Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add pregHosh/Solitarius-mcp --skill rlgit clone --depth 1 https://github.com/pregHosh/Solitarius-mcpWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/preghosh/solitarius-mcp/rl)<a href="https://agentmods.dev/skills/preghosh/solitarius-mcp/rl"><img src="https://agentmods.dev/badge/skills/preghosh/solitarius-mcp/rl.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00039 | $0.02301 |
| Opus 5 | $0.00019 | $0.01151 |
| Sonnet 5 | $0.00008 | $0.00460 |
| Haiku 4.5 | $0.00004 | $0.00230 |
Grade A, and why
rl scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 6d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 251 lines — stays where its author put it; the contents beside it link to each section on GitHub.
REINVENT4 Reinforcement Learning
Run staged reinforcement learning by writing a TOML config and running the reinvent CLI.
No MCP server needed — activate the reinvent4 conda env first.
Workflow
RL requires interactive scoring function design BEFORE writing the config. Follow this protocol.
1. Resolve paths
readlink -f <relative_path>
All paths in the TOML must be absolute.
2. Scoring function design (MANDATORY)
Walk the user through these steps interactively — do NOT skip ahead.
Step 1 — Identify objectives
Ask: "What properties must your molecules satisfy?" Classify each answer:
- Hard constraints (e.g., no reactive groups, must contain scaffold) →
custom_alertsorMatchingSubstructure— these are filters (zero total score on failure) - Soft objectives (e.g., MW 300-500, QED > 0.6) → scored components with transforms
Step 2 — Match objectives to components
Check what is available:
# List built-in scoring components
python -c "
import reinvent_plugins.components as rpc
import pkgutil
for m in pkgutil.iter_modules(rpc.__path__):
print(m.name)
"
Built-in components (always available):
| Property | Component type | Suggested transform |
|---|---|---|
| Drug-likeness | QED |
sigmoid, low=0.5, high=0.9, k=0.25 |
| LogP | SlogP |
double_sigmoid, low=-0.5, high=5.0, coef_div=5.0, coef_si=20.0, coef_se=20.0 |
| Molecular weight | MolecularWeight |
double_sigmoid, low=200, high=500, coef_div=500.0, coef_si=20.0, coef_se=20.0 |
| TPSA | TPSA |
reverse_sigmoid, low=60, high=140, k=0.1 |
| H-bond acceptors | HBondAcceptors |
reverse_sigmoid, low=5, high=10, k=0.5 |
| H-bond donors | HBondDonors |
reverse_sigmoid, low=3, high=5, k=0.5 |
| Rotatable bonds | NumRotBond |
reverse_sigmoid, low=5, high=10, k=0.5 |
| Synth. accessibility | SAScore |
reverse_sigmoid, low=2, high=6, k=0.5 |
| Fsp3 | Csp3 |
sigmoid, low=0.2, high=0.6, k=0.25 |
| Tanimoto similarity | TanimotoSimilarity |
sigmoid, low=0.3, high=0.7, k=0.25 |
| PAINS/alerts | custom_alerts |
filter — zeros total score on match |
| Substructure match | MatchingSubstructure |
filter — zeros total score if substructure absent |
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 6d ago First seen · 251 lines · 0 tokens per session scan A 84d6a0e522d5
rl is a skill published in the GitHub repository pregHosh/Solitarius-mcp (0 stars, last pushed 28d ago), licensed Apache-2.0. It adds 39 tokens to every session and 2,301 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
text-based-molecule-editing
Modify molecules based on natural language descriptions using MolT5/BioT5 models. Use this skill when: (1) User wants to modify a molecule to improve specific properties (solubility, potency, etc.), (2) User provides a molecule and asks to "make it more X" or "improve Y", (3) User wants to generate molecule variants…
biomed-research
Use when answering biomedical research questions that need source-backed evidence from local MCP servers, including gene, disease, drug, variant, phenotype, study, or clinical-trial questions.
tooluniverse
Access 1000+ scientific tools through ToolUniverse for drug discovery, protein analysis, genomics, literature search, clinical data, ADMET prediction, molecular docking, and more. Use when the user needs biomedical or scientific research capabilities.
datamol
Pythonic wrapper around RDKit with simplified interface and sensible defaults. Preferred for standard drug discovery including SMILES parsing, standardization, descriptors, fingerprints, clustering, 3D conformers, parallel processing. Returns native rdkit.Chem.Mol objects. For advanced control or custom parameters…
drug-design
End-to-end drug discovery pipeline orchestration. Deterministic Python script that auto-chains structure prediction, pocket detection, de novo design, docking, scoring, and ADMET filtering into reproducible workflows.
pocket-detection
Multi-method binding pocket detection and druggability assessment. Grid-based, fpocket, and P2Rank detection with druggability scoring, visualization, and cross-structure comparison.