thierryc/glyphs-mcp/glyphs-mcp-outlines-docs
Claude Code and agent skill by thierryc/glyphs-mcp
- community
Checking that the SKILL.md file is still in the repository.
Inspect curves and safely prepare typed outline path edits.
Install
npx vskill@latest install thierryc/glyphs-mcp/glyphs-mcp-outlines-docsSKILL.md
Glyphs Outlines
When the target depends on the current selection, read compact context and native selection, then bounded selected-node details as needed. Record whether selection informed the target. Explicit glyph/layer/path/node targets always win; unrelated selection never retargets a mutation. Skip context/selection reads for fully explicit targets; read their exact geometry and guards instead.
For layer identity, advances and bounds, use read_entities and the native layer read reference. An outlineHash is a change guard; it does not reveal node/component geometry. Report unavailable evidence explicitly before choosing a native UI route.
For selection inspection, use the selection read reference. Require selection.context.v1 in this connection’s negotiated read capabilities. If unavailable, explain that the private installation needs its bridge, sidecar and skills updated together; do not substitute an earlier selection workflow. Recommend the compact object counts; request bounded node details only when needed. Distinguish empty selections, no Edit View and incomplete evidence. This read-only workflow needs no job or Save.
Reuse the verified $glyphs connection, capabilities and intended document_id already in context; do not repeat setup for this skill. If missing, use connection setup and document targeting. Reuse the ID for reads of the same font; rediscover after document_not_found or target change, not a missing glyph. Never silently substitute another open font. Each read is fresh; check source/dirty state when preparing edits.
Prefer the shared conversation workflow for typed outline edits, with one request and compact polling. Explicit low-level job work can use start_job, get_job and apply_job. Application is a reversible live change; it does not save. Use accept_job only when the user's task authorizes saving the reviewed whole document; it closes the rollback window. Use discard_job for whole-job restoration or cancellation. Native Undo and Redo are grouped per glyph. Never save merely to satisfy preparation, and never save, export, close, or overwrite a font unless the user's task authorizes it. Do not retry an uncertain write or save as a new job; reconcile the existing job identity first.
For general path reads and kind="outline_edit", follow the typed path-editing workflow. Require paths.list.v1, path.geometry.v1, and bridge writeCapabilities containing outline.edit.v1 before mutation. Background reads/edits additionally require outline.background.read.v1 / outline.background.edit.v1 and explicit surface:"background"; layer IDs always identify the owning foreground. Resolve explicit native layer IDs, path and raw node indices, then retain exact returned pathHash guards. Reads remain available on dirty or unsaved fonts; jobs still require a saved clean baseline. For ordinary node removal, also require outline.remove-node.v1 and use remove_node; reserve raw delete_nodes for explicitly requested topology surgery because it does not invoke Glyphs' keep-shape behavior.
For bulk transforms or a reviewed sequence of cleanup/anchor/component edits, offer direct native scripting under the shared route and recovery guide. One callback per surface can perform all requested steps locally, avoiding per-path tool calls and layer snapshots. Preserve fractions and untargeted contents, then verify geometry and compatibility. Use typed outline_edit for individual-node work and its topology checks and selective recovery.
For a complete-target Glyphs command, prefer an advertised closed native action over scripting. The outline routes are correct_path_direction, round_coordinates, add_extremes, cleanup_paths, remove_overlap, add_missing_anchors, align_components, decompose_components, decompose_corners, make_components, connect_open_paths, and swap_foreground_background. Follow the native-action reference, preserving its exact layer targeting and action-specific risk review. Do not use these selection-independently defined actions to emulate a selection-sensitive command.
For curvature display, load the focused Curve Inspector guide. Use Reference Inspector to compare against Last Saved, a Font File, Local Git or Public GitHub. Keep draw callbacks display-only. Changes labels belong only to the edited occurrence, never repeated preview glyphs. Native outline visibility must survive startup, activation, node edits, panning, Undo and Redo. Use typed outline_edit, start_nodes, or slant only for their advertised operations; do not invent curve-edit tools.
For bulk vertical mirroring, resolve the exact master and selected surface, then use the vertical-flip example with an explicit pivot. Default to combined path-bounds center per background. Direct native scripts capture a detached pre-run copy at every target count, without Save. Verify foregrounds, other masters, anchors and components remain unchanged. Restore before run restores the whole pre-run font including existing unsaved edits and discards later unsaved edits; typed operations retain their selective recovery.
Security
- Security scanPassed, 100 of 100Pattern checks on the skill's files.
- Model reviewNot runA model review is scheduled for the next rescan.
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingskills/glyphs-mcp-outlines-docs/SKILL.md
Popularity
- Repository stars
- 54
- Repository forks
- 7
- 7-day trend score
- 19
Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/thierryc/glyphs-mcp/glyphs-mcp-outlines-docs)