agentleFS
BaratiLabBaratiLab/mechanical-engineering-skills/skills/rocketsmith/SKILL.md

rocketsmith

Skill16 starsChanged 16 days agoMITView on GitHub

Design 3D-printed hobby high-power rockets (NAR/TRA Class 2, certified commercial motors only) — airframe, nose cone and fin geometry, Barrowman stability with fin sizing to a target caliber, 1-D flight simulation (apogee, rail-exit speed, thrust-to-weight, descent rate), printed-part mass from wall/infill settings, measured-mass and flown-apogee calibration, and printable STEP/STL for nose cone and fin can. Based on RocketSmith (Barati Farimani Lab). Use for rocket design-for-AM, stability checks, flight prediction, or post-flight calibration.

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 3 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Every design is a *candidate for review*, never a go-for-launch.

    line 21
  2. Very specific:

    Use the manufacturer's RASP .eng file from thrustcurve.org (--eng).

    line 25
  3. Distinctive:

    Check that the motor letter matches the user's certification level.

    line 27
  4. Very specific:

    Run check design.json for CG and CP, stability at liftoff and burnout, rail-exit speed, thrust-to-weight, Mach and descent rate.

    line 32
  5. Very specific:

    If stability is out of range, fins design.json --target 1.25 [--lever span_mm|root_mm|tip_mm] --out sized.json resizes the fins.

    line 34
  6. Distinctive:

    Use Cd 0.45–0.55 for smooth airframes.

    line 39

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

RocketSmith: design for AM of high-power rockets

RocketSmith (Pak et al., arXiv:2606.00097, 2026) designed four FDM-printed high-power rockets. All four launched stably and two were recovered reflyable. Altimeter apogees came in at 80–84% of the predictions.

Its key idea: the LLM orchestrates and every number comes from a tool, because LLMs that iterate numbers themselves plateau. scripts/rocket.py provides those tools. It needs numpy and matplotlib, plus CadQuery for parts. Details and the paper's flight data are in references/paper-notes.md.

Scope and safety. This skill covers airframe design for certified commercial motors within the flyer's certification:

  • H–I motors need L1 and J–L need L2; Class 2 ends at 40,960 N·s total impulse.
  • Motor, propellant, igniter and ejection-charge design or sizing are out of scope. Point users to the motor manufacturer, their certification and the Range Safety Officer.
  • Every design is a candidate for review, never a go-for-launch.

Workflow (mirrors RocketSmith's skills)

  1. Motor selection.
    • Use the manufacturer's RASP .eng file from thrustcurve.org (--eng).
    • motors lists the built-in certification summaries for the paper's motors, H100W and J425R.
    • Check that the motor letter matches the user's certification level.
  2. Design in design.json, with the nose cone, body segments, fins, internal masses and print settings. Size parts to the user's printer build volume (the paper split HP2 into three parts to fit an Ender 3 and an Ender 5 Plus).
  3. Stability analysis:
    • Run check design.json for CG and CP, stability at liftoff and burnout, rail-exit speed, thrust-to-weight, Mach and descent rate.
    • If stability is out of range, fins design.json --target 1.25 [--lever span_mm|root_mm|tip_mm] --out sized.json resizes the fins. When fins can't reach the target, move mass forward (avionics bay, nose weight) or lengthen the airframe.
  4. Predict the flight: simulate sized.json --plot flight.png [--cd 0.55 --rail 1.83].
    • Use Cd 0.45–0.55 for smooth airframes.
    • Printed rockets flew at Cd ≈ 0.8 in the paper's data. Say which value you used.
  5. Generate structures: cad sized.json --out parts/ writes the nose cone (ogive or other shape, with shoulder) and the fin can (tube, fins and an integrated motor mount), as STEP and STL.
    • Then add the details the paper's agent added on request: shear-pin and heat-insert pilot holes, eyebolt holes, rail-button bosses, avionics slots. Use cadsmith-text-to-cad or edit the CadQuery code.
    • Print in PETG or ABS, not PLA (heat). Use 4–5 walls, gyroid infill and 0.2 mm layers.
  6. Mass calibration: slice the parts, weigh them after printing, then calibrate design.json --weighed nose=412,fincan=1265 [--cg 520]. Re-check stability with the real masses; the paper's stability shifted with every manufactured part.
  7. After flight: calibrate design.json --weighed "" --apogee 276 fits the airframe's real Cd for the next prediction.

Review checklist to hand the user (from the paper's failures)

  • HP1: the eyebolt pulled out of the printed baffle and the lower airframe split. Make baffles and anchor points stronger than the printed walls, or anchor to the motor mount.
  • HP2: the recovery bay didn't separate. Check shear-pin count against separation force, and check the fit.
  • HP3: a friction-fit nose cone worked, at lower ejection resistance.
  • Weathercocking: stability much above 2 cal in wind costs altitude and drift.

Cite: Pak et al., RocketSmith, arXiv:2606.00097 (2026).

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.