From research-workflow
Use when a research session produces meaningful results and you need durable manifests, payloads, plot scripts, and a completion note so later sessions can reason from artifacts instead of memory. Don't use for the in-the-moment command discipline (→ evidence-first-execution), the go/no-go close-out (→ verification-gate), or pinning the runtime environment itself (→ reproducible-environment-contract).
How this skill is triggered — by the user, by Claude, or both
Slash command
/research-workflow:artifact-first-reproducibilityThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
If the result matters, leave artifacts behind — enough to rebuild it without you. Use the ecosystem's existing homes, not ad-hoc files:
If the result matters, leave artifacts behind — enough to rebuild it without you. Use the ecosystem's existing homes, not ad-hoc files:
validation/validate_*.py — the standalone script that produced the result.validation/plots/*.png — the figures, regenerable from that script..claude-work/TASK_X.Y_COMPLETE.md — the completion note (results, plot references, pass/fail)..claude-work/TASK_X.Y_COMPLETE.md)A small committed JSON beside the outputs — paths and diagnostics, plus a link to the pinned environment:
{
"command": "uv run validation/validate_plummer.py --N 1000 --seed 42",
"config_hash": "sha256:9f2c…",
"key_diagnostics": {"virial_Q": 0.501, "abs_dE_over_E": 7.3e-5},
"outputs": {"plot": "validation/plots/plummer_virial.png",
"payload": "validation/data/plummer_N1000.npz"},
"env": "validation/env.json"
}
The manifest references the env.json / lockfile / seeds — it never restates them. Those are owned by reproducible-environment-contract; duplicating them here lets the two copies drift.
.npz vs HDF5. Match what the package already uses; the constraint is resolvable + committed, not the schema.evidence-first-execution / verification-gate — the discipline that produces what gets archived here.reproducible-environment-contract — owns the env.json/lockfile/seeds this manifest links to.decision-log-and-commits — where decision rationale and commits live.npx claudepluginhub drannarosen/research-workflow --plugin research-workflowGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.