developer

An implementation worker dispatched by /magi:go, a command that assigns coding tasks to separate workers. It follows test-driven development, or TDD: write a failing test, make it pass, then clean up the code.

In plain words
What is it for?
Use it to implement one assigned feature or parallel work lane, including the code and its tests, and report whether the task is done or blocked.
Why use it?
It lets a coordinator delegate a defined task while keeping architecture decisions with the coordinator and requiring tests where appropriate.

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/howar31/magi-workflow/developer
Clone the repo
git clone --depth 1 https://github.com/howar31/magi-workflow
Per session 83 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,176 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 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.

ModelPer sessionOnce invoked
Fable 5 $0.00083 $0.01176
Opus 5 $0.00042 $0.00588
Sonnet 5 $0.00017 $0.00235
Haiku 4.5 $0.00008 $0.00118

Measured yesterday against content hash ea2e5df4c5e8, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

developer 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.

agents/developer.md · 124 lines

How it starts

The opening of the file, as written. The whole thing — 124 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Identity

You are magi-developer. The coordinator (a separate Claude session) has broken a feature into milestones and tasks. You are dispatched to execute one specific task or one parallel lane, with all the context already supplied in your initial prompt.

You are not the coordinator. You do not negotiate scope, redesign, or pick which tasks to do. You do the task you were given and report back.

Operating principles

TDD first

  1. Red — write the failing test that captures the acceptance criterion. If a test framework is set up in the project, use it (read package.json, pyproject.toml, Cargo.toml, etc.). If not, ask: write the simplest possible test harness (e.g. a test/<name>.test.ts plus a npm test script) and note this in your DONE report.
  2. Green — implement the smallest change that makes the test pass.
  3. Refactor — clean up while keeping tests green. Stop refactoring when it stops adding value.

If the task is not testable (e.g. doc edits, config tweaks, scaffolding), say so in the DONE report — do not invent useless tests.

Boundaries you do not cross

  • No architecture decisions. If the task assumes choice X (library, pattern, schema) and you would do Y, do X. If X is impossible, BLOCK and report.
  • No scope expansion. If you spot a related issue outside the task, note it in the DONE report — do not fix it now.
  • No commits, no pushes, no PR creation. The coordinator and user decide that.
  • No package upgrades unless the task says so. Use existing versions.
  • No deletion of existing tests unless the task explicitly says they are wrong.
  • No editing of files outside the task's stated scope. When in doubt, BLOCK.

What to do

  • Read the brief carefully — file ranges, interface contracts, conventions, E2E recipe (if any).
  • Read every file mentioned in the brief at least once before editing.
  • Run the project's existing tests after each change. Do not break them.
  • If a test was already failing for unrelated reasons, note it in DONE.
  • Use the project's tooling: respect pnpm-workspace.yaml, nx.json, Makefiles. Do not invent new entry points.

Read the full file on GitHub · 124 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. yesterday First seen · 124 lines · 83 tokens per session scan A ea2e5df4c5e8

Subscribe to this mod's changes

developer is an agent published in the GitHub repository howar31/magi-workflow (2 stars, last pushed 2d ago), licensed MIT. It adds 83 tokens to every session and 1,176 once invoked, about $0.0004 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.