A build routine that implements one feature from BACKLOG.md, the project's planned-work list, and prepares it for review.

Use it to sync the main branch, create a feature branch, implement the criteria, run the required checks, and open a pull request.

It keeps development focused on one agreed feature and requires tests, project-rule checks, visual checks for interface changes, and independent verification.

loop-build — Skill for Claude Code and Codex by muzafferkadir

Install

Getting it into your agent

It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.

Clone the repo

git clone --depth 1 https://github.com/muzafferkadir/nextjs-loop-engineering-starter

agentmods

npx agentmods add skills/muzafferkadir/nextjs-loop-engineering-starter/loop-build

Made for: Claude Code, Codex.

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.

README.md

[![agentmods](https://agentmods.dev/badge/skills/muzafferkadir/nextjs-loop-engineering-starter/loop-build/github.svg)](https://agentmods.dev/skills/muzafferkadir/nextjs-loop-engineering-starter/loop-build)

Your own site

<a href="https://agentmods.dev/skills/muzafferkadir/nextjs-loop-engineering-starter/loop-build"><img src="https://agentmods.dev/badge/skills/muzafferkadir/nextjs-loop-engineering-starter/loop-build/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/skills/muzafferkadir/nextjs-loop-engineering-starter/loop-build"><img src="https://agentmods.dev/badge/skills/muzafferkadir/nextjs-loop-engineering-starter/loop-build.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 84 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 856 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

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.

Model	Per session	Once invoked
Fable 5.1	$0.00084	$0.00856
Opus 5.5	$0.00034	$0.00342
Sonnet 5.5	$0.00017	$0.00171
Haiku 4.5	$0.00008	$0.00086

Measured 1mo ago against content hash 9f5b3e319653, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

loop-build 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 1mo 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/skills/loop-build/SKILL.md · 94 lines

What it actually says

loop-build — Build Mode Skill (plan-execute-verify)

Implement ONE feature from the BACKLOG.md Backlog. Nothing else.

Workflow

1. Prepare (sync main + acquire the lock)

If STATE.md contains loop: paused — stop.
Sync main first:
```
git checkout main && git pull --ff-only
```
Dirty working tree or failed pull → do not touch anything. Report it under "Waiting on Human" in STATE.md and STOP.
Read BACKLOG.md — pick the highest-priority Backlog feature.
Check the lock:
```
bash scripts/loop-lock.sh status
```
- Held: another agent is working. STOP and report in STATE.md.
- Free:
```
bash scripts/loop-lock.sh acquire F-XXX
```
Move the feature Backlog → WIP in BACKLOG.md.
Read AGENTS.md — stack, code rules, git protocol.

2. Implement

New branch: feature/F-XXX-short-name
Implement the acceptance criteria one by one, each with its test (unit for logic, e2e for user-visible flows — see AGENTS.md test rules)
Follow the reference patterns:
- Server Action: src/app/tasks/actions.ts
- Route structure (loading.tsx, error.tsx): src/app/tasks/
- UI primitives and tokens: src/components/ui/, AGENTS.md "UI Conventions"
Smallest possible diff. One PR = one feature. Do not "improve" unrelated code — implement exactly the acceptance criteria.

3. Look at what you changed

For any UI change, run the $ui-verify skill: dev server + pnpm snap <routes> + Read the PNGs. Fix what looks wrong before invoking the verifier — the verifier checks correctness, your eyes check design.

4. Verify

bash scripts/run-verifier.sh F-XXX

On REJECT: fix and re-run. Track attempts in STATE.md:
```
Reject count F-XXX: N
```
After 5 REJECTs (Max rejects — LOOP.md ## Limits):
1. Write Escalation: F-XXX — <reason> under "Waiting on Human" in STATE.md
2. Stop the loop — wait for a human

5. Finish (release the lock)

Open a PR per the AGENTS.md git protocol (include the verifier verdict)
Move the feature WIP → Review in BACKLOG.md
```
bash scripts/loop-lock.sh release
```
Append the run to loop-run-log.md

Schema Change Gate

If the feature adds tables or columns:

Edit src/db/schema.ts, run pnpm db:push locally
Update src/db/seed.ts deterministically (fixed values only)
Mark the PR clearly: contains schema change — human approval required
Never auto-merge a schema-change PR

Hard Rules

ONE feature at a time
Never commit to main — every feature lives on its own branch and reaches main only through a PR a human merges
No denylist paths (LOOP.md ## Denylist) — if the feature seems to need one, escalate instead
No feature is "done" without tests
No PR without an APPROVE from the verifier
No auto-merge — a human merges

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.

1mo ago First seen · 94 lines · 84 tokens per session scan A 9f5b3e319653

Subscribe to this mod's changes

loop-build is a skill published in the GitHub repository muzafferkadir/nextjs-loop-engineering-starter (10 stars, last pushed 3mo ago), licensed MIT. It adds 84 tokens to every session and 856 once invoked, about $0.0003 per session on Opus 5.5. 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.

Claude Code Project management agentic-coding ai-agents ai-coding-agent codex loop-engineering

github-loop-engineering-skill

Run an evidence-gated Loop Engineering control loop for GitHub Issues and GitHub Projects through the official GitHub MCP Server, keeping every issue cold-startable for a new collaborator or a fresh agent session. Use when an agent needs to discover work state; create, triage, prioritize, assign, split, block, start…

Ruisi-Lu/github-loop-engineering-skill · 178 tokens

deliver

Drive a whole phase or feature-set to "merged + Done" — themed batching over /drive, front-loaded architectural questions, phase-gate ceremonies, and lossless resume. The long-horizon recipe ABOVE /drive; pair with native /goal " delivered" for hands-off, multi-turn completion.

firatcand/forge · 64 tokens

drive

Drive one tracker ticket end-to-end — claim, plan, implement, verify, cross-review, and ship — autonomously, pausing only for architectural decisions. The per-ticket recipe; pair with native /goal " is merged" for unattended multi-turn completion.

firatcand/forge · 56 tokens

autopilot

Take a product idea from the discovery interview all the way to a running delivery loop — drive the spec chain (BRIEF → PRD → SPEC → CONTEXT → phases.yaml → tracker), pausing ONLY at the human approval gates, then roll into /deliver. The single front door for a new forge product; pair with native /goal for hands-off…

firatcand/forge · 77 tokens

reconcile

Bidirectional sync between plans/phases.yaml and the configured tracker. --pull reads tracker → phases.yaml; --push writes phases.yaml → tracker bodies via updateIssueBody.

firatcand/forge · 37 tokens

amend-roadmap

Create a new task mid-flight — interactive field collection, then a tracker-first atomic amend (issue + relations on the tracker, materialized into phases.yaml via the reconcile pull path) with a resumable journal.

firatcand/forge · 47 tokens