New: 98,000+ skills indexed from GitHub repos with 100+ stars · every skill safety-checkedBrowse top skillsTop skills
Esc to closeBrowse all skills
Submit a skillBrowse skills
Submit a skillBrowse skills

Agent skill

Run Organization

by harrisongzhang in harrisongzhang/TheVirtualBiotech

How to organise a session run so a human can audit it — directory layout, artifact naming, recording the analysis plan, filing claim-evidence objects, and the end-of-run checklist.

MITAuto-check passedResearch & Science

Install Run Organization

skills CLI
$ npx skills add harrisongzhang/TheVirtualBiotech --skill run-organization -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install harrisongzhang/TheVirtualBiotech run-organization --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/harrisongzhang/TheVirtualBiotech.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.claude/skills/run-organization .claude/skills/run-organization && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
run-organization
GitHub stars
122
Token cost
~1.8k tokens
SKILL.md length
710 words
Files
1
Skills in repo
4
Repo updated
First seen
Licence
MIT

At a glance

How to organise a session run so a human can audit it — directory layout, artifact naming, recording the analysis plan, filing claim-evidence objects, and the end-of-run checklist.

  • Works in 4 steps: Declare the plan before dispatching → Keep artifacts in the right place → File claim-evidence objects → …
  • Orchestrating specialists (the CSO role)
  • SKILL.md covers Why this exists, The run directory, 1. Declare the plan before… and 2. Keep artifacts in the right…, plus 3 more sections
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

Run Organization is an agent skill from harrisongzhang/TheVirtualBiotech. How to organise a session run so a human can audit it — directory layout, artifact naming, recording the analysis plan, filing claim-evidence objects, and the end-of-run checklist. Use when orchestrating specialists (the CSO role), whenever you are about to dispatch work or synthesise findings, or when you need to know where an artifact belongs.

Its SKILL.md is about 1.8k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Research & Science. The repository describes itself as: Multi-agent AI system for drug-target identification and due diligence. The licence is MIT.

When your agent uses it

  • Orchestrating specialists (the CSO role)
  • Whenever you are about to dispatch work
  • Synthesise findings
  • You need to know where an artifact belongs

Example prompts

  • “/run-organization”

Requirements

  • Python 3

Workflow steps

4 steps, taken from the step headings in SKILL.md.

  1. Declare the plan before dispatching
  2. Keep artifacts in the right place
  3. File claim-evidence objects
  4. End-of-run checklist

What it can do on your machine

Read from SKILL.md and the folder at commit 71f9da6. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md (its code samples are python).

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Run Organization loads about 1.8k tokens when it runs. Until then it costs about 91 tokens; SKILL.md has 710 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~91
When it runs · the whole SKILL.md, loaded when a task matches
~1.8k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from harrisongzhang/TheVirtualBiotech at commit 71f9da6, republished under its MIT licence (© harrisongzhang). 710 words, ~1,799 tokens.

Download SKILL.mdSave it as .claude/skills/run-organization/SKILL.md (or your agent's skills folder).
name
run-organization
description
How to organise a session run so a human can audit it — directory layout, artifact naming, recording the analysis plan, filing claim-evidence objects, and the end-of-run checklist. Use when orchestrating specialists (the CSO role), whenever you are about to dispatch work or synthesise findings, or when you need to know where an artifact belongs.

Run Organisation

Why this exists

A reviewer ran the system, got a pile of files, and could not tell how they were organised, which artifact supported which claim from which specialist, or how the analysis flowed together. They also could not replay it.

That is a fair description of what an unmanaged run produces. This skill is the standard that prevents it. The rule behind every instruction here:

Every run must be understandable by someone who was not there, opening the directory cold, six months later.

The run directory

Every session gets one directory. You are working inside it now.

runs/<RUN_ID>/
├── MANIFEST.json      every artifact: hash, producing agent, producing tool call
├── README.md          generated map of the run — do not hand-edit
├── audit.html         self-contained report for someone with no environment
├── inputs/
│   ├── query.txt      the user's turns, verbatim
│   ├── plan.json      the analysis DAG you declared      ← you write this
│   └── config.json    models, prompt hashes, MCP servers, git commit
├── work/<agent>/      one subtree per specialist — never flat
│   ├── code/scripts/
│   ├── data/{raw,processed}/
│   └── results/{figures,tables,reports}/
├── evidence/
│   ├── claims.json    claim → evidence                    ← you write this
│   └── provenance.json  tool call → agent, derived from the trace
├── logs/              trace.jsonl, cost_report.json, transcript.md
└── report/FINAL_REPORT.md

MANIFEST.json, provenance.json, README.md and audit.html are generated. Your two responsibilities are plan.json and claims.json.

1. Declare the plan before dispatching

Before dispatching two or more specialists, call mcp__provenance__write_plan.

python
write_plan(
  goal="Assess the safety risk of targeting IL-33 in asthma",
  steps=[
    {"id": "s1", "agent": "single-cell-analyst",
     "task": "IL33/IL1RL1 expression across lung and critical-organ cell types",
     "depends_on": [], "expected_outputs": ["il33_celltype_expression.csv"]},
    {"id": "s2", "agent": "fda-safety-officer",
     "task": "Clinical precedent AEs read against the expression profile",
     "depends_on": ["s1"]},
  ])
  • depends_on: [] → can start immediately. Steps that do not depend on each other are dispatched in parallel.
  • Use depends_on only for real data dependencies. Over-declaring serialises work that could have run concurrently and makes the run slower for no auditing benefit.
  • The plan is validated on write: cycles, unknown ids and duplicates are rejected.
  • Deviating is allowed and expected. If a specialist's findings change what should happen next, do that — deviations are recorded, not forbidden. If the change is substantial, call write_plan again with the revised plan.
  • Skip the plan for single-specialist queries and clarification exchanges.

2. Keep artifacts in the right place

Each specialist writes only under work/<its-own-name>/. Its prompt tells it so, and the system records anything written elsewhere and attributes it anyway — but a file in the wrong place is still a file the next reader has to puzzle over.

When you delegate, if a specialist needs an earlier one's output, give it the path:

Load the expression table from
work/single-cell-analyst/results/tables/il33_celltype_expression.csv

Do not tell a specialist to "check the workspace" and hope. Name the file.

Naming

Name for content, never for sequence or status.

GoodBadWhy
il33_celltype_expression.csvanalysis2.csvsays what is in it
gwas_credible_sets_chr9.parquetcs_2fd0.parquetreadable six months later
safety_ae_summary.mdresults_final_v3.md"final v3" tells a reader nothing

If you find yourself appending _v2, the first file was either superseded (say so in its description) or the two differ in a way the name should state.

Show full SKILL.md (350 more words)Show less

3. File claim-evidence objects

Every substantive factual assertion in your synthesis becomes a claim with the evidence behind it. See the evidence-citation skill for the specialist-side contract that produces citable artifacts in the first place.

Sequence:

  1. mcp__provenance__list_artifacts — get the exact paths. Do not guess a filename; a path that does not exist is rejected.
  2. Write your synthesis with inline anchors: ...highly expressed in lung mast cells[[claim:C1]]...
  3. mcp__provenance__record_claims with the claim objects.
python
record_claims(claims=[
  {"id": "C1",
   "text": "IL1RL1 is most highly expressed in lung mast cells (mean 2.4 CPM)",
   "agent": "single-cell-analyst",
   "confidence": "strong",
   "evidence": [
     {"kind": "table",
      "path": "work/single-cell-analyst/results/tables/il33_celltype_expression.csv",
      "note": "row: mast cell"},
     {"kind": "figure",
      "path": "work/single-cell-analyst/results/figures/il33_celltype.png"}]}])

Rules that matter:

  • Evidence is validated on write. ok: false means a path or tool id is wrong — fix it and call again.
  • Never resolve a rejection by deleting the evidence. A claim you cannot support is a finding: state it in prose as unsupported or uncertain. Filing an unsupported claim as though it were supported is the specific failure this whole mechanism exists to prevent.
  • confidence: strong = direct measurement; moderate = inference; weak = suggestive or indirect.
  • Every [[claim:Cn]] anchor must correspond to a filed claim. A dangling anchor renders as a visibly broken marker and is reported as a defect in the README.

4. End-of-run checklist

Before your final response:

  • write_plan called, if two or more specialists ran
  • Every substantive assertion carries a [[claim:Cn]] anchor
  • record_claims returned ok: true for all of them
  • list_artifacts shows no unexplained files — anything a specialist produced that carries a finding should have a description
  • Anything you could not establish is stated as a gap, not omitted

Anti-patterns

Filing a claim with weak evidence to clear the checklist. The confidence field exists so you can be honest. weak with a real artifact beats strong with a stretched one.

Citing a specialist's prose. If a specialist asserted a number but wrote no file, there is nothing to cite. Report it as an unsupported statement, or send the specialist back to produce the artifact.

One claim covering a whole paragraph. One claim = one checkable assertion. If the text spans three findings from two specialists, that is three claims.

Silently dropping a failed analysis. A specialist that timed out or returned nothing is part of the run's story. Say so, and adjust your confidence.

© harrisongzhang, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in .claude/skills/run-organization of harrisongzhang/TheVirtualBiotech.

Open the folder on GitHubat commit 71f9da6

Compare with similar skills

Run Organization next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Run Organization compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Run Organization this skillharrisongzhang/TheVirtualBiotech122—~1.8kAutomated safety check: PassMIT
Hypothesis Generationspacering-net/codeg3.9k14 repos~3.6kAutomated safety check: NotesMIT
GitHub Deep Researchbytedance/deer-flow84k4 repos~1.3kAutomated safety check: PassMIT
Nature Paper CardYuan1z0825/nature-skills47k2 repos~2.1kAutomated safety check: PassApache-2.0
Content Research Writerweapp-tailwindcss/weapp-tailwindcss1.9k25 repos~3.5kAutomated safety check: PassMIT
Last30daysmvanhorn/last30days-skill64k—~7.9kAutomated safety check: NotesMIT

Similar skills

  • Hypothesis Generation

    spacering-net/codeg

    Structured hypothesis formulation from observations. An agent skill from spacering-net/codeg.

    3.9k GitHub starsUsed in 14 repos~3.6k tokens
    Research & ScienceAuto-check: notes

  • GitHub Deep Research

    bytedance/deer-flow

    Researches a GitHub repository over four rounds using the GitHub API and web search, then writes a structured markdown report with timeline, metrics and Mermaid diagrams.

    84k GitHub starsUsed in 4 repos~1.3k tokens
    Research & ScienceAuto-check passed

  • Nature Paper Card

    Yuan1z0825/nature-skills

    Builds a structured deep-reading card for one scientific paper, covering methods, how experiments support claims, limitations and research ideas, with a script to prepare the source.

    47k GitHub starsUsed in 2 repos~2.1k tokens
    Research & ScienceAuto-check passed

  • Content Research Writer

    weapp-tailwindcss/weapp-tailwindcss

    Assists in writing high-quality content by conducting research, adding citations, improving hooks, iterating on outlines, and providing real-time feedback on each section.

    1.9k GitHub starsUsed in 25 repos~3.5k tokens
    Research & ScienceAuto-check passed

  • Last30days

    mvanhorn/last30days-skill

    Research what people actually say about any topic in the last 30 days.

    64k GitHub stars~7.9k tokensUpdated yesterday
    Research & ScienceAuto-check: notes

  • Peer Review

    spacering-net/codeg

    Structured manuscript/grant review with checklist-based evaluation.

    3.9k GitHub starsUsed in 17 repos~5.9k tokens
    Research & ScienceAuto-check: notes

More from harrisongzhang/TheVirtualBiotech

  • Evidence Citation

    harrisongzhang/TheVirtualBiotech

    How to report findings so they can be cited — writing evidence to files before asserting it, choosing where outputs belong, describing what each artifact shows, and returning findings with…

    122 GitHub stars~1.4k tokensUpdated 24 days ago
    Auto-check passed

  • Single Cell Analysis

    harrisongzhang/TheVirtualBiotech

    Statistical analysis and reporting for single-cell RNA-seq data.

    122 GitHub stars~3.2k tokensUpdated 24 days ago
    Auto-check passed

  • Single Cell Data Prep Qc

    harrisongzhang/TheVirtualBiotech

    Single-cell RNA-seq data preparation and quality control pipeline.

    122 GitHub stars~2.9k tokensUpdated 24 days ago
    Auto-check passed

Categories

Questions about Run Organization

What does Run Organization do?

How to organise a session run so a human can audit it — directory layout, artifact naming, recording the analysis plan, filing claim-evidence objects, and the end-of-run checklist. Run Organization is an agent skill from harrisongzhang/TheVirtualBiotech. How to organise a session run so a human can audit it — directory layout, artifact naming, recording the analysis plan, filing claim-evidence objects, and the end-of-run checklist.

When should I use Run Organization?

Run Organization fits situations like: orchestrating specialists (the CSO role); whenever you are about to dispatch work; synthesise findings; you need to know where an artifact belongs.

How do I install Run Organization in Claude Code?

Run `npx skills add harrisongzhang/TheVirtualBiotech --skill run-organization -a claude-code`. Or copy the skill folder (.claude/skills/run-organization in harrisongzhang/TheVirtualBiotech) into .claude/skills/run-organization in your project. Claude Code loads it when a task matches its description.

How do I install Run Organization in Codex?

Run `npx skills add harrisongzhang/TheVirtualBiotech --skill run-organization -a codex`. Or copy the skill folder (.claude/skills/run-organization in harrisongzhang/TheVirtualBiotech) into .agents/skills/run-organization in your project. Codex loads it when a task matches its description.

Can I use Run Organization in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add harrisongzhang/TheVirtualBiotech --skill run-organization -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/run-organization, .gemini/skills/run-organization, .github/skills/run-organization and .opencode/skills/run-organization in your project.

What does Run Organization need to run?

SKILL.md names no scripts, command-line tools or credentials: Run Organization is instructions for the agent only. Our summary lists: Python 3.

Does Run Organization access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Run Organization safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Run Organization use?

Run Organization is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Run Organization use?

About 1.8k tokens (SKILL.md is roughly 7.2k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Run Organization?

Skills that share tags, products or a category with Run Organization: Hypothesis Generation (spacering-net/codeg, 3.9k stars), GitHub Deep Research (bytedance/deer-flow, 84k stars), Nature Paper Card (Yuan1z0825/nature-skills, 47k stars) and Content Research Writer (weapp-tailwindcss/weapp-tailwindcss, 1.9k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Run Organization?

harrisongzhang (a GitHub user) maintains it in harrisongzhang/TheVirtualBiotech, which has 122 GitHub stars. The repository holds 4 skills in this directory. The repository was last updated on September 17, 2026.

Source: harrisongzhang/TheVirtualBiotech on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.