karmaterminal/summon-human
v0.1.0MIT
Lets an agent ask frond-alarm to phone figs, the Thornfield frond's human, and read what happened. A skill gives the one command for "I want figs here now", how to word the call, and the room @-mention for what can wait. A standard-library script files one request in the alarm's own format, with up to three questions for figs, and reads the outcome and figs's answers from what its worker wrote. The only credential is your GitHub token, and the alarm accepts only the callers it lists.
Changelog
0.1.0 (2026-10-25): first release
- One skill,
summon-human: one command for "I want figs here now", since figs has said any seat may call, at any hour, for any reason (thornfield-plugins#16); the room @-mention as the light path for a question that can wait; how to word the 240 characters the call speaks and the questions it asks; and what each outcome and answer state means and what to do next. summon.py check: your login, the caller name the alarm lists for you and how the call announces you, and whether GitHub will keep thecall-requestlabel for you. It files nothing.summon.py submit: one issue in the alarm's issue-first format (karmaterminal/frond-alarm#9), byte for byte as the alarm'ssubmit.shwrites it, then the optional wake.--askadds up to 3 questions for figs (karmaterminal/frond-alarm#31). It refuses before filing what the alarm would reject, a second open live request of yours, and a reused request id. It closes an issue again whose label GitHub dropped, and reports a create it can't confirm as unknown instead of retrying it.summon.py status: the outcome, and figs's answers once the alarm posts them, from the labels and comments the alarm's worker wrote, with--waitpolling until both are in. Labels and comments from anyone else are named in a warning and never read as the outcome or as answers.