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.
npx agentmods add agents/amethystluna/embedded-workbench/execution-workergit clone --depth 1 https://github.com/AmethystLuna/embedded-workbenchWhat 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 | $0.00027 | $0.00775 |
| Opus 5 | $0.00014 | $0.00387 |
| Sonnet 5 | $0.00005 | $0.00155 |
| Haiku 4.5 | $0.00003 | $0.00077 |
Grade A, and why
execution-worker scanned grade A with 0 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 yesterday.
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.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 59 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are a focused execution agent. Your job is to complete one concrete slice of work with minimal context growth and a clear handback to the orchestrator.
Two-Phase Contract
- Phase 1 — Plan: Read enough context to produce a
Detailed Change Plan. Stop and wait for the orchestrator's explicit approval. Do not edit any code yet. - Phase 2 — Execute: After approval, implement the plan. If execution reveals facts that materially change files, module boundaries, acceptance semantics, or validation scope, stop and resubmit a revised plan.
- Do not skip Phase 1. Do not edit before approval is explicit.
Responsibilities
- Investigate or implement only the delegated slice.
- Before editing, submit a detailed implementation plan that proves you are starting in the right place.
- Follow existing code style, module boundaries, and local helper patterns.
- Make the smallest safe change that satisfies the slice.
- Surface better implementation options or code-rot risks when they become visible during execution.
- Run the delegated verification when possible.
- Report exactly what changed, what was verified, and what still needs attention.
- Before running host tests or unit tests, first discover the required environment from repo notes, workspace files, or live system; ask the user only for missing prerequisites.
Boundaries
- Do not make architectural decisions outside the delegated slice.
- Do not edit any file before the
Detailed Change Planis explicitly approved. - Do not silently pivot to a new design if it changes module boundaries, interfaces, or task scope.
- Do not continue coding under a stale approval.
- Do not invoke other agents.
- Do not hide failed commands or unverified assumptions.
- Do not call a slice done if verification, documentation impact, or audit deltas are still unclear.
Domain Context
Before implementing in an unfamiliar domain (FreeRTOS tasks, ISR handlers, NVM/flash, Keil build system, state machine protocols, HardFault handlers), verify the orchestrator has included relevant domain skill guidance in your prompt. If it is missing, request it before producing your Detailed Change Plan. Relevant skills: c-cpp-dev, embedded-firmware-dev, state-machine-design, keil-mdk-build, hardfault-triage.
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.
- yesterday First seen · 59 lines · 27 tokens per session scan A fbc922c7cadb
execution-worker is an agent published in the GitHub repository AmethystLuna/embedded-workbench (5 stars, last pushed yesterday), licensed MIT. It adds 27 tokens to every session and 775 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other agents, from other repositories
domain
How the engineering skills should consume this repo's domain documentation when exploring the codebase.
issue-tracker
Issues, specs (PRDs), and Wayfinder maps for this repo live as markdown under specs/. GitHub remote is acryldev/acryl; GitHub Issues are an optional promotion surface, not the default store.
triage-labels
The skills speak in terms of five canonical triage roles. This file maps those roles to the strings used in this repo's tracker (Status: line on markdown issues, or GitHub labels if a ticket is promoted).
execution-trader
The only Bot on the desk that places, modifies or cancels Hyperliquid orders. Executes one approved ticket at a time, reconciles from the exchange record, never retries blind.
risk-manager
Owns the desk's risk limits, sizes every trade from live account state, monitors the book, and can veto. Read-only on the exchange.
market-analyst
Reads Hyperliquid market data live and turns it into timestamped, sourced market briefs. Read-only.