Lobster Farmer Feeder

Feed lobsters in the Lobster Farmer game by calling the local CLI command `lobster-farmer feed` with `--model`, `--input-tokens`, `--output-tokens`, and opti...

Security Scan

This instruction-only skill is internally consistent: it only tells the agent to run a local Lobster Farmer CLI to feed lobsters and does not request extra credentials, installs, or network endpoints.

Purpose & Capability

The SKILL.md and agent manifest align with the stated purpose (invoking the local `lobster-farmer` CLI to feed lobsters). One minor omission: the instructions assume a Node.js runtime and a project-local CLI binary (bin/lobster-farmer.cjs) or an installed `lobster-farmer` command, but the skill does not declare these as required binaries. This is expected for an instruction-only skill but the user should confirm those prerequisites exist before running it.

Instruction Scope

Instructions stay within scope: check service status, start it if necessary, and run `lobster-farmer feed` with model/token/emotion args. The instructions reference only project-local paths and the CLI; they do not ask the agent to read unrelated files, environment variables, or send data to external endpoints.

Install Mechanism

No install spec or external downloads — lowest-risk model. Nothing will be written to disk by the skill itself; it only instructs running an existing local CLI/service.

Credentials

The skill requests no environment variables or credentials. It does imply the ability to execute local commands and to start/stop a local service, which is proportionate to the stated purpose.

Persistence & Privilege

always is false and the skill does not request persistent or elevated platform privileges. Note: as with any skill that runs local commands, the agent (when invoked) needs permission to execute CLI commands and manage the local service.

Assessment

This skill is instruction-only and appears coherent, but check these before installing: 1) Confirm you have the Lobster Farmer CLI (or Node.js and the project with bin/lobster-farmer.cjs) in the expected project root—otherwise the commands will fail. 2) Be aware the agent will run local commands and may start a local service (ensure you trust the project and its binaries). 3) There are no external endpoints or credentials requested by the skill, but if you run it, validate the CLI is from a trusted source. 4) If you do not want an agent to execute host commands autonomously, avoid enabling autonomous invocation for this skill or review agent permissions.

Feed model-specific lobsters by invoking the local CLI and report the resulting growth metrics.

Workflow

1. Confirm command context.

• Run from the Lobster Farmer project root where bin/lobster-farmer.cjs exists.
• Prefer the installed command lobster-farmer; use node ./bin/lobster-farmer.cjs as fallback.

2. Ensure service is reachable.

• Check status first:

lobster-farmer status

• If stopped, start service:

lobster-farmer start

• Respect explicit user port preference with --port.

3. Execute feed command.

• Single feed:

lobster-farmer feed --model "<model>" --input-tokens <n> --output-tokens <m> [--emotion "<text>"] [--port <p>]

• Short flags:

lobster-farmer feed -m "<model>" -i <n> -o <m> [-e "<text>"] -p <p>

4. Return result to user.

• Include model name, input/output token values, emotion, lobster total tokens, feed count, and size.
• If requested, run multiple feed commands sequentially and summarize each result.

Parameter Rules

• Require model as non-empty string.
• Require input_tokens + output_tokens > 0.
• emotion is optional; if omitted, UI should show ?.
• Keep token values as integers and non-negative.
• If user gives only a single total token value, default to: input_tokens = total, output_tokens = 0, unless user specifies another split.

Error Handling

• feed request failed: fetch failed: Service is not running or port is wrong. Start server and retry.
• API validation errors: Relay backend message directly (for example invalid token values).
• Port mismatch: Retry with explicit --port matching the running service.

Quick Examples

• Feed one lobster:

lobster-farmer feed --model gpt-4.1 --input-tokens 800 --output-tokens 400

• Feed with emotion:

lobster-farmer feed --model gpt-4.1 --input-tokens 800 --output-tokens 400 --emotion "focused"

• Feed different models:

lobster-farmer feed --model claude-3.7 --input-tokens 1200 --output-tokens 300
lobster-farmer feed --model qwen-max --input-tokens 600 --output-tokens 200