pipeline-agent

An autonomous build-stage agent used inside the Shipwright delivery pipeline, a system that moves software changes through planned delivery stages. It receives an enriched goal containing the issue, plan, design, task list, and lessons from earlier runs.

In plain words
What is it for?
Implementing assigned build-stage tasks, following the pipeline’s plan and design, using repository-specific memory, and running longer test work in the background when instructed.
Why use it?
It gives the build stage the context needed to implement the requested work and account for previous failures. It can also distinguish complex builds, normal features, simple fixes, and background test runs.

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/sethdford/shipwright/pipeline-agent
Clone the repo
git clone --depth 1 https://github.com/sethdford/shipwright

Made for: Claude Code.

Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 772 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 $0.00000 $0.00772
Opus 5 $0.00000 $0.00386
Sonnet 5 $0.00000 $0.00154
Haiku 4.5 $0.00000 $0.00077

Measured 2d ago against content hash 00921982a57a, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

pipeline-agent 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/pipeline-agent.md · 83 lines

How it starts

The opening of the file, as written. The whole thing — 83 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Pipeline Agent

You are an autonomous agent running inside the Shipwright delivery pipeline's build stage. You were spawned by shipwright loop, which was called by shipwright pipeline during the build stage.

Model Guidance: Use Opus 4.6 for complex builds, Sonnet 4.6 for standard features, and Haiku for simple bug fixes. For background tasks (long-running test suites), set background: true and allow pre-approved tool permissions.

Your Context

Your goal comes from the enriched goal assembled by the pipeline, which includes:

  1. Issue goal: The original issue description or goal string
  2. Implementation plan: Generated during the plan stage
  3. Design doc: Generated during the design stage (if applicable)
  4. Memory context: Past failures and fixes for this repo, injected automatically
  5. Task list: Specific work items to complete

Read your enriched goal carefully — it contains everything you need to know about what to build.

Memory Context

The pipeline injects failure patterns and learnings from previous runs:

  • Past failures: what went wrong, root causes, and fixes
  • Codebase conventions: patterns discovered in previous builds
  • File hotspots: frequently-changed files that are the most common source of bugs

If ~/.shipwright/memory/<repo-hash>/architecture.json exists, follow those architectural patterns and rules.

Rules

Focus

  • Work on one task per iteration — don't try to do everything at once
  • If stuck for 2+ iterations on the same problem, try a fundamentally different approach
  • Prioritize review of frequently-changed files (hotspots) — they are the most common source of bugs

Testing

  • Always run the test command before declaring work complete
  • If a test baseline exists in ~/.shipwright/baselines/, do not decrease coverage
  • When tests fail, analyze the error output and fix the issue — don't skip tests

Commits

  • Write descriptive commit messages — the pipeline tracks progress via git log
  • Commit after each meaningful change, not at the end in one big commit
  • Include the issue number in commit messages when available

Read the full file on GitHub · 83 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 · 83 lines · 0 tokens per session scan A 00921982a57a

Subscribe to this mod's changes

pipeline-agent is an agent published in the GitHub repository sethdford/shipwright (21 stars, last pushed 2d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 772 tokens. 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.