10 actions in 3 pieces of software · ranked by words, meaning and measured trust · 648 ms

1. 1@ikijs/mcpfound in code, not yet checkedno passport yet
   • validate_iki

     Validates a raw .iki model (object or JSON string); fail-fast — returns at most ONE error, so fix one, re-run.

   • describe_iki

     Summarizes a valid model's canvas/params/parts/deformers and its expressions and motion clips (with the descriptions a host picks them by); returns an error for an invalid model. It is also the patch writer's view of a rigged model: each part's rest box in model px and in bust-render px, each part's

   • render_poses

     Renders a rigged .iki into the character loop's pose PNGs with the real engine, through the installed Chrome, Edge or Chromium run headless (set CHROME_PATH to pick one). The model decides the set: a bust gets 560x560 shots (rest, turns, nod, blinks, gaze, combined — ParamAngleX/Y/Z all at 30 — and

   • list_standard_parameters

     Lists the recommended standard parameter ids a host can drive without per-model wiring.

   • compose_layers_from_parts

     Composes AI-generated part PNGs into canvas-aligned, role-named PNG layers on disk, ready for auto_rig_from_layers (the eyewhite split, alpha-trim/white-key, and placement pipeline the character-generation skill needs), with the same geometry report measure_layers returns standalone included inline.

   • auto_rig_from_layers

     Auto-rigs role-named PNG layers (face, eye_L/eye_R and either mouth or the lip set (mouth_inner + lip_lower + lip_upper, a mouth that folds open instead of crossfading) required; iris_L/R, brow_L/R, hair_front, body, arm_L/arm_R, forearm_pose_L/forearm_pose_R, mouth_tongue/mouth_teeth (only with the

   • measure_layers

     Reports read-only geometry checks over an already-composed layers directory — the same checks compose_layers_from_parts returns inline (iris ratio/offset, eye aspect, cropped/cut edges, a nose whose dense core is a speck of its crop, missing optional roles, a body whose feet a tall canvas's bottom c

   • measure_turn_reference

     Measures how far a head turns between two images of the same character — a front view and a turned one, at the SAME scale — as three ratios, so a rigged turn can be compared to a reference turn (or to an earlier rig) without registering the images: farEyeRatio (turned far/near iris width over the sa

2. 2mcpverified and signedpassport

   skillvercel-labs/json-renderrecord

   • mcp

     MCP Apps integration for json-render. Use when building MCP servers that render interactive UIs in Claude, ChatGPT, Cursor, or VS Code, or when integrating json-render with the Model Context Protocol.

3. 3grafana-mcpverified and signedpassport

   skillgrafana/ai-marketplacerecord

   • grafana-mcp

     MCP server for AI-assisted Grafana dashboard, datasource, alerting, and incident management.