Draft a Hohmann transfer

A worked end-to-end draft: turn a request for a two-burn Hohmann transfer into a validated GMAT .script. This needs a model configured — the example uses Anthropic.

From the library

from gmat_copilot import draft

result = draft(
    "A Hohmann transfer from a 400 km circular Earth orbit to a 35786 km circular orbit: "
    "raise apogee with an impulsive burn, coast to apogee, then circularise with a second "
    "burn. Report the semi-major axis before and after.",
    model="anthropic:claude-...",
)

print(result.script)        # the generated GMAT .script
print(result.lint.clean)    # did it lint clean?
result.save("hohmann.script")

The phrasing carries the parts a correct transfer needs: the two orbits, two impulsive burns, the coast to apogee between them, and the reported quantity. The retrieval layer grounds the draft in the relevant GMAT resources (Spacecraft, ImpulsiveBurn, Propagator, ReportFile) and command syntax (Maneuver, Propagate ... {Sat.Apoapsis}), so the model writes against real syntax.

Inspect what came back through the result schema:

if not result.lint.clean:
    for d in result.lint.diagnostics:
        print(f"{d.line}:{d.column}: {d.severity.value}: {d.rule}: {d.message}")

for chunk in result.retrieval.chunks:
    print(f"{chunk.score:.3f}  {chunk.source}")   # the grounding the draft used

By default draft runs in strict mode: a draft that does not lint clean raises DraftRejected, with the offending result attached so you can inspect it.

from gmat_copilot import DraftRejected

try:
    result = draft("...", model="anthropic:claude-...")
except DraftRejected as exc:
    print(exc)                       # why it was rejected
    print(exc.result.script)         # the best-effort draft
    print(exc.result.lint.warnings)  # and its diagnostics

Pass strict=False to get the best-effort draft with its diagnostics attached instead of an exception.

From the CLI

gmat-copilot "A Hohmann transfer from a 400 km circular Earth orbit to a 35786 km circular \
orbit: raise apogee, coast to apogee, then circularise; report SMA before and after." \
  --model anthropic:claude-... -o hohmann.script

The script is written to -o (default mission.script; -o - for stdout) and a one-line lint summary is printed to stderr. Strict mode (the default) exits non-zero and writes nothing if the draft does not lint clean; --permissive writes the best-effort draft anyway.

Generation produces the script — review and run it in GMAT before trusting the trajectory. The lint gate catches malformed scripts, not wrong physics.