| name | illustrator-claw |
| description | Public-safe Illustrator Claw setup and automation workflow for connecting an Illustrator Claw agent runner to this Illustrator MCP repository. Use when Codex needs to document, configure, or verify Illustrator Claw workspace setup; design Illustrator Claw automations; connect Illustrator Claw to Illustrator MCP tools; use the Illustrator menu command index; or keep Illustrator Claw-related repo changes free of secrets, private paths, hostnames, logs, and local runtime state. |
Illustrator Claw
Use this skill to make Illustrator Claw work with this public Illustrator automation
repo without leaking private setup details.
Load References
Read only what the task needs:
docs/illustrator-claw-public-setup.md for beginner connection steps.
docs/illustrator-claw-automation-blueprints.md for safe automation patterns.
workflows/design-qa-with-ollama.md when tasks include local PNG QA via Ollama vision.
workflows/ollama-qwen-gemma-vision.md when the user wants Qwen VL or Gemma 3 specifically.
docs/public-boundary.md before any public repo edit.
docs/adobe-illustrator-mcp-tools.md for callable Illustrator MCP tools.
docs/illustrator-menu-command-links.md for Illustrator menu command values.
data/illustrator-menu-commands.csv when the generated menu index needs to be
rebuilt or checked.
Public Boundary
Keep these out of committed files and public answers:
- API keys, bearer tokens, auth files, session files, logs, caches, local
databases, and environment files.
- Private Illustrator Claw workspace names, hostnames, gateway details, account names,
local absolute paths, schedules, and agent logs.
- Client artwork, document names, screenshots, thumbnails, exports, and private
production prompts.
Use placeholders such as [private Illustrator Claw workspace] or [local MCP URL]
when an example needs a value.
Connection Workflow
- Confirm whether the task is public documentation or live private setup.
- For public docs, explain components and verification steps without copying
private values.
- For live setup, verify the actual Illustrator Claw runtime or workspace before
trusting exported notes.
- Confirm the repo checkout is readable by Illustrator Claw.
- Keep MCP server URLs and bearer tokens in the user's private Illustrator Claw
settings or secret store.
- Verify the connected tool list before attempting document or artwork calls.
- Test Illustrator automation on a blank or disposable document first.
Automation Defaults
- Start with read-only inventory: list tools, list documents, inspect the active
artboard, capture a preview, and run preflight.
- Prefer dry runs before transforms, recolors, cleanup, and exports.
- Require human approval before destructive, bulk, locked-art, hidden-art, or
irreversible edits.
- Store generated logs, screenshots, previews, and exports outside this public
repo unless they are fully redacted public examples.
- Prefer explicit MCP functions over menu commands when both can do the job.
Menu Commands
When using menu command data:
- Search
docs/illustrator-menu-command-links.md by menu path or command
value.
- Use the
value field as the command identifier.
- Check
docRequired, selRequired, minVersion, maxVersion, and ignore.
- Avoid
ignore rows unless the user is testing compatibility.
- Rebuild the index with
python3 tools/build-menu-command-links.py after
changing data/illustrator-menu-commands.csv.
Repo Edit Checks
Before committing Illustrator Claw-related repo changes:
git diff --check
rg -n "auth[.]json|session[_]index|state_[0-9]+[.]sqlite|logs_[0-9]+[.]sqlite|BEGIN (RSA|OPENSSH|PRIVATE) KEY" README.md CONTRIBUTING.md NOTICE.md SECURITY.md docs data tools workflows skills
If the menu command generator changed, also run:
python3 -m py_compile tools/build-menu-command-links.py
python3 tools/build-menu-command-links.py