          ← detailed install for every platform
├── LICENSE                           ← free for non-commercial use; no commercial use, no resale
├── AGENTS.md                         ← OpenClaw compatibility entry point
├── blacklist.txt                     ← domains the skill will never fetch from
├── setup.sh                          ← interactive API key configuration
├── .env.example                      ← key template
├── references/
│   ├── sources.md                    ← every primary data source
│   ├── search-operators.md           ← Google operators for legislative work
│   └── output-formats.md             ← file format rules for saved outputs
└── scripts/
    ├── congress_lookup.py            ← Congress.gov API helper
    └── fec_lookup.py                 ← FEC API helper

---

License

Free for non-commercial use — including personal use, journalism, advocacy, research, education, students, civic work, and nonprofits.

Commercial use is not permitted under this license. You may not sell, repackage, rebrand, or redistribute the Software as your own product, and you may not bundle it with paid services or use it as a material part of a commercial deliverable.

For commercial licensing inquiries, contact hm@smbconsultants.ai.

Full terms in LICENSE.

---

Lawful use

This skill uses publicly available, indexed information from government data portals, news archives, and public-record databases. No intrusion. No unauthorized access. No data brokers.

Use it for journalism, civic research, academic work, advocacy, or just understanding what your representatives are actually doing.

Don't use it to harass or stalk individuals. That use is explicitly prohibited by the license.

---

Support and contributions

Built by Harold Mansfield
Website: SMBConsultants.ai

Questions, bug reports, feature requests: https://github.com/SamaritanOC/We-The-People/issues

If this skill is useful to your work, consider supporting its development at The Samaritan Project.