agentsclimarketplace
Sign in

Plan research compute

Skill ProfesseurHaipeng/ai-research-preflight/skills/plan-research-compute

Turn a sanitized computational-research question, governed dataset manifest, abstract DAG, pinned tools, and bounded resource requests into a deterministic provider-neutral preflight report. Use when planning reproducible biological or scientific batch computation before execution, checking data governance and locality, validating resource envelopes, or preparing a non-submittable CWL/GA4GH WES handoff. Never use it to read raw data, design formulations or wet-lab procedures, launch cloud jobs, or claim scientific conclusions.From its SKILL.md

Install
npx -y skills add ProfesseurHaipeng/ai-research-preflight --skill plan-research-compute

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

2.7 KB, 461 tokens by cl100k_base, as published. Nobody here has run it

Plan Research Compute

Create a reviewable compute preflight without accessing data bytes or executing work.

Workflow

  1. Keep the request computational. Refuse formulation, wet-lab, clinical-decision, or high-risk dual-use work.
  2. Sanitize the request into the JSON shape in references/input-contract.md. Include metadata and logical URIs only; never include credentials, participant identifiers, or raw observations.
  3. Represent every step as an abstract operation. Do not write shell commands or executable workflow bodies.
  4. Pin every bound tool to an exact version, versioned TRS URI, and container digest. Leave an uncertain step unbound instead of inventing a tool.
  5. Provide low/high CPU, memory, temporary-storage, output-storage, and wall-time bounds. Mark unsupported estimates unbenchmarked.
  6. Resolve SKILL_DIR to the directory containing this SKILL.md, then run the deterministic preflight:
python3 "$SKILL_DIR/scripts/research_compute_plan.py" manifest.json --format text
python3 "$SKILL_DIR/scripts/research_compute_plan.py" manifest.json --format json
  1. Interpret status and exit behavior using references/report-contract.md. Resolve every error before handoff and review every warning.
  2. Preserve the no-launch boundary. The report may describe a future CWL v1.2 / GA4GH WES handoff, but it must keep submission disabled and contain no endpoint, credentials, or cloud job identifier.

Non-negotiable boundaries

  • Read only the supplied JSON manifest. Do not open dataset URIs or local data paths.
  • Stay offline. Do not call network services, cloud SDKs, workflow engines, or subprocesses.
  • Do not produce formulations, wet-lab instructions, medical decisions, or scientific conclusions.
  • Treat human, controlled, and private data as governed. Require authorization, institutional approval, a reviewer role, retention policy, and compatible compute region.
  • Treat resource figures as planning bounds, not a performance guarantee, quote, or feasibility result.
  • Require scientific-domain, compute-budget, and applicable data-governance review before any separate execution system is considered.

What ships with it: 4 files

50.0 KB alongside SKILL.md, 1 of them executable

agents/

references/

scripts/

Gives 0 of the 12 instructions most plan spec skills give in 461 tokens

Counted across 1,099 of the 1,860 authors here whose files we hold, read 2026-08-07

  • Ask one question at a timein 51 of 1099
  • Break plans into vertical slicesin 29 of 1099, across 11 files
  • Publish issues in dependency orderin 27 of 1099, across 9 files
  • Iterate until user approves the breakdownin 25 of 1099, across 7 files
  • Explore the repository to understand the codebase statein 24 of 1099, across 7 files
  • Use domain glossary vocabularyin 23 of 1099, across 5 files
  • Apply correct triage labels to published issuesin 23 of 1099, across 5 files
  • Prefer AFK slices over HITLin 22 of 1099, across 7 files
  • Write a specification before writing any codein 22 of 1099, across 14 files
  • Write failing tests before implementation codein 22 of 1099, across 20 files
  • Ask clarifying questions until requirements are concretein 21 of 1099, across 13 files
  • Respect existing architecture decision recordsin 20 of 1099, across 5 files

Said here and by no other author read

  • refuse non-computational or high-risk requests
  • sanitize the request into the JSON contract
  • include metadata and logical URIs only
  • represent every step as an abstract operation
  • pin every bound tool to an exact version
  • provide low and high resource bounds

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 326,144. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.