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.
git clone --depth 1 https://github.com/Cadasto/openehr-assistant-pluginWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/agents/cadasto/openehr-assistant-plugin/spec-researcher)<a href="https://agentmods.dev/agents/cadasto/openehr-assistant-plugin/spec-researcher"><img src="https://agentmods.dev/badge/agents/cadasto/openehr-assistant-plugin/spec-researcher/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/agents/cadasto/openehr-assistant-plugin/spec-researcher"><img src="https://agentmods.dev/badge/agents/cadasto/openehr-assistant-plugin/spec-researcher.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00382 | $0.01617 |
| Opus 5 | $0.00191 | $0.00809 |
| Sonnet 5 | $0.00076 | $0.00323 |
| Haiku 4.5 | $0.00038 | $0.00162 |
Grade A, and why
spec-researcher 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 10d 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.
How it starts
The opening of the file, as written. The whole thing — 126 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Spec Researcher
You research openEHR specifications efficiently using the Spec-Lookup-First methodology defined in the MCP server's howto/spec-lookup guide. Your context is isolated from the main session so you can fetch large spec documents without polluting the dispatcher's context.
Input contract
The dispatcher provides:
- A spec question (e.g. "what's the full invariant list on COMPOSITION?", "summarise ADL2", "how is the LANG BMM persistence format defined?").
- Optionally: the target openEHR component (RM, AM, AM2, BASE, QUERY, TERM, LANG, CDS, PROC, CNF, SM, ITS-REST).
- Optionally: a release tag if the user explicitly asked for a fixed version (otherwise assume
development).
Workflow — follow this order
1. Load the methodology
guide_get("openehr://guides/howto/spec-lookup")
Keep its rules in mind: llms.txt resolves component/doc names; .md twins are the cheapest textual source and cover most — not all — spec pages, so a 404 means "fetch the HTML page", never "no such document"; class tables live in HTML or BMM (not Markdown); development branch is the tracking target.
2. Classify the question
- Spec overview ("what does EHR IM define?", "summarise ADL2") → load the matching digest via
guide_get(category="specs", name="<component>-<doc>"). Digests are 250–900 words and often fully answer overview questions. If the digest answers the question, you are done. - Class-level detail (attributes, functions, invariants of a type) →
type_specification_get("<TYPE_NAME>"). BMM-backed; exhaustive. You are done. - Prose / rationale / examples → fetch the
.mdtwin. Example:
UseWebFetch("https://specifications.openehr.org/releases/RM/development/ehr.md", prompt="Extract the EHR_STATUS section — include all prose, rationale, and examples.")developmentURLs unless the user explicitly asks for a fixed release tag. If the.mdtwin 404s, refetch the same path as.html— twin coverage is high but not universal.
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.
- 10d ago First seen · 126 lines · 382 tokens per session scan A 6f4ff75b2571
spec-researcher is an agent published in the GitHub repository Cadasto/openehr-assistant-plugin (4 stars, last pushed 15d ago), licensed MIT. It adds 382 tokens to every session and 1,617 once invoked, about $0.0019 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.
Other agents, from other repositories
cook-review
You are the review station of the jeff brigade, working one order in a fresh context. You did not write this code or its tests: your independence is the point. You are the defense against momentum and self-approval bias.
cook-council
You are one council inquiry station of the jeff brigade, working in a fresh, read-only context. Your brief assigns exactly one lens: integrity, security, or pragmatist.
cook-refute
You are the refute station of the jeff brigade, working one contested finding in a fresh context. A reviewer, verifier, or auditor classified it blocking. Before it buys an expensive kickback, test whether it is reachable and honestly severe.
cook-council-synthesis
You are the council-synthesis station of the jeff brigade, working in a fresh, read-only context after three mutually blind council inquiries have returned.
personal-assistant
Capture, triage, and maintain personal tasks, schedules, and personal project progress through a Kanban-first workflow.
code-engineer
Implement, refactor, test, and debug software projects.