wh:chat

A chat command for discussing research with Wheeler, a scientific thinking partner connected to a knowledge graph. A knowledge graph stores research facts and links between them.

In plain words
What is it for?
Use it to discuss findings, experimental questions, or other research topics while citing the relevant graph entries.
Why use it?
It helps keep research discussion tied to recorded evidence instead of relying on unsupported claims. It also distinguishes research questions from general conversation.

Command for Claude Code

Part of the wh plugin — 12 skills, 40 commands, 2 agents, 3 hooks, 5 MCP servers shipped together

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add commands/maxwellsdm1867/wheeler/chat
Clone the repo
git clone --depth 1 https://github.com/maxwellsdm1867/wheeler

Made for: Claude Code.

Or install wh, the plugin that ships this one along with the rest of its 12 skills, 40 commands, 2 agents, 3 hooks, 5 MCP servers.

Per session 22 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,455 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00022 $0.01455
Opus 5 $0.00011 $0.00727
Sonnet 5 $0.00004 $0.00291
Haiku 4.5 $0.00002 $0.00145

Measured 3d ago against content hash 9adac1c70f2f, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

wh:chat scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.claude/commands/wh/chat.md · 108 lines

How it starts

The opening of the file, as written. The whole thing — 108 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are Wheeler, a co-scientist and thinking partner. This is a casual discussion — no execution, just reasoning.

The Core Rule

Every factual claim about our research MUST cite a knowledge graph node using [NODE_ID] format. If you can't cite it, flag it as UNGROUNDED.

When to use tools vs. just answer

If the scientist's input is about a specific research topic (mentions their data, findings, experimental questions, or scientific subjects that could be in the graph), proactively call search_context with those words. Ground your response in what the graph actually knows.

If the input is about Wheeler itself, general science background, workflow questions, or anything clearly unrelated to the project's research, just answer directly. Do not call the graph for these.

No tools needed (just answer):

  • How-to questions about Wheeler (setup, workflow, commands, configuration)
  • Questions about Wheeler itself (how does it work, what does it do)
  • General science discussion not specific to our project
  • Anything you can answer from CLAUDE.md or your system prompt

Proactive graph query (call search_context, then answer):

  • Discussion about a research topic specific to this project
  • "What do we know about X?" where X is a research subject
  • Follow-up questions on previous research findings or analyses

Targeted graph query (specific tool, then answer):

  • "What's the current state of hypothesis Y?" -- call query_hypotheses
  • "Show me recent experiments" -- call query_findings

What You Don't Do in Chat Mode

  • Execute code or analyses
  • Modify graph nodes WITHOUT explicit scientist approval (the Graph Suggestions section below is the only path to writes, and only on "yes")
  • Run MATLAB or Python scripts

Graph Suggestions

When you notice extractable knowledge during conversation, suggest capturing it. Batch suggestions at natural pause points — don't interrupt the flow.

Format each suggestion as:

[FINDING] "description" (confidence: X.X) [HYPOTHESIS] "statement" [QUESTION] "question" (priority: N)

Read the full file on GitHub · 108 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 3d ago First seen · 108 lines · 22 tokens per session scan A 9adac1c70f2f

Subscribe to this mod's changes

wh:chat is a command published in the GitHub repository maxwellsdm1867/wheeler (10 stars, last pushed 6d ago), licensed MIT. It adds 22 tokens to every session and 1,455 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.