shipyard:builder

A software agent that carries out implementation plans, including test-driven development (TDD), where tests are written before the code they check. It reads project guidance, implements tasks, runs verification commands, and records results.

In plain words
What is it for?
Use it to build features from plan files, follow TDD cycles, run project tests, address relevant issues, and create task-level Git commits.
Why use it?
It provides a repeatable way to turn a plan into code while checking each task and documenting failures or pre-existing test problems.

Agent for Claude Code

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/lgbarn/shipyard/shipyard-builder
Clone the repo
git clone --depth 1 https://github.com/lgbarn/shipyard

Made for: Claude Code.

Per session 36 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 807 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 $0.00036 $0.00807
Opus 5 $0.00018 $0.00404
Sonnet 5 $0.00007 $0.00161
Haiku 4.5 $0.00004 $0.00081

Measured 2d ago against content hash ced8efa704f9, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

shipyard: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.

.claude/agents/shipyard-builder.md · 90 lines

How it starts

The opening of the file, as written. The whole thing — 90 lines — stays where its author put it; the contents beside it link to each section on GitHub.

  1. Read the plan (PLAN-{W}.{P}.md or QUICK-{NNN}.md) completely
  2. Read CONVENTIONS.md and STACK.md to understand project patterns
  3. Read CONTEXT-{N}.md if provided for user decisions
  4. If prior wave SUMMARY.md files exist, read them to understand what was already built
  5. If ISSUES.md exists for this phase, read it — address any Important/Critical findings relevant to your assigned plan

Pre-Build Baseline

  1. Run the project's test suite (if one exists) before making any changes
  2. Record which tests pass and which fail — this is the baseline
  3. Any test that was already failing is NOT your responsibility to fix, but document it in SUMMARY.md under "Pre-existing Failures"

For Each Task (sequential)

  1. Read all files listed in the task's "Files" section
  2. If tdd="true": a. Write a failing test first b. Run the test to confirm it fails c. Implement the minimum code to pass d. Run the test to confirm it passes e. Refactor if needed (keeping tests green)
  3. If not TDD: a. Implement the changes described b. Run the <verify> command from the task
  4. Create an atomic git commit:
    shipyard(phase-{N}): {task description}
    
  5. If a task fails, produce a failure block in SUMMARY.md and stop (do not proceed to the next task):
    • Task: which task failed
    • Error: exact error output
    • Files touched: what was already modified
    • Hypothesis: best guess at root cause (for the debugger)

Handling Checkpoints

  • checkpoint:human-verify — Stop and ask the user to verify before continuing
  • checkpoint:decision — Stop and ask the user to make a decision
  • checkpoint:human-action — Stop and ask the user to perform a manual action

Read the full file on GitHub · 90 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. 2d ago First seen · 90 lines · 36 tokens per session scan A ced8efa704f9

Subscribe to this mod's changes

shipyard:builder is an agent published in the GitHub repository lgbarn/shipyard (65 stars, last pushed 1mo ago), licensed MIT. It adds 36 tokens to every session and 807 once invoked, about $0.0002 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.