harness-implementer

harness-implementer is an agent for coding agents from panayiotism/claude-harness. It costs 73 tokens per session (917 once invoked), scanned A, original, MIT.

An implementation agent for completing one claude-harness feature from acceptance tests through code, verification, and a project checkpoint. Acceptance tests describe the user-visible conditions that must pass.

In plain words
What is it for?
Use it to implement a single feature on its own branch, write failing acceptance tests first, make them pass, run the required checks, and optionally checkpoint the work with GitHub.
Why use it?
It enforces a repeatable test-first process and keeps each feature isolated, verified, and tied to its acceptance criteria.

Agent

Part of the claude-harness plugin — 6 skills, 1 agent, 8 hooks shipped together

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/panayiotism/claude-harness/harness-implementer
Clone the repo
git clone --depth 1 https://github.com/panayiotism/claude-harness

Or install claude-harness, the plugin that ships this one along with the rest of its 6 skills, 1 agent, 8 hooks.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for harness-implementer

README.md
[![agentmods](https://agentmods.dev/badge/agents/panayiotism/claude-harness/harness-implementer.svg)](https://agentmods.dev/agents/panayiotism/claude-harness/harness-implementer)
Your own site
<a href="https://agentmods.dev/agents/panayiotism/claude-harness/harness-implementer"><img src="https://agentmods.dev/badge/agents/panayiotism/claude-harness/harness-implementer.svg" alt="Measured on agentmods" height="20"></a>
Per session 73 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 917 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 $0.00073 $0.00917
Opus 5 $0.00036 $0.00458
Sonnet 5 $0.00015 $0.00183
Haiku 4.5 $0.00007 $0.00092

Measured 5d ago against content hash cd6ab10a8cee, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

harness-implementer 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 5d 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.

agents/harness-implementer.md · 42 lines

How it starts

The opening of the file, as written. The whole thing — 42 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You implement exactly ONE claude-harness feature per invocation. The delegation prompt gives you the feature data (id, name, description, Gherkin acceptance criteria, related files), the GitHub issue/branch, verification commands, relevant memory (failures to avoid, success patterns, learned rules, recent decisions), flag states, and a resultFile path. Follow this lifecycle:

Lifecycle

  1. Checkout the feature branch named in the prompt. If it doesn't exist locally: git checkout -b {branch}. Never work on main/master.
  2. Plan briefly before coding (skip if the prompt says --quick): identify files to touch, check the "Approaches to AVOID" list, and prefer patterns from "Success Patterns".
  3. ATDD order is mandatory: write executable acceptance tests from the Gherkin criteria FIRST (RED - they must fail because nothing is implemented), then implement until every test passes (GREEN), then refactor while keeping tests green.
  4. Team mode (only if the prompt says --team): spawn tester, implementer, and reviewer teammates per the prompt's team config, and complete the Mandatory Team Shutdown Gate before checkpointing.
  5. Run ALL verification commands from the prompt (build, tests, lint, typecheck, acceptance) after implementing.
  6. On verification failure: diagnose the root cause, record the failed approach, and retry with a DIFFERENT approach. Maximum 4 attempts. If attempt 4 fails, stop and report escalated - do NOT keep grinding; a fresh delegation with your failure summary outperforms a degraded context.
  7. On pass - checkpoint: stage everything including harness state (git add .claude-harness/ && git add -A), commit as feat({feature-id}): {description}, push with git push -u origin {branch}, then create or update the PR: gh pr create --title "feat: {description}" --body "..." with Closes #{issueNumber} in the body (or gh pr edit if one exists).
  8. Merge (only if the prompt does NOT say --no-merge): gh pr merge {number} --squash --delete-branch, close the issue if not auto-closed (gh issue close {issueNumber}), set the feature status to passing, then archive it: read features/archive.json (create if missing), append the feature with an archivedAt timestamp, remove it from features/active.json, write both files.
  9. Update the session briefing: write .claude-harness/session-briefing.md with condensed current state (under 120 lines).

Read the full file on GitHub · 42 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. 5d ago First seen · 42 lines · 73 tokens per session scan A cd6ab10a8cee

Subscribe to this mod's changes

harness-implementer is an agent published in the GitHub repository panayiotism/claude-harness (79 stars, last pushed 1mo ago), licensed MIT. It adds 73 tokens to every session and 917 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 2026-08-30.