skill

An installable skill that teaches a coding agent how to connect to and work with a Canvas Drop instance, including through MCP, a standard way for tools to communicate with agents.

In plain words
What is it for?
Use it to install the Canvas Drop skill, connect an agent to an instance, and deploy or extend canvases there.
Why use it?
It packages the connection instructions and examples in the format agents can discover automatically, avoiding manual setup and correction.

Agent

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/markpasternak/canvas-drop/skill
Clone the repo
git clone --depth 1 https://github.com/markpasternak/canvas-drop
Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 733 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 1 finding. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00000 $0.00733
Opus 5 $0.00000 $0.00367
Sonnet 5 $0.00000 $0.00147
Haiku 4.5 $0.00000 $0.00073

Measured 2d ago against content hash 266214df55d4, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

skill scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

curl -fLOJ "{base}/skill.zip"
docs/site/agents/skill.md · 56 lines

How it starts

The opening of the file, as written. The whole thing — 56 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agent skill

Install this skill so your coding agent deploys and extends canvases against this instance, first try, without manual correction. It ships in the standard skill format (a SKILL.md with name/description frontmatter and a when-to-use trigger), so an agent loads it automatically when a task matches.

Download and install

GET /skill.zip is public — no session or API key required — so an agent can fetch and unpack it in one step:

curl -fLOJ "{base}/skill.zip"
unzip canvas-drop-skill.zip -d ~/.claude/skills/

Replace {base} with this instance's base URL. The download is named canvas-drop-skill.zip and unpacks to a single canvas-drop/ folder containing SKILL.md plus an optional examples/ directory. Point your agent at the unpacked SKILL.md, or drop the folder wherever your agent discovers skills. The skill is self-contained: it refers to this instance by base URL and asks the user for {base} when it doesn't know it.

What's inside

The zip is built from an explicit allowlist (SKILL.md plus examples/*.md), so it never carries a stray secret. The skill covers four ways to work against this instance:

  • Connect over MCP. Add {base}/mcp, sign in once through the instance's own login, then call identity-scoped tools (whoami, list_canvases, create_canvas, deploy_canvas, get_canvas_file, rollback_canvas, the draft-editor loop, version export/deletion, and more — 44 tools in all) with no key to paste. See the MCP server.
  • Deploy over HTTP with a per-canvas API key: PUT {base}/v1/canvases/{id}/deploy (Bearer auth, ZIP body) publishes immediately. Companion read-back and recovery routes (GET {base}/v1/canvases/{id}, …/versions, …/files, POST …/rollback, POST …/unpublish) let an agent confirm what went live and undo it. See Deploy API.
  • Add backend capability with the zero-config browser SDK. One <script src="/sdk/v1.js"> tag defines the global window.canvasdrop and rides the signed-in session cookie, so the five primitives (KV, files, me(), AI, realtime) work with no keys in client code. See the SDK overview.
  • The golden rules. Never put a secret in canvas files. Canvases are static only, with no server build step. Every primitive is off until the owner enables Backend plus that feature, so a disabled call throws CapabilityDisabledError (code: "CAPABILITY_DISABLED", status 403).
  • Typed errors. Branch on a stable err.code / err.status rather than parsing messages. Full table at Error codes.

Read the full file on GitHub · 56 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 56 lines · 0 tokens per session scan A 266214df55d4

Subscribe to this mod's changes

skill is an agent published in the GitHub repository markpasternak/canvas-drop (7 stars, last pushed 19d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 733 tokens. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.