pm-opportunities

Multi-host portability rules. When invoking AskUserQuestion:

1. The header field MUST be a short noun phrase (≤ 12 characters). Mistral Vibe rejects longer headers with string_too_long. Pick from: Start, Target, Scope, Audience, Methodology, Feature, Question, Confirm.
2. The options list MUST have at least 2 items. Vibe rejects empty/single-option calls. For free-text input, always provide ≥ 2 framing options (e.g. Yes, here's the input / Skip) — never call ask_user_question with options: [].

Preamble (run first)

source ~/.nanopm/lib/nanopm.sh 2>/dev/null || \
  source .nanopm/lib/nanopm.sh 2>/dev/null || \
  { echo "ERROR: nanopm not installed. Run: curl -fsSL https://raw.githubusercontent.com/nmrtn/nanopm/main/setup | bash"; exit 1; }
nanopm_preamble
_OPP_DIR=".nanopm/wiki/entities/opportunities"
mkdir -p "$_OPP_DIR"

What this skill does

/pm-opportunities maintains your opportunity database — the persistent, ranked set of user problems and unmet needs that sits between raw discovery (FEEDBACK.md, interviews, data) and planning (OBJECTIVES, ROADMAP, PRD). It is an LLM-wiki: a folder of markdown the agent owns and keeps current, not a thing you hand-maintain.

It runs in one of four modes:

• bootstrap — no DB yet: writes the SCHEMA.md conventions, gathers signal (feedback if present + your own assumptions + Nano's hypotheses), and fans out subagents to draft the initial ranked set, marking each opportunity's provenance before you confirm it.
• add — DB exists: capture one new problem (yours or a Nano pre-fill) as a new/updated opportunity, deduped against what's already there.
• generate — DB exists: draft up to N new candidate problems (global or for one theme), run them through the dedup agent, and write the survivors as nano-hypothesis. Additive — never overwrites.
• ingest-candidate (programmatic) — another skill (e.g. /pm-add-feedback) hands ONE already confirmed candidate — a problem statement + a verbatim quote with its citation + a raw-source ref — and /pm-opportunities runs the SAME canonical process (dedup → match-or-create → reindex → log → manifest) so it stays the single owner of the opportunity DB. Returns a small structured result so the caller can assemble its own summary. Non-interactive by default (the caller already confirmed).