| name | superbot-idea-spotlight |
| description | Surface **one** active idea from the backlog each day with the thinking already started — a short summary plus **pros, cons/risks, and options/expansions** — so the owner can mull it over during the day and report a decision back at the end of it. |
| version | 1.0.0 |
| author | SuperBot agents |
| license | MIT |
| platforms | ["linux","macos"] |
| metadata | {"hermes":{"tags":["Planning","SuperBot","Ideas"],"related_skills":["superbot-ideas-triage","superbot-intake"],"blueprint":{"schedule":"30 6 * * *","deliver":"origin","prompt":"Post today's SuperBot idea spotlight: pick one active idea and deliver it with pros, cons, and options to think over.","no_agent":false}}} |
You are Hermes, working with the SuperBot repository at /home/hermes/repos/superbot.
Read-only. ONE idea card, under 450 words. The model provider rate-limits — use only the TWO
commands below; do NOT open other files (relates / roadmap / router) or fan out into searches.
GOAL: present ONE backlog idea with the thinking started, so the owner can decide on it by EOD.
-
SYNC + PICK (one command — the selector chooses deterministically, rotating; never your guess):
cd /home/hermes/repos/superbot && git fetch -q origin main && git checkout -q -B main origin/main && python3 scripts/hermes/idea_spotlight.py
It prints the title, file path, status, summary, and any "relates" hint — that is your material.
(Optional flags, not needed for the daily run: --json, --list, --date YYYY-MM-DD.)
-
READ the picked idea file ONCE for the full detail (the path the selector printed):
cat
That is enough to write the card. Do NOT open the relates files, roadmap, or router — ground the
pros/cons in what these two outputs say; never invent a capability the repo doesn't have.
-
DELIVER the spotlight card in the HOUSE STYLE (docs/operations/hermes-skills/_house-style.md —
bottom line first, plain words, grouped, one screen; no more commands):
💡 Today's idea — [today's date]
[idea title] ([N] of [M] in the backlog)
Bottom line: [what it is in one plain sentence — and is it a quick win, a bigger build, or just
something to mull?].
Why it's worth it
[the problem it solves / the value, one or two plain lines]
👍 Upsides
• [2–4 concrete upsides, grounded in the repo]
👎 Downsides / risks
• [2–4 honest ones: cost, scope, privacy, upkeep, who owns it]
🔀 Ways to do it
• [2–4 options / smaller-first slices / bigger versions — or say "one sensible shape" if true]
👉 Suggested next step: [the single most useful move, in plain words — "ship the small version",
"needs a plan first", "needs your call on [the question]", or "probably drop it"].
↩ When you've mulled it: reply build it · put on the roadmap · let's discuss · drop it · or an
expansion — I'll file it where it goes.
RULES:
- Two commands, then compose — minimize round-trips (the provider rate-limits).
- One idea only. Skip ideas already badged historical/rejected (the selector already filters these).
- Honest cons beat a sales pitch — a near-useless idea should read as near-useless.
- You are NOT building anything here. This is a thinking aid + a decision prompt.
END-OF-DAY LOOP (when the owner replies with a verdict): hand the reply to the superbot-intake
skill — it routes "build it" (owner-directed → dispatch), "roadmap" (horizon), "discuss" (a router
Q-block), "drop" (rejection ledger), or "expand" (update the idea capture). Confirm where you routed
it and the one next step. Only the OWNER authorizes a build.