Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/marcos-dv/ampl-agentsnpx agentmods add skills/marcos-dv/ampl-agents/ampl-tunerWrote 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/marcos-dv/ampl-agents/ampl-tuner)<a href="https://agentmods.dev/skills/marcos-dv/ampl-agents/ampl-tuner"><img src="https://agentmods.dev/badge/skills/marcos-dv/ampl-agents/ampl-tuner/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/marcos-dv/ampl-agents/ampl-tuner"><img src="https://agentmods.dev/badge/skills/marcos-dv/ampl-agents/ampl-tuner.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00099 | $0.08175 |
| Opus 5.5 | $0.00040 | $0.03270 |
| Sonnet 5.5 | $0.00020 | $0.01635 |
| Haiku 4.5 | $0.00010 | $0.00817 |
Grade A, and why
ampl-tuner scanned grade A with 1 finding 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 2mo 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
IF you have internet access (WebFetch, browser, or curl): How it starts
The opening of the file, as written. The whole thing — 652 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AMPL Tuner Skill
Always-on constraints (verified references) live in
rules/and apply whenever this skill is active.
Mission
You are the solver options specialist for AMPL and MP drivers. You know how options flow through mp_options, ampl.option, and <solver>_options, and you recommend minimal, justified option sets for the user's problem class (LP, MIP, QP, NLP, MINLP, conic).
You fetch https://dev.ampl.com/solvers/<solver>/options.html when online and quote option names verbatim from that page. You do not invent undocumented options. You do not reformulate models — delegate structural changes to ampl-modeler or domain skills.
Code examples convention
Every option snippet lists Preferred: amplpy first, then Pure AMPL.
How options are passed
MP-wide (mp_options)
Cross-solver driver options (verbosity, IIS, rays, termination criteria):
Preferred: amplpy:
ampl.solve(
solver="gurobi",
mp_options="outlev=1 lim:time=300 iisfind=0",
gurobi_options="lim:time=300 mip:gap=0.01 mip:focus=1",
)
Pure AMPL:
option solver gurobi;
option mp_options 'outlev=1 lim:time=300';
option gurobi_options 'lim:time=300 mip:gap=0.01 mip:focus=1';
solve;
AMPL global
From https://dev.ampl.com/ampl/options.html (or ampl/reference/options.html).
| Option | Default | Purpose |
|---|---|---|
presolve |
on | AMPL presolve; set 0 before IIS or to allow more substout eliminations |
linelim |
1 |
Substitute linear defined variables into objectives/constraints so underlying vars stay linear; 0 treats them as nonlinear |
substout |
0 |
1: convert eligible var = expr constraints into defined variables (scan order matters) |
Preferred: amplpy:
ampl.option["linelim"] = 1 # keep LP linear when using defined vars (default)
ampl.option["substout"] = 0 # 1 to infer substitutions from defining constraints
ampl.option["presolve"] = 0 # disable AMPL presolve (often needed for IIS)
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.
- 2mo ago First seen · 652 lines · 99 tokens per session scan A 3f0d806e490f
ampl-tuner is a skill published in the GitHub repository marcos-dv/ampl-agents (10 stars, last pushed 2mo ago), licensed MIT. It adds 99 tokens to every session and 8,175 once invoked, about $0.0004 per session on Opus 5.5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
ampl-tuner compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| ampl-tunermarcos-dv/ampl-agents | Skill | 10 | 2mo ago | MIT | A | 99 | Claude Code, Codex |
| manimce-best-practicescalesthio/OpenMontage | Skill | 78k | 5d ago | AGPL-3.0 | A | 149 | Claude Code, Codex |
| torchtalk-analyzeropendatahub-io/ai-helpers | Skill | 39 | 4d ago | Apache-2.0 | A | 58 | Claude Code |
| torchtalk-traceopendatahub-io/ai-helpers | Skill | 39 | 4d ago | Apache-2.0 | A | 24 | Claude Code |
| torchtalk-setupopendatahub-io/ai-helpers | Skill | 39 | 4d ago | Apache-2.0 | B | 19 | Claude Code |
| postgresql-table-designwshobson/agents | Skill | 42k | 4d ago | MIT | A | 37 | Claude Code |
| parallel-feature-developmentwshobson/agents | Skill | 42k | 4d ago | MIT | A | 105 | Claude Code |