openJiuwen-ai/sciencediscovery / skills/antibody-design
antibody-design
UnverifiedPrepare, launch, monitor, and summarize the real RFdiffusion to ProteinMPNN to Protenix antibody pipeline on a local or remote ScienceDiscovery Runner with sandboxed Ascend NPUs.
| File | Kind | Size |
|---|---|---|
| AGENT_BOOTSTRAP.md | markdown | 1.9 KB |
| SKILL.md | skill md | 11.5 KB |
| references/real_pipeline_config.example.json | config | 506 B |
| requirements.txt | manifest | 863 B |
| scripts/antibody_pipeline_manager.py | script | 47.8 KB |
| scripts/check_real_pipeline_env.sh | script | 11.3 KB |
| scripts/install_hmmer_for_protenix_msa.sh | script | 4.6 KB |
| scripts/install_pipeline_deps.sh | script | 7.7 KB |
| scripts/pdb_to_protenix_json.py | script | 4.0 KB |
| scripts/protenix_py312_compat.py | script | 2.0 KB |
| scripts/rfdiffusion_mindspore_io.patch | other | 7.4 KB |
| scripts/run_after_rfdiffusion.sh | script | 20.4 KB |
| scripts/run_full_antibody_pipeline.sh | script | 17.1 KB |
| scripts/run_sandbox_pipeline.sh | script | 2.3 KB |
| scripts/sandbox_npu_runtime.sh | script | 1.4 KB |
| scripts/screen_protenix_results.py | script | 18.3 KB |
| scripts/test_antibody_pipeline_manager.py | script | 20.3 KB |
| scripts/test_validate_managed_environment.py | script | 3.1 KB |
| scripts/validate_managed_environment.py | script | 4.2 KB |
scripts/run_full_antibody_pipeline.sh
View on GitHub#!/usr/bin/env bash
# Copyright (C) 2026-2026 Huawei Technologies Co., Ltd
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
set -euo pipefail
usage() {
cat <<'EOF'
Usage:
run_full_antibody_pipeline.sh [options]
Run the full antibody design pipeline:
RFdiffusion -> ProteinMPNN -> MindScience Protenix -> preliminary screening
The target antigen PDB, antibody framework PDB, and hotspot list are explicit
scientific inputs. This script never falls back to bundled example PDB files.
Options:
--run-dir DIR Pipeline run directory. Default: APP_DIR/pipeline_runs/antibody_protenix_<N>_<timestamp>
--num-designs N Total designs to generate. Default: 600
--app-dir DIR MindSPONGE applications directory. Default: MINDSCIENCE_APP_DIR or derived from ANTIBODY_PIPELINE_HOME
--scripts-dir DIR Helper scripts directory. Default: directory containing this script
--python PYTHON Python executable from a ScienceDiscovery scientific-env revision. Required
--npus LIST Ascend physical device list. Default: 0,1,2,3,4,5,6,7
--workers-per-npu N RFdiffusion workers per NPU. Default: 2
--target-pdb PDB Target antigen PDB. Required.
--framework-pdb PDB Antibody framework PDB. Required.
--ckpt PATH RFdiffusion checkpoint path.
--hotspots LIST RFdiffusion hotspot list using target-PDB chain labels. Required.
--design-loops LIST RFdiffusion design loops. Default: [H1:8,H2:6,H3:16]
--final-step N RFdiffusion inference.final_step. Default: 160
--diffuser-t N RFdiffusion diffuser.T. Default: 200
--protenix-dir DIR MindScience Protenix application directory. Default: APP_DIR/protenix or PROTENIX_APP_DIR
--protenix-ckpt FILE Protenix checkpoint. Default: PROTENIX_CKPT or PROTENIX_DIR/release_data/checkpoint/ms_model_v0.5.0.ckpt
--protenix-use-msa BOOL Pass through to inference.py --use_msa. Default: PROTENIX_USE_MSA or false
--protenix-n-sample N Pass through to inference.py --n_sample. Default: PROTENIX_N_SAMPLE or 1
--protenix-seeds LIST Pass through to inference.py --seeds. Default: PROTENIX_SEEDS or 42
--hmmer-home DIR HMMER home for Protenix MSA. Default: HMMER_HOME or ANTIBODY_PIPELINE_HOME/tools/hmmer
--pipeline-env FILE Optional extra env file. Disabled by default in ScienceDiscovery managed-env mode; do not use host env.sh
--cann-set-env FILE Optional legacy Ascend set_env.sh. Sandbox execution leaves this unset.
--force Remove existing pipeline outputs under --run-dir before rebuilding.
-h, --help Show this help.
EOF
}
PIPELINE_HOME="${ANTIBODY_PIPELINE_HOME:-}"
APP_DIR="${MINDSCIENCE_APP_DIR:-}"
if [[ -z "$APP_DIR" && -n "$PIPELINE_HOME" ]]; then
APP_DIR="$PIPELINE_HOME/models/mindscience/MindSPONGE/applications"
fi
APP_DIR="${APP_DIR:-/opt/antibody_pipeline/models/mindscience/MindSPONGE/applications}"
SCRIPTS_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
resolve_python() {
local candidate
for candidate in \
"${SCIENCE_AGENT_MANAGED_PYTHON:-}" \
"${PYTHON_BIN:-}" \
"${ANTIBODY_PIPELINE_PYTHON:-}"; do
if [[ -n "$candidate" && -x "$candidate" ]]; then
printf '%s\n' "$candidate"
return 0
fi
done
printf '%s\n' ""
}
truthy_env() {
case "${1,,}" in
0|false|no|off) return 1 ;;
*) return 0 ;;
esac
}
require_managed_python() {
if [[ -z "${PYTHON_BIN:-}" || ! -x "$PYTHON_BIN" ]]; then
echo "error: --python must be a valid ScienceDiscovery scientific-env revision Python" >&2
echo " expected: /opt/science-env/bin/python or a scientific-env revision Python" >&2
exit 2
fi
if [[ -n "${PIPELINE_HOME:-}" ]]; then
case "$PYTHON_BIN" in
"$PIPELINE_HOME"/venv/*|"$PIPELINE_HOME"/python_user/*|"$PIPELINE_HOME"/bin/*)
echo "error: refusing host pipeline Python: $PYTHON_BIN" >&2
echo " create/select a ScienceDiscovery scientific environment and pass its bin/python" >&2
exit 2
;;
esac
fi
if truthy_env "${ANTIBODY_REQUIRE_SCIENCEDISCOVERY_ENV:-${ANTIBODY_REQUIRE_SCIENCEAGENT_ENV:-1}}"; then
case "$PYTHON_BIN" in
/opt/science-env/bin/python|/opt/science-env/bin/python3|/opt/science-env/bin/python3.*|*/scientific-envs/revisions/*/bin/python|*/scientific-envs/revisions/*/bin/python3|*/scientific-envs/revisions/*/bin/python3.*) ;;
*)
echo "error: Python is not from a ScienceDiscovery scientific environment revision: $PYTHON_BIN" >&2
echo " expected: /opt/science-env/bin/python or a scientific-env revision Python" >&2
exit 2
;;
esac
fi
}
require_no_host_pipeline_env() {
if truthy_env "${ANTIBODY_REQUIRE_SCIENCEDISCOVERY_ENV:-${ANTIBODY_REQUIRE_SCIENCEAGENT_ENV:-1}}" && [[ -n "${PIPELINE_ENV:-}" ]]; then
echo "error: refusing --pipeline-env in ScienceDiscovery managed-env mode: $PIPELINE_ENV" >&2
echo " do not source host env.sh; install Python deps into the ScienceDiscovery scientific environment" >&2
exit 2
fi
}
PYTHON_BIN="$(resolve_python)"
RUN_DIR=""
NUM_DESIGNS=600
NPUS="0,1,2,3,4,5,6,7"
WORKERS_PER_NPU=2
TARGET_PDB=""
FRAMEWORK_PDB=""
RF_CKPT="./models/RFdiffusion_Ab.ckpt"
HOTSPOTS=""
DESIGN_LOOPS="[H1:8,H2:6,H3:16]"
FINAL_STEP=160
DIFFUSER_T=200
PROTENIX_DIR="${PROTENIX_APP_DIR:-}"
PROTENIX_CKPT="${PROTENIX_CKPT:-}"
PROTENIX_USE_MSA="${PROTENIX_USE_MSA:-false}"
PROTENIX_N_SAMPLE="${PROTENIX_N_SAMPLE:-1}"
PROTENIX_SEEDS="${PROTENIX_SEEDS:-42}"
HMMER_HOME="${HMMER_HOME:-}"
CANN_SET_ENV="${CANN_SET_ENV:-}"
PIPELINE_ENV="${ANTIBODY_PIPELINE_ENV:-}"
FORCE=0
# ScienceDiscovery managed-env mode does not auto-source host pipeline env.sh.
while [[ $# -gt 0 ]]; do
case "$1" in
--run-dir) RUN_DIR="$2"; shift 2 ;;
--num-designs) NUM_DESIGNS="$2"; shift 2 ;;
--app-dir) APP_DIR="$2"; shift 2 ;;
--scripts-dir) SCRIPTS_DIR="$2"; shift 2 ;;
--python) PYTHON_BIN="$2"; shift 2 ;;
--npus) NPUS="$2"; shift 2 ;;
--workers-per-npu) WORKERS_PER_NPU="$2"; shift 2 ;;
--target-pdb) TARGET_PDB="$2"; shift 2 ;;
--framework-pdb) FRAMEWORK_PDB="$2"; shift 2 ;;
--ckpt) RF_CKPT="$2"; shift 2 ;;
--hotspots) HOTSPOTS="$2"; shift 2 ;;
--design-loops) DESIGN_LOOPS="$2"; shift 2 ;;
--final-step) FINAL_STEP="$2"; shift 2 ;;
--diffuser-t) DIFFUSER_T="$2"; shift 2 ;;
--protenix-dir) PROTENIX_DIR="$2"; shift 2 ;;
--protenix-ckpt|--checkpoint) PROTENIX_CKPT="$2"; shift 2 ;;
--protenix-use-msa) PROTENIX_USE_MSA="$2"; shift 2 ;;
--protenix-n-sample) PROTENIX_N_SAMPLE="$2"; shift 2 ;;
--protenix-seeds) PROTENIX_SEEDS="$2"; shift 2 ;;
--hmmer-home) HMMER_HOME="$2"; shift 2 ;;
--pipeline-env) PIPELINE_ENV="$2"; shift 2 ;;
--cann-set-env) CANN_SET_ENV="$2"; shift 2 ;;
--force) FORCE=1; shift ;;
-h|--help) usage; exit 0 ;;
*) echo "error: unknown option: $1" >&2; usage >&2; exit 2 ;;
esac
done
abs_path() {
case "$1" in
/*) printf '%s\n' "$1" ;;
*) readlink -m "$PWD/$1" ;;
esac
}
if ! [[ "$NUM_DESIGNS" =~ ^[0-9]+$ ]] || [[ "$NUM_DESIGNS" -lt 1 ]]; then
echo "error: --num-designs must be a positive integer" >&2
exit 2
fi
RF_DIFFUSION_DIR="$APP_DIR/rf_diffusion"
PROTENIX_DIR="${PROTENIX_DIR:-$APP_DIR/protenix}"
PROTENIX_CKPT="${PROTENIX_CKPT:-$PROTENIX_DIR/release_data/checkpoint/ms_model_v0.5.0.ckpt}"
if [[ -z "$TARGET_PDB" ]]; then
echo "error: --target-pdb is required; ask the user to provide a workspace-local antigen PDB" >&2
exit 2
fi
if [[ -z "$FRAMEWORK_PDB" ]]; then
echo "error: --framework-pdb is required; ask the user to provide a workspace-local antibody framework PDB" >&2
exit 2
fi
if [[ -z "$HOTSPOTS" ]]; then
echo "error: --hotspots is required; ask the user for chain-labelled target residues such as [B45,B46,B49]" >&2
exit 2
fi
TARGET_PDB="$(abs_path "$TARGET_PDB")"
FRAMEWORK_PDB="$(abs_path "$FRAMEWORK_PDB")"
RF_CKPT="$(abs_path "$RF_CKPT")"
if [[ -z "$HMMER_HOME" && -n "$PIPELINE_HOME" ]]; then
HMMER_HOME="$PIPELINE_HOME/tools/hmmer"
fi
HMMER_HOME="${HMMER_HOME:-}"
require_no_host_pipeline_env
if [[ -n "$PIPELINE_ENV" && -f "$PIPELINE_ENV" ]]; then
set +u
# shellcheck disable=SC1090
source "$PIPELINE_ENV" >/dev/null 2>&1 || echo "warning: pipeline env returned non-zero: $PIPELINE_ENV" >&2
set -u
fi
PIPELINE_HOME="${ANTIBODY_PIPELINE_HOME:-$PIPELINE_HOME}"
if [[ ! -x "$PYTHON_BIN" ]]; then
PYTHON_BIN="$(resolve_python)"
fi
require_managed_python
if [[ -z "$RUN_DIR" ]]; then
RUN_DIR="$APP_DIR/pipeline_runs/antibody_protenix_${NUM_DESIGNS}_$(date +%Y%m%d_%H%M%S)"
fi
RUN_DIR="$(abs_path "${RUN_DIR%/}")"
SCRIPTS_DIR="$(abs_path "$SCRIPTS_DIR")"
RF_OUT="$RUN_DIR/01_rfdiffusion"
LOG_DIR="$RUN_DIR/logs"
mkdir -p "$RF_OUT" "$LOG_DIR"
AFTER_SCRIPT="$SCRIPTS_DIR/run_after_rfdiffusion.sh"
if [[ ! -f "$AFTER_SCRIPT" ]]; then
echo "error: missing downstream wrapper: $AFTER_SCRIPT" >&2
exit 2
fi
count_top_pdbs() {
find "$RF_OUT" -maxdepth 1 -type f -name "output_*.pdb" | wc -l
}
run_rfdiffusion_inline() {
local out_dir="$1"
local total_designs="$2"
local shard_root="${out_dir}_shards"
local ascend_device_env="${ASCEND_DEVICE_ENV:-ASCEND_RT_VISIBLE_DEVICES}"
local rfdiffusion_script="run_inference.py"
IFS=',' read -r -a npu_array <<< "$NPUS"
if [[ "${#npu_array[@]}" -lt 1 ]]; then
echo "error: NPUS is empty" >&2
exit 2
fi
if ! [[ "$WORKERS_PER_NPU" =~ ^[0-9]+$ ]] || [[ "$WORKERS_PER_NPU" -lt 1 ]]; then
echo "error: WORKERS_PER_NPU must be a positive integer" >&2
exit 2
fi
mkdir -p "$out_dir/logs" "$shard_root"
export PYTHONPATH="$APP_DIR/../..:$RF_DIFFUSION_DIR/env:${PYTHONPATH:-}"
local total_workers=$((${#npu_array[@]} * WORKERS_PER_NPU))
if [[ "$total_workers" -gt "$total_designs" ]]; then
total_workers="$total_designs"
fi
echo "RFdiffusion inline parallel settings"
echo " RF_DIFFUSION_DIR : $RF_DIFFUSION_DIR"
echo " OUT_DIR : $out_dir"
echo " TOTAL_DESIGNS : $total_designs"
echo " NPUS : $NPUS"
echo " WORKERS_PER_NPU : $WORKERS_PER_NPU"
echo " TOTAL_WORKERS : $total_workers"
echo " PYTHON_BIN : $PYTHON_BIN"
echo " SHARD_ROOT : $shard_root"
echo " TARGET_PDB : $TARGET_PDB"
echo " FRAMEWORK_PDB : $FRAMEWORK_PDB"
echo " RF_CKPT : $RF_CKPT"
echo " HOTSPOTS : $HOTSPOTS"
echo " DESIGN_LOOPS : $DESIGN_LOOPS"
echo " FINAL_STEP : $FINAL_STEP"
echo " DIFFUSER_T : $DIFFUSER_T"
cd "$RF_DIFFUSION_DIR"
local pids=()
local worker=0
local remaining="$total_designs"
local npu slot workers_left shard_count shard_dir shard_prefix log_path
for npu in "${npu_array[@]}"; do
for slot in $(seq 1 "$WORKERS_PER_NPU"); do
if [[ "$worker" -ge "$total_workers" ]]; then
break 2
fi
workers_left=$((total_workers - worker))
shard_count=$(((remaining + workers_left - 1) / workers_left))
shard_dir="$shard_root/shard_$(printf "%03d" "$worker")"
shard_prefix="$shard_dir/output"
log_path="$out_dir/logs/rfdiffusion_shard_$(printf "%03d" "$worker")_npu${npu}.log"
mkdir -p "$shard_dir"
echo "[launch] worker=$worker npu=$npu designs=$shard_count log=$log_path"
(
export "$ascend_device_env=$npu"
if [[ "$ascend_device_env" == "DEVICE_ID" ]]; then
export DEVICE_ID="$npu"
else
export DEVICE_ID="0"
fi
"$PYTHON_BIN" "$rfdiffusion_script" \
--config-name antibody \
"inference.output_prefix=$shard_prefix" \
"inference.num_designs=$shard_count" \
"inference.ckpt_override_path=$RF_CKPT" \
"antibody.target_pdb=$TARGET_PDB" \
"antibody.framework_pdb=$FRAMEWORK_PDB" \
"ppi.hotspot_res=$HOTSPOTS" \
"antibody.design_loops=$DESIGN_LOOPS" \
"inference.final_step=$FINAL_STEP" \
"diffuser.T=$DIFFUSER_T"
) >"$log_path" 2>&1 &
pids+=("$!")
worker=$((worker + 1))
remaining=$((remaining - shard_count))
done
done
local fail=0
local pid
for pid in "${pids[@]}"; do
if ! wait "$pid"; then
fail=1
fi
done
if [[ "$fail" -ne 0 ]]; then
echo "error: one or more RFdiffusion shards failed; inspect $out_dir/logs" >&2
exit 1
fi
local idx=0
local pdb base trb new_base
while IFS= read -r -d '' pdb; do
base="${pdb%.pdb}"
trb="${base}.trb"
new_base="$out_dir/output_$(printf "%06d" "$idx")"
cp "$pdb" "${new_base}.pdb"
if [[ -f "$trb" ]]; then
cp "$trb" "${new_base}.trb"
fi
idx=$((idx + 1))
done < <(find "$shard_root" -mindepth 2 -maxdepth 2 -type f -name "output_*.pdb" -print0 | sort -z)
if [[ "$idx" -ne "$total_designs" ]]; then
echo "error: collected $idx PDB files, expected $total_designs" >&2
exit 1
fi
echo "Collected $idx RFdiffusion designs into $out_dir"
}
write_artifact_manifest() {
local pipeline_workspace
local manifest
pipeline_workspace="$(dirname "$(dirname "$RUN_DIR")")"
manifest="$pipeline_workspace/artifact_manifest.txt"
mkdir -p "$pipeline_workspace"
: >"$manifest"
add_file() {
local path="$1"
if [[ -f "$path" ]]; then
realpath --relative-to="$pipeline_workspace" "$path" >>"$manifest"
fi
}
add_tree() {
local dir="$1"
local pattern="$2"
if [[ -d "$dir" ]]; then
find "$dir" -type f -name "$pattern" -print0 | sort -z | while IFS= read -r -d '' path; do
realpath --relative-to="$pipeline_workspace" "$path" >>"$manifest"
done
fi
}
add_file "$RUN_DIR/05_screening/protenix_screening_report.md"
add_file "$RUN_DIR/05_screening/protenix_screening_summary.csv"
add_tree "$RUN_DIR/04_protenix_output" "*.cif"
sort -u "$manifest" -o "$manifest"
echo "Artifact manifest: $manifest"
echo "Artifact count: $(wc -l < "$manifest")"
}
start_ts=$(date +%s)
echo "Full antibody Protenix pipeline"
echo " RUN_DIR : $RUN_DIR"
echo " NUM_DESIGNS : $NUM_DESIGNS"
echo " APP_DIR : $APP_DIR"
echo " SCRIPTS_DIR : $SCRIPTS_DIR"
echo " PYTHON_BIN : $PYTHON_BIN"
echo " NPUS : $NPUS"
echo " WORKERS_PER_NPU : $WORKERS_PER_NPU"
echo " TARGET_PDB : $TARGET_PDB"
echo " FRAMEWORK_PDB : $FRAMEWORK_PDB"
echo " RF_CKPT : $RF_CKPT"
echo " HOTSPOTS : $HOTSPOTS"
echo " DESIGN_LOOPS : $DESIGN_LOOPS"
echo " FINAL_STEP : $FINAL_STEP"
echo " DIFFUSER_T : $DIFFUSER_T"
echo " PROTENIX_DIR : $PROTENIX_DIR"
echo " PROTENIX_CKPT : $PROTENIX_CKPT"
echo " PROTENIX_USE_MSA : $PROTENIX_USE_MSA"
echo " PROTENIX_N_SAMPLE : $PROTENIX_N_SAMPLE"
echo " PROTENIX_SEEDS : $PROTENIX_SEEDS"
echo " HMMER_HOME : $HMMER_HOME"
echo " PIPELINE_ENV : $PIPELINE_ENV"
echo " CANN_SET_ENV : $CANN_SET_ENV"
echo
rf_count=$(count_top_pdbs)
if [[ "$FORCE" == "1" && "$rf_count" -ne 0 ]]; then
rm -rf "$RF_OUT" "${RF_OUT}_shards"
mkdir -p "$RF_OUT"
rf_count=0
fi
if [[ "$rf_count" -eq "$NUM_DESIGNS" ]]; then
echo "[skip] RFdiffusion already has $rf_count top-level PDB files"
else
if [[ "$rf_count" -ne 0 ]]; then
echo "error: RFdiffusion output has $rf_count PDB files, expected $NUM_DESIGNS." >&2
echo " Use a fresh --run-dir, or finish/clean the partial RF run first." >&2
exit 1
fi
echo "[$(date '+%F %T')] Step 1/2: RFdiffusion"
echo " log: $LOG_DIR/01_rfdiffusion.log"
run_rfdiffusion_inline "$RF_OUT" "$NUM_DESIGNS" >"$LOG_DIR/01_rfdiffusion.log" 2>&1
rf_count=$(count_top_pdbs)
if [[ "$rf_count" -ne "$NUM_DESIGNS" ]]; then
echo "error: RFdiffusion produced $rf_count PDB files, expected $NUM_DESIGNS" >&2
echo " Check $LOG_DIR/01_rfdiffusion.log" >&2
exit 1
fi
fi
echo "[$(date '+%F %T')] Step 2/2: ProteinMPNN + Protenix + screening"
echo " log: $LOG_DIR/02_after_rfdiffusion.log"
after_cmd=(
bash "$AFTER_SCRIPT"
--run-dir "$RUN_DIR"
--rf-dir "$RF_OUT"
--target-pdb "$TARGET_PDB"
--num-designs "$NUM_DESIGNS"
--app-dir "$APP_DIR"
--scripts-dir "$SCRIPTS_DIR"
--python "$PYTHON_BIN"
--protenix-dir "$PROTENIX_DIR"
--protenix-ckpt "$PROTENIX_CKPT"
--protenix-use-msa "$PROTENIX_USE_MSA"
--protenix-n-sample "$PROTENIX_N_SAMPLE"
--protenix-seeds "$PROTENIX_SEEDS"
--hmmer-home "$HMMER_HOME"
--pipeline-env "$PIPELINE_ENV"
--cann-set-env "$CANN_SET_ENV"
--npus "$NPUS"
--hotspots "$HOTSPOTS"
)
if [[ "$FORCE" == "1" ]]; then
after_cmd+=(--force)
fi
"${after_cmd[@]}" >"$LOG_DIR/02_after_rfdiffusion.log" 2>&1
write_artifact_manifest
end_ts=$(date +%s)
elapsed=$((end_ts - start_ts))
echo
echo "Done."
echo "Elapsed seconds: $elapsed"
echo "Run dir: $RUN_DIR"
echo "Reports:"
echo " $RUN_DIR/05_screening/protenix_screening_report.md"
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deployment | Stars: 323k | Updated | Claude CodeCodex | Verified | Install |
| xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like "the xlsx in my downloads") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved. | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |