← Back to Skills Marketplace
ashipiling

DrugFlow

by ashipiling · GitHub ↗ · v1.0.2 · MIT-0
cross-platform ✓ Security Clean
187
Downloads
0
Stars
0
Active Installs
3
Versions
Install in OpenClaw
/install drugflow-api
Description
Multi-flow API workflow skill for this DrugFlow Django repository. Use when an agent needs executable end-to-end API procedures such as login/register, works...
README (SKILL.md)

DrugFlow Skills

Route requests to the correct DrugFlow API flow and execute with minimal ambiguity.

Flow Selection

  1. Read references/index.md first.
  2. Match user intent to one flow.
  3. Load only that flow's reference files.
  4. Prefer script execution from scripts/\x3Cflow>/ when available.

Current Flows

  1. Common APIs: reusable auth/workspace/balance/jobs APIs available.
  2. Virtual screening: complete flow available.
  3. Docking: complete flow available.
  4. ADMET: complete flow available.
  5. Rescoring: complete flow available.
  6. Structure extract: complete flow available (img2mol backend type).
  7. Molecular factory: complete flow available (with atom-selection helpers).

Common APIs Workflow

  1. Read references/flows/common-apis/call-flow.md.
  2. Read references/flows/common-apis/payloads.md.
  3. Reuse scripts/common/drugflow_api.py for:
  • signin
  • signup
  • list_workspaces / create_workspace / ensure_workspace
  • get_balance
  • list_jobs
  1. Use scripts/common/test_common_apis.py for direct smoke tests.

Virtual Screening Workflow

  1. Read references/flows/virtual-screening/call-flow.md.
  2. Read references/flows/virtual-screening/payloads.md.
  3. Use scripts/virtual-screening/run_vs_flow.py for end-to-end execution.
  4. Always include ws_id for /api/jobs list/detail.
  5. For /api/jobs create, pass name, type, args (JSON string), ws_id; in non-private mode include expect_tokens and avail_tokens.

Docking Workflow

  1. Read references/flows/docking/call-flow.md.
  2. Read references/flows/docking/payloads.md.
  3. Use scripts/docking/run_docking_flow.py for end-to-end execution.
  4. Create docking jobs through POST /api/jobs with multipart fields pdb, ligands, pdb_content, and args.
  5. Site-driven docking box note: when --site is provided but center/size/radius are omitted, the script auto-derives the docking box from that site in local PDB.
  6. Always include ws_id on job list/detail requests and pass expect_tokens/avail_tokens in non-private mode.

ADMET Workflow

  1. Read references/flows/admet/call-flow.md.
  2. Read references/flows/admet/payloads.md.
  3. Use scripts/admet/run_admet_flow.py for end-to-end execution.
  4. ADMET job type is fixed to admet-dl.
  5. Support two input modes:
  • direct smiles list
  • dataset mode via dataset_id + smiles_col
  1. For /api/jobs create, pass name, type=admet-dl, args, ws_id, and in non-private mode expect_tokens/avail_tokens.

Rescoring Workflow

  1. Read references/flows/rescoring/call-flow.md.
  2. Read references/flows/rescoring/payloads.md.
  3. Use scripts/rescoring/run_rescoring_flow.py for end-to-end execution.
  4. Create rescoring jobs through POST /api/jobs with:
  • type=rescoring
  • form fields pdb, ligands, smiles_col
  • args.mode=semi and args.rescoring_functions
  1. Script enforces input files: --pdb-file must be .pdb, --ligands-file must be .sdf.
  2. Always include ws_id; in non-private mode include expect_tokens and avail_tokens.

Structure Extract Workflow

  1. Read references/flows/structure-extract/call-flow.md.
  2. Read references/flows/structure-extract/payloads.md.
  3. Use scripts/structure-extract/run_structure_extract_flow.py for end-to-end execution.
  4. User-facing "结构提取" maps to backend job type=img2mol.
  5. For create, pass name, type=img2mol, args (dataset_id, page_list), ws_id, and in non-private mode expect_tokens/avail_tokens.
  6. dataset_id must be img2mol-compatible and include extras.osskey.

Molecular Factory Workflow

  1. Read references/flows/molecular-factory/call-flow.md.
  2. Read references/flows/molecular-factory/payloads.md.
  3. Use scripts/molecular-factory/run_molecular_factory_flow.py:
  • atom-info
  • extract-partial
  • draw-atom-index
  • create-job
  1. Default to non-docking molecular factory unless user explicitly asks for docking:
  • args.need_docking=false
  • args.pdb_use.*=false
  1. Default generation models:
  • args.molgen_algos=["Frag-GPT","REINVENT"]
  1. Use helper APIs first to confirm selected_atoms/start_atoms, then submit molecular_factory job.
  2. Always pass ws_id; in non-private mode include expect_tokens and avail_tokens.

Output Contract

  1. Return method + endpoint + required parameters for each step.
  2. Return key ids and state: ws_id, job_id, state, result count.
  3. When running scripts, return command + important outputs.

Expansion Rules

  1. Add new flow docs under references/flows/\x3Cflow>/ with call-flow.md and payloads.md.
  2. Add runnable scripts under scripts/\x3Cflow>/.
  3. Update references/index.md and this file's Current Flows section.

References

  1. references/index.md
  2. references/flows/common-apis/call-flow.md
  3. references/flows/common-apis/payloads.md
  4. references/flows/virtual-screening/call-flow.md
  5. references/flows/virtual-screening/payloads.md
  6. references/flows/docking/call-flow.md
  7. references/flows/docking/payloads.md
  8. references/flows/admet/call-flow.md
  9. references/flows/admet/payloads.md
  10. references/flows/molecular-factory/call-flow.md
  11. references/flows/molecular-factory/payloads.md
  12. references/flows/rescoring/call-flow.md
  13. references/flows/rescoring/payloads.md
  14. references/flows/structure-extract/call-flow.md
  15. references/flows/structure-extract/payloads.md
Usage Guidance
This skill appears to do what it claims: run DrugFlow API flows and upload/submit files to a DrugFlow server. Before installing or invoking it, verify the source (there's no homepage and the owner is unknown), and only point scripts to a trusted base_url. Be aware that running flows will read local PDB/ligand/smiles files and transmit them (and will send supplied email/password) to the configured server — avoid running it against untrusted endpoints or with sensitive files/credentials. If you need higher assurance, review the repository locally (the included Python files are human-readable) and run the scripts in an isolated environment or container.
Capability Analysis
Type: OpenClaw Skill Name: drugflow-api Version: 1.0.2 The drugflow-api skill bundle provides a comprehensive toolkit for an AI agent to interact with the DrugFlow platform for molecular docking, virtual screening, and ADMET analysis. The bundle includes a shared API client (scripts/common/drugflow_api.py) and specialized execution scripts (e.g., scripts/docking/run_docking_flow.py, scripts/admet/run_admet_flow.py) that handle authentication, file uploads, and job polling. The behavior is well-documented in the references/ directory and SKILL.md, and all network and file operations are strictly aligned with the stated purpose of performing drug discovery workflows. No evidence of data exfiltration, malicious prompt injection, or unauthorized persistence was found.
Capability Assessment
✓ Purpose & Capability
The name/description (end-to-end DrugFlow API flows) matches the included scripts and reference docs. The scripts perform sign-in, workspace/balance/job operations, create jobs, upload datasets and poll results — all expected for this purpose.
ℹ Instruction Scope
SKILL.md and the scripts direct the agent to run flow scripts that will read local files (PDB, ligands, smiles files) and send them to the provided base_url. This behavior is coherent with docking/structure/VS flows, but it means the agent will access and transmit local files when asked to run those flows.
✓ Install Mechanism
No install spec; the skill ships Python scripts only. No external installers, downloads, or package installs are requested by the skill metadata.
ℹ Credentials
The skill does not request environment variables or platform credentials. It expects runtime parameters (base_url, email, password) passed to scripts — which is appropriate for authenticating to a DrugFlow deployment, but those credentials will be transmitted to whatever base_url is supplied, so the target URL must be trusted.
✓ Persistence & Privilege
always is false and there is no installation hook or config mutation. The skill does network I/O and file I/O when executed but does not request elevated system privileges or persistent autorun.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install drugflow-api
  3. After installation, invoke the skill by name or use /drugflow-api
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.2
- Added default behavior for molecular factory: runs non-docking jobs unless docking is explicitly requested (`args.need_docking=false`, `args.pdb_use.*=false`). - Set default generation models for molecular factory: `args.molgen_algos=["Frag-GPT","REINVENT"]`. - No code changes; documentation updates only.
v1.0.1
Expanded with four new end-to-end API flows: ADMET, rescoring, structure extraction, and molecular factory. - Added detailed references and payload docs for ADMET, rescoring, structure extract, and molecular factory flows. - Added runnable scripts for each new flow under `scripts/<flow>/`. - Updated documentation to cover full workflows, input parameters, and script usage for all new flows. - Extended the main skill description and current flows list to include these capabilities.
v1.0.0
init
Metadata
Slug drugflow-api
Version 1.0.2
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 3
Frequently Asked Questions

What is DrugFlow?

Multi-flow API workflow skill for this DrugFlow Django repository. Use when an agent needs executable end-to-end API procedures such as login/register, works... It is an AI Agent Skill for Claude Code / OpenClaw, with 187 downloads so far.

How do I install DrugFlow?

Run "/install drugflow-api" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is DrugFlow free?

Yes, DrugFlow is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does DrugFlow support?

DrugFlow is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created DrugFlow?

It is built and maintained by ashipiling (@ashipiling); the current version is v1.0.2.

More Skills
self-improving agent
Captures learnings, errors, and corrections to enable continuous improvement. Use when: (1) A command or operation fails unexpectedly, (2) User corrects Clau...
⬇ 463283 · by pskoett
Skill Vetter
Security-first skill vetting for AI agents. Use before installing any skill from ClawdHub, GitHub, or other sources. Checks for red flags, permission scope, and suspicious patterns.
⬇ 259410 · by spclaudehome
Polymarket
Query Polymarket prediction markets. Check odds, find trending markets, search events, track price movements.
⬇ 232503 · by joelchance
Self-Improving + Proactive Agent
Self-reflection + Self-criticism + Self-learning + Self-organizing memory. Agent evaluates its own work, catches mistakes, and improves permanently. Use when...
⬇ 200423 · by ivangdavila
Github
Interact with GitHub using the `gh` CLI. Use `gh issue`, `gh pr`, `gh run`, and `gh api` for issues, PRs, CI runs, and advanced queries.
⬇ 191115 · by steipete
ontology
Typed knowledge graph for structured agent memory and composable skills. Use when creating/querying entities (Person, Project, Task, Event, Document), linkin...
⬇ 190156 · by oswalpalash

💬 Comments