Scope Guard — catch scope creep + draft the boundary reply
Decide whether a request is scope creep — "is this in scope", "is this scope creep", "should I push back on this client request", "draft a boundary reply". Returns in-scope, creep or grey-zone with a
SKILL.md.claude/skills/scope-guard/SKILL.md
GitHubScope Guard
Does: spots scope creep in a client request and drafts a polite, firm boundary reply. Solves: saying yes to free extra work without noticing. Triggers on: "is this in scope" + the client message · "draft a boundary reply"
How Maxim does this (not a generic "say no" template)
Compares the incoming request against the agreed scope (paste the SOW/proposal, or describe it). Returns a verdict — in-scope / creep / grey-zone — with the specific clause or expectation the request exceeds, then drafts a reply via negotiation-specialist that holds the line without damaging the relationship (offers a change-order path, not a flat refusal).
Behavioral overlay
- Frameworks (cited per ADR-007): Loss Aversion (frame the change-order as protecting their timeline/quality, not your time) · Anchoring (re-anchor to the original agreement) · Commitment & Consistency (Cialdini — "as we scoped…") · Boundary-setting (assertive, non-defensive language).
- Tone-safe: the reply is firm + warm by default; it never threatens or guilt-trips.
- Confidence tag (ADR-010) on the verdict — 🟡 if the original scope wasn't supplied (it'll ask).
Output
Verdict (in-scope / creep / grey-zone) · the exact expectation exceeded · a ready-to-send boundary reply with a change-order option.