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.
npx agentmods add commands/grainulation/wheat/researchgit clone --depth 1 https://github.com/grainulation/wheatWhat 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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00000 | $0.00605 |
| Opus 5 | $0.00000 | $0.00302 |
| Sonnet 5 | $0.00000 | $0.00121 |
| Haiku 4.5 | $0.00000 | $0.00060 |
Grade A, and why
research 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 yesterday.
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.
How it starts
The opening of the file, as written. The whole thing — 75 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/research — Deep dive on a topic
You are researching a topic for the current Wheat sprint. Read CLAUDE.md for sprint context and claims.json for existing claims.
Process
-
Understand the request: The user's argument tells you what to research. Could be a technology, a comparison, a question, a process.
-
Research deeply: Use web search, read documentation, check connected repos (see Connectors in CLAUDE.md). Be thorough — this is the foundation for later decisions.
-
Extract claims: Every finding becomes a typed claim. Be specific and verifiable. Bad: "Auth0 is popular." Good: "Auth0 serves 15,000+ customers as of 2025."
-
Detect conflicts with existing claims: Check claims.json. If your new findings contradict existing claims, set
conflicts_withon both the new and existing claim.
Adding claims
Append claims to claims.json with IDs continuing the r### sequence (check existing claims for the next number). Each claim:
{
"id": "r001",
"type": "factual|estimate|risk|recommendation",
"topic": "<topic category>",
"content": "<specific, verifiable finding>",
"source": {
"origin": "research",
"artifact": "research/<topic-slug>.md",
"connector": null
},
"evidence": "web",
"status": "active",
"phase_added": "research",
"timestamp": "<ISO timestamp>",
"conflicts_with": [],
"resolved_by": null,
"tags": ["<relevant tags>"]
}
If the finding came from a connector (GitHub repo, Jira, etc.), set evidence to "documented" and fill in the connector field.
Run the compiler
npx @grainulation/wheat compile --summary
Check for new conflicts introduced. Report them to the user.
Generate HTML explainer
Create research/<topic-slug>.html — a self-contained HTML explainer using the dark scroll-snap template style. This should be:
- Beautiful and presentable (stakeholders will see this)
- Organized into logical sections (scroll-snap slides)
- Include key findings, comparisons, tradeoffs
- Reference claim IDs so findings are traceable
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.
- yesterday First seen · 75 lines · 0 tokens per session scan A db7d1889d6c7
research is a command published in the GitHub repository grainulation/wheat (20 stars, last pushed 4mo ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 605 tokens. 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 2026-08-30.
Other commands, from other repositories
advisor
Advisory gate for triage or plan decisions. Spawns a second-opinion agent that challenges assumptions, surfaces risks, and proposes alternatives before the decision commits. Based on Anthropic advisor tool pattern.
guide
You are a triage agent who continuously prioritizes loom:issue issues by applying loom:urgent to the top 3 priorities.
pm-review
Review local git diff with reasoning + snippets; cross-check; dispose findings into the rules library.
new
Creates a new item in a module that supports the new capability.
card
Open one self-contained OverClick card.
linear-bulk
Apply the same change to many Linear issues at once. Dry-run by default.