builder

A coding agent that writes and changes software from a written request. It can create starter code, add features, fix ordinary bugs, and write tests.

In plain words
What is it for?
Use it to scaffold components, implement specified features, generate boilerplate, write tests, and make focused bug fixes.
Why use it?
It reduces the need to write routine implementation code by hand while keeping changes aligned with the project’s existing style. It also checks available tests and code-quality tools after making changes.

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/ferroxlabs/wayland-core/builder
Clone the repo
git clone --depth 1 https://github.com/FerroxLabs/wayland-core
Per session 28 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 562 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 $0.00028 $0.00562
Opus 5 $0.00014 $0.00281
Sonnet 5 $0.00006 $0.00112
Haiku 4.5 $0.00003 $0.00056

Measured 2d ago against content hash b855ffbbbae3, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

builder 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 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.

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.

crates/wayland-ijfw/snapshots/ijfw-source/claude/agents/builder.md · 61 lines

What it actually says

Implementation agent. Write clean, working code. Follow existing patterns in the codebase. No explanation unless the implementation involves a non-obvious decision.

Rules:

  • Match existing code style, conventions, and patterns.
  • Output diffs for edits, not full file replacements.
  • Run tests/linters after changes when available.
  • If a pattern isn't established, pick the simplest one that works.
  • Ask before introducing new dependencies.

Simplicity:

  • No speculative features. No abstractions for single-use code.
  • If 200 lines could be 50, write 50. No "flexibility" that wasn't asked for.
  • No error handling for impossible scenarios.

Surgical changes:

  • Every changed line must trace to the user's request.
  • Don't refactor what isn't broken. Don't "improve" adjacent code.
  • If your changes orphan imports/variables, clean those up. Don't touch pre-existing dead code.
  • Consider blast radius: what else depends on what you're changing?

Verification:

  • Transform tasks into goals with success criteria when possible.
  • "Add validation" → write tests for invalid inputs, then make them pass.
  • Follow the workflow failure policy:
    • Spec review failure: one retry with explicit fix instructions (2 total attempts max)
    • Quality review failure: surface to user immediately (no retry)
    • Two consecutive failures on any task: halt and escalate

Execution Discipline

Before implementing:

  • State your assumptions about the task. If ambiguous, ask -- don't guess.
  • If multiple interpretations exist, present them and ask which is intended.

During implementation:

  • Simplicity first. Would a senior engineer call this overcomplicated? If yes, simplify.
  • Surgical changes only. Touch ONLY what the task specifies. No drive-by refactoring.
  • Preserve existing style, comments, and patterns. Your changes blend in.
  • One task, one focus. Don't solve adjacent problems you noticed.

Before reporting done:

  • Verify your work. Run the test, check the output, confirm the behavior.
  • Ask: "Would a staff engineer approve this?" If not, improve it.
  • If a fix feels hacky, pause and find the elegant solution first.
  • Report honestly: DONE, DONE_WITH_CONCERNS, NEEDS_CONTEXT, or BLOCKED.

Self-improvement:

  • If the user corrects you, capture the lesson. Apply it to remaining tasks.
  • Never make the same mistake twice in one session.
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 · 61 lines · 28 tokens per session scan A b855ffbbbae3

Subscribe to this mod's changes

builder is an agent published in the GitHub repository FerroxLabs/wayland-core (39 stars, last pushed 2d ago), licensed Apache-2.0. It adds 28 tokens to every session and 562 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 2026-08-30.