aster
Pointer to the canonical aster agent skill in skills/aster/. Use when an agent must define a parameterized antenna design space, run dry or real AEDT-backed design studies through a guarded solver, keep measured artifacts as evidence, apply explicit bandwidth/pattern acceptance rules, or re-score recorded runs without re-solving. The web GUI is not included; real runs require the ansys-aedt-mcp guard plus AEDT 2026 R1.
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Check it deeper on your machine
It has changed since its scan on 21 Nov 2026.
As a Claude Code plugin, install it and then ask Claude Code:
/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/Hawse1337/aster with scanskillsafety
Or from a terminal, clone it and run it:
git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/Hawse1337/aster'
It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.
Badge for your README
Markdown
[](/listings/hawse1337--aster)HTML
<a href="/listings/hawse1337--aster"><img src="/api/listings/hawse1337--aster/badge.svg" alt="findsafeskills safety rating" /></a>Install
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
SKILL.md · Manifest · 18 lines
- ---
- name: aster
- description: Pointer to the canonical aster agent skill in skills/aster/. Use when an agent must define a parameterized antenna design space, run dry or real AEDT-backed design studies through a guarded solver, keep measured artifacts as evidence, apply explicit bandwidth/pattern acceptance rules, or re-score recorded runs without re-solving. The web GUI is not included; real runs require the ansys-aedt-mcp guard plus AEDT 2026 R1.
- ---
- # aster skill
- The canonical skill (with `references/` and `scripts/`) lives in
- [`skills/aster/`](skills/aster/SKILL.md).
- Install it into an agent's skill directory:
- ```powershell
- aster skill install # → %USERPROFILE%\.dsh\skills\aster
- ```
- Or copy `skills/aster` there manually. Full documentation: [README.md](README.md) ·
- [user_doc/](user_doc/README.md) · [https://hawse1337.github.io/aster/](https://hawse1337.github.io/aster/)
README.md · README · 171 lines
- <p align="center">
- <img src="assets/banner.svg" alt="aster — an antenna design bench, evidence first" width="100%">
- </p>
- <p align="center">
- <a href="https://github.com/Hawse1337/aster/actions/workflows/ci.yml"><img alt="tests" src="https://github.com/Hawse1337/aster/actions/workflows/ci.yml/badge.svg"></a>
- <a href="https://github.com/Hawse1337/aster/releases/latest"><img alt="release" src="https://img.shields.io/github/v/release/Hawse1337/aster?color=e4572e"></a>
- <a href="https://hawse1337.github.io/aster/"><img alt="docs" src="https://img.shields.io/badge/docs-online-2a9d8f"></a>
- <a href="LICENSE"><img alt="license: MIT" src="https://img.shields.io/badge/license-MIT-2a9d8f"></a>
- <img alt="python 3.10+" src="https://img.shields.io/badge/python-3.10%2B-23262b">
- </p>
- # Keep the measurement. Change the target.
- > **aster** was previously published as **antopt**. The package import, CLI and
- > environment prefix are now `aster` / `ASTER_*`; the old repository URL redirects.
- `aster` is a design bench for antenna studies. You describe a parameter space and what "good"
- means, the library drives candidates through a guarded AEDT session, and every solve leaves its raw
- artifacts behind. When the requirement changes — a different band, a tighter ripple, a flatter
- elevation cut — you re-judge the history instead of burning another night of solver time.
- > **中文简介**:[README.zh-CN.md](README.zh-CN.md)
- > The web GUI / desktop application is under validation and is **not** part of this repository.
- 
- ## Three forms, one bench
- | form | how you get it | use it when |
- |---|---|---|
- | **Python library + CLI** | wheel/sdist from [Releases](https://github.com/Hawse1337/aster/releases), then `aster doctor` | you script your own loop, run CI, or work in a notebook |
- | **Agent skill** | `aster skill install` (or copy `skills/aster/`) | an AI agent should run design studies with evidence rules |
- | **Docs + releases** | [hawse1337.github.io/aster](https://hawse1337.github.io/aster/) | you want the contract, the workflow and pinned artifacts |
- ## The bench
- Four verbs, one loop:
- - **Design** — `ParamSpec` / `ParamSpace` describe the knobs, `Objective` / `Constraint` describe the
- target. Nothing here knows about AEDT.
- - **Solve** — a candidate becomes a real run through the companion guard
- ([ansys-aedt-mcp](https://github.com/Hawse1337/ansys-aedt-mcp)). Sessions are addressed explicitly;
- budgets are spent, not assumed.
- - **Keep** — S11 Touchstone, far-field grids and native report images stay with the run record. A
- number without its artifact is a rumour.
- - **Judge** — acceptance rules are code (`band_acceptance`, `pattern_acceptance`, `acceptance`), so a
- verdict can be recomputed, compared and challenged.
- The archive is the point. A *campaign* is just a Python recipe over that loop; swap the evaluator
- (dry model or real solve) and the same recipe still runs.
- ## What it refuses to hide
- | risk | how aster handles it |
- |---|---|
- | "the result looks great" | every measured run keeps `result.s1p`, `farfield_views.json` and the native report PNGs |
- | "we changed the target, re-run everything" | records are re-scored; the solver is not called again unless you ask |
- | "the surrogate says it is good" | dry/surrogate results are labelled `PREDICTED`; validation state is explicit |
- | "the solve died, we lost the point" | failed runs are recorded as failed, with the reason, not dropped |
- | "which AEDT session was that?" | the guard requires one explicit PID or gRPC port per call |
- ## Try it in one minute
- ```powershell
- py -3.10 -m venv .venv
- .\.venv\Scripts\python.exe -m pip install -U pip
- .\.venv\Scripts\python.exe -m pip install -e ".[optimizer,surrogate,plot]"
- aster version
- aster doctor # what is missing before a real campaign
- aster dry-demo --samples 64
- ```
- `dry-demo` is a complete study on an analytic objective — Latin-hypercube design, Pareto archive,
- no AEDT required. The same shape in code:
- ```python
- from aster.core import Objective, ParamSpace, ParamSpec, latin_hypercube
- from aster.optimize import Evaluation, pareto_archive
- space = ParamSpace([
- ParamSpec("ring_radius", "mm", 20.0, 60.0, 40.0),
- ParamSpec("ring_height", "mm", 1.0, 10.0, 4.0),
- ])
- objectives = [Objective("gain", "peak_gain", "maximize"),
- Objective("bw", "s11_bandwidth", "maximize")]
- runs = [Evaluation("c%02d" % i, "ring", p,
- metrics={"peak_gain": predict_gain(p),
- "s11_bandwidth": predict_bw(p)}, feasible=True)
- for i, p in enumerate(latin_hypercube(space, 32, seed=1))]
- front = pareto_archive(runs, objectives)
- ```
- ## Runtime safety, by construction
- Real solves go through the companion guard
- [ansys-aedt-mcp](https://github.com/Hawse1337/ansys-aedt-mcp), not through a loose PyAEDT call in
- the same process:
- - the guard records the AEDT sessions **it** launched and refuses to attach to unmanaged ones
- (`guard_status` classifies every live `ansysedt.exe`);
- - `run_python_code` / `run_python_script` are refused by default at the guard's MCP and worker
- layers; a human operator can enable them for a supervised session;
- - run tokens bind analysis submissions to project/design/setup when
- `AEDT_GUARD_ENFORCE=1`;
- - launch/release events are audited and PyAEDT app objects are checked against the verified desktop.
- If the guard MCP tools are missing, **stop** — do not fall back to shell-level PyAEDT or
- `ansysedt.exe`. Fail-closed DSH/Codex configurations and the threat model are in the guard repo.
- See [user_doc/safety.md](user_doc/safety.md) for the boundary and residual risks.
- ## Real studies
- 1. Install the guard system and its virtual environment (PyAEDT 1.10.0, AEDT 2026 R1).
- 2. Point `ASTER_AEDT_EXE`, `ASTER_GUARD_REPO`, `ASTER_VENV_PYTHON`,
- `ASTER_REFERENCE_PROJECT`, `ASTER_CAMPAIGN_ROOT` at your machine.
- 3. Run the worker against a campaign directory; each candidate gets its own run folder and archive
- row.
- The contract, the run lifecycle and the artifact layout are in
- [user_doc/aedt_integration.md](user_doc/aedt_integration.md), with a walkthrough in
- [examples/aedt_campaign.md](examples/aedt_campaign.md).
- ## Drop it into an agent
- The skill lives in [`skills/aster/`](skills/aster/): `SKILL.md` plus `references/` (API map, AEDT
- contract, evidence rules) and a runnable `scripts/dry_demo.py`. Install it with one command:
- ```powershell
- aster skill install # → %USERPROFILE%\.dsh\skills\aster
- aster skill path # print the destination
- ```
- Or copy `skills/aster/` into your agent's skill directory by hand. The skill does **not** cover the
- GUI, which is not published.
- ## Repository map
- ```text
- assets/ banner + loop illustrations
- aster/ the library
- core.py parameter spaces, objectives, constraints
- optimize.py evaluations, dominance, Pareto archive
- campaign.py campaign recipes
- runtime/ measured-run manager + job store (no web server)
- worker.py command-line measured campaign worker
- aedt_host.py AEDT process helpers
- skills/aster/ bundled skill (mirrored from skills/aster at build time)
- cases/ benchmark/ interop/ optimizer/ parametric/ physics/ surrogate/ topology/
- skills/aster/ canonical agent skill (SKILL.md + references + scripts)
- tools/ sync_skill_bundle.py (mirror the skill into the package)
- user_doc/ installation, architecture, workflow, AEDT integration, skill, validation
- examples/ dry study + measured-campaign walkthrough
- tests/ offline suite (no AEDT needed)
- mkdocs.yml documentation site
- ```
- ## Status
- - Offline suite: **782 passed, 80 subtests**; CI runs it on Windows / CPython 3.12.
- - Measured runs are exercised on AEDT 2026 R1 through the guard; any solver number quoted from a
- study must be accompanied by its run artifacts.
- - Wheel and sdist: [v1.54.0 release](https://github.com/Hawse1337/aster/releases/tag/v1.54.0);
- documentation: [hawse1337.github.io/aster](https://hawse1337.github.io/aster/).
- ## License
- MIT — see [LICENSE](LICENSE). If you build a study on top of this bench, keep the artifacts; a bench
- without evidence is just furniture.
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.