Skip to content
Agentmart
Submit a tool

genomilab

Run or continue patient-authorized, genome-informed GenomiLab investigations in the current Claude, Codex, or other MCP agent task.

Skill by exon-research

IndexedApache-2.0★ 489
Write a review

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/exon-research-genomilab

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Run or continue patient-authorized, genome-informed GenomiLab investigations in the current Claude, Codex, or other MCP agent task. Use when a patient asks to open the Research Desk, investigate a condition against their active genome, review an existing investigation, supply follow-up information, revise a hypothesis, or publish a revised investigation response.

Use cases: research, productivity, agent-building

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/genomilab

What's inside

Skill name
genomilab
Description
Run or continue patient-authorized, genome-informed GenomiLab investigations in the current Claude, Codex, or other MCP agent task. Use when a patient asks to open the Research Desk, investigate a condition against their active genome, review an existing investigation, supply follow-up information, revise a hypothesis, or publish a revised investigation response.
License
Apache-2.0
Instructions
343 lines in SKILL.md · view on GitHub

Files (2)

  • SKILL.md
  • openai.yaml · agents/

SKILL.md (first lines)

# GenomiLab Research Desk

Keep the current host agent in control of the conversation, native specialist
subagents, planning, streaming, follow-ups, and cancellation. Use GenomiLab for
typed capabilities, patient authorization, Active Genome Index access, durable
evidence and hypothesis state, validated briefs, and the patient portal.

The portal is for patient onboarding, exact approvals, integration setup, and
monitoring committed investigation milestones. Never start a second agent task
from the portal.

## Start or resume

1. Call `genomilab.open_workspace`.
2. If it returns `status="setup_required"`, keep setup in core Genomi. Select
   or finish the current user's Active Genome Index. If none exists, ask the
   user for the local VCF or another supported genome-source path and use core
   Genomi intake; pointing the host at that path is the only genome handoff.
   Do not open an investigation without a query-ready selected index.
3. Show the returned portal link when the patient needs onboarding or approval.
4. Call `genomilab.create_investigation` for a new question, or
   `genomilab.inspect_investigation` for an existing investigation.
5. If no profile observation exists, ask for one concise patient-reported fact
   before preparing authorization. Do not fabricate a symptom, diagnosis,
   phenotype, family history, or molecular finding.

Do not ask for a VCF path in GenomiLab. Genome intake remains in core Genomi;
GenomiLab uses the selected Active Genome Index.

## Chair the specialist board

For every new investigation, act as chair and form 2–5 native host subagents
with adaptive, non-overlapping domain roles. Give each specialist an explicit
role and bounded task chosen for the question; do not use a fixed board when a
different evidence mix is more relevant. Use stable logical `specialist_id`
values, not native task or thread identifiers. Record the board once with
`genomilab.form_specialist_board` before submitting a plan. These are persistent
specialist identities: reuse the same IDs in every investigation round, while
giving each specialist a new bounded assignment for that round.

On resume, inspect the investigation first. Treat its pre-authorization
`specialist_board` as a structural redacted marker only: board existence plus
`status` and `member_count`; a static chair description may also appear. If
that marker says a board exists, do not call
`genomilab.form_specialist_board` again. Renew current-session authorization
through the flow below. After `private_context_status` is
`approved_for_session`, inspect again; only then read and reuse the full
specialist IDs, roles, tasks, and current-work states, and reconstitute the
corresponding native subagents if the host requires it.

The chair alone owns the patient conversation, authorization, all private AGI
reads, and canonical plan, hypothesis, gap, and brief commits. Give specialists
public questions or only the minimum approved evidence needed for their task.
Specialists return their analysis to the chair; they never read the AGI
directly, interact with the portal, request patient approval, or commit the
canonical investigation artifacts.

Call `genomilab.report_specialist_progress` with the current `round_id` only at
meaningful work milestones, using `working`, `blocked`, or `completed` and a
short `current_work` label.
GenomiLab derives the initial `assigned` state. Do not send raw agent messages,
chain of thought, token streams, native task IDs, or per-call chatter. The
portal monitors only these committed board milestones.

When a specialist returns, use `genomilab.record_specialist_report` to commit
that round's findings and gaps with exact evidence and profile anchors. A
specialist report is traceable synthesis, not a new evidence record or a
clinical conclusion. Commit all assigned reports before starting another round.

## One investigation authorization

Call `genomilab.prepare_authorization` after the patient profile contains the
facts needed for the question. The portal presents the exact selected profile
slice, current genome scope, and current underlying-agent destination.
Give the patient the portal launch link returned by that call: its one-time
token targets the exact signed candidate, while the candidate itself remains
redacted from the agent-facing result and URL. Do not prepare another candidate
merely to display the review.
For an existing investigation in a new agent session, omit
`observation_revision_ids` to renew the pinned profile, AGI, scope, and purpose
without silently expanding to newer profile facts. Use
`genomilab.record_patient_observations` for an intentional context update.
Each local stdio MCP `initialize` handshake is a new GenomiLab agent session.
It closes the prior session's private authority even if the host reports the
same client name and version, so do not reinitialize the stdio MCP connection
mid-investigation. HTTP MCP initialization is public-tools-only and cannot
create or replace the private GenomiLab runtime.

The agent must not approve this candidate. Ask the patient to review it in the
portal, then poll with `genomilab.inspect_investigation`. Continue when
`private_context_status` is `approved_for_session`.

Do not request another approval for routine local planning, evidence work,
hypothesis updates, or brief publication. A new patient approval is appropriate
only when the profile or genome snapshot/scope changes. Exact external-provider
egress can require a separate just-in-time portal approval.

## Plan and execute

Read `capability_catalog` from `genomilab.inspect_investigation`. Submit only
requests that are currently advertised as available:

```json
{
  "investigation_id": "investigation-...",
  "focus_question": "What does the approved profile establish?",
  "specialist_assignments": [
    {
      "specialist_id": "specialist-timeline",
      "task": "Reconstruct the approved clinical timeline"
    },
    {
      "specialist_id": "specialist-evidence",
      "task": "Review relevant public evidence"
    }
  ],
  "requests": [
    {
      "id": "profile-review",

From exon-research/genomi, Apache-2.0.

More from exon-research/genomi

Signals

Stars
489
Forks
61
Open issues
0
Last commit (this folder)
Aug 17, 2026 (2 months ago)
Last release
Jul 1, 2026 (3 months ago)

From GitHub (fetched Oct 24, 2026).

Details

License
Apache-2.0
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from exon-research

Cite this page

genomilab skill. Agent Mart. https://agentmart.dev/skills/exon-research-genomilab (last updated 2026-10-24)