What is workshop AGENTS.md?

Instructions for using or developing Workshop, a local interface and service for observing agent activity. They distinguish installing and running it from developing its source code, with separate commands for each.

Use them when installing Workshop, starting its interface, connecting hosted observability, or building, testing, linting, and type-checking the project.

They prevent an agent from taking the wrong setup path and define the checks required before submitting development changes.

workshop AGENTS.md — Instructions file by raindrop-ai

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 instructions/raindrop-ai/workshop/agents-md

Clone the repo

git clone --depth 1 https://github.com/raindrop-ai/workshop

Made for: Codex, OpenCode.

Per session 451 This file is loaded in full into every session.

When invoked 451 The same file — it is already loaded in full.

Security scan C 2 findings. 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.

Model	Per session	Once invoked
Fable 5	$0.00451	$0.00451
Opus 5	$0.00226	$0.00226
Sonnet 5	$0.00090	$0.00090
Haiku 4.5	$0.00045	$0.00045

Measured 2d ago against content hash a996a70f768b, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade C, and why

workshop AGENTS.md scanned grade C with 2 findings 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.

Downloads and executes remote codehighSupply chain

curl | sh runs whatever the server returns today, which is not necessarily what it returned when this was reviewed.

curl -fsSL https://raindrop.sh/install | bash

Makes network callslowCapability

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

curl -fsSL https://raindrop.sh/install | bash

AGENTS.md · 68 lines

What it actually says

Agent guide

Read this first to decide what the user actually wants.

Did the user ask you to use Workshop? (the common case)

Just run the one-liner:

curl -fsSL https://raindrop.sh/install | bash

That installs the raindrop CLI. Then raindrop workshop starts the UI.

Do not clone this repo or bun install for this. Building from source is only for developing Workshop itself.

Hosted observability instead? raindrop cloud setup connects to app.raindrop.ai (no local daemon).

Did the user ask you to develop Workshop?

Build from source:

bun install
bun run dev   # daemon on :5899, Vite UI on http://localhost:5900

Commands

bun run build         # build
bun run test          # test
bun run lint          # lint
bun x tsc --noEmit    # typecheck

See .devin/blueprint.yml and .cursor/README.md for full dev-environment setup.

How to work

Touch only what the task requires. Match existing style.
Every changed line must trace to the user's request.

Verification (Definition of Done)

Before creating a PR, run and confirm:

bun run build         # MUST exit 0
bun run test          # MUST pass
bun run lint          # MUST exit 0
bun x tsc --noEmit    # MUST exit 0

For UI changes: visually verify changed pages render correctly and check neighboring pages.

Then: review your diff — no unintended changes, no debug code, changes scoped to the task.

Do not touch (without explicit permission)

.github/workflows/ — propose, don't apply
The raindrop CLI installer (https://raindrop.sh/install)
Publish/release configuration

Agent operating notes

Keep a running memory of mistakes here. Add a concrete guardrail when you make a repeatable stumble.

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.

2d ago First seen · 68 lines · 451 tokens per session scan C a996a70f768b

Subscribe to this mod's changes

workshop AGENTS.md is an instructions file published in the GitHub repository raindrop-ai/workshop (1,067 stars, last pushed 10d ago), licensed MIT. It adds 451 tokens to every session, about $0.0023 per session on Opus 5. A static security scan graded it C with 2 findings (downloads and executes remote code, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

LLM raindrop tracing

workshop AGENTS.md

Getting it into your agent

What it costs to keep this loaded

Grade C, and why

Downloads and executes remote codehighSupply chain

Makes network callslowCapability

What it actually says

Agent guide

Did the user ask you to use Workshop? (the common case)

Did the user ask you to develop Workshop?

Commands

How to work

Verification (Definition of Done)

Do not touch (without explicit permission)

Agent operating notes

What this file has done since we first saw it

Other instructions, from other repositories

openinference AGENTS.md

raindrop-mcp AGENTS.md

raindrop-mcp mcp-inspector.instructions.md

raindrop-mcp GEMINI.md

raindrop-mcp mcp-dev.instructions.md

raindrop-mcp mcp-refactor.instructions.md