GlassFrog MCP Server

GlassFrog

Operational map for working with Holacracy records inside GlassFrog.

Data model anchors

• Circles — nested teams from the Anchor Circle outward; optionally pull membership payloads when aligning people allocation.
• Roles — accountabilities bounded to circles (is_core differentiates seeded circle roles vs custom scopes).
• Assignments — person-to-role pairings spanning circles; foundational when determining who staffs a constrained role seat.
• People — member directory keyed for cross-reference via assignments output.
• Projects — deliverables bound to circles and roles; statuses move through operational states including archive-friendly outcomes.
• Metrics — measured cadences scoped to circles/roles plus optional organization-wide overlays when global reporting is surfaced.
• Checklist items — recurring rhythms (weekly/monthly/quarterly accountability checks).
• Custom fields — supplemental schema carried on upstream records (typically read-facing).

Order of operations

1. Refresh the hierarchy and optional membership expansion when seat planning touches multiple circles simultaneously.
2. Pull assignments beside people metadata to reconcile human-facing names with linkage identifiers before mutating workloads.
3. Filter projects into a manageable slice (for example circle-scoped) before attempting accountability edits.
4. When project creation rejects because a role seat is already staffed, correlate assignment rows against that role identifier and supply the staffing person linkage on retry.
5. Prefer archive-style status transitions for projects when lineage matters instead of invoking destructive removals.
6. Metrics and checklist entries always anchor on circles; widen coverage with explicit global overlays only when dashboards expect organization-level rows.

Working constraints

• Standard responses embed links collections—resolve textual labels via follow-up lookups instead of trusting ids alone silently.
• Update paths emit JSON Patch operations internally; callers keep supplying scalar fields—the platform emits the compliant patch payloads.
• Read endpoints return exhaustive arrays without pagination knobs—expect larger payloads rather than iterative paging.