hyper-implement-loop

A workflow for carrying out an implementation plan, reviewing the resulting code, and fixing blocking findings in repeated rounds. It combines building and code review into one process.

In plain words
What is it for?
Use it when you want an implementation completed, checked against the main branch, and revised until the review finds no blocking problems or the workflow limit is reached.
Why use it?
It reduces the manual coordination needed to move from a plan to reviewed, corrected code.

Skill for Claude CodeCodex

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 skills/zeikar/hyperclaude/hyper-implement-loop
Any agent
npx skills add zeikar/hyperclaude --skill hyper-implement-loop
Clone the repo
git clone --depth 1 https://github.com/zeikar/hyperclaude

Made for: Claude Code, Codex.

Per session 78 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 4,280 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 $0.00078 $0.04280
Opus 5 $0.00039 $0.02140
Sonnet 5 $0.00016 $0.00856
Haiku 4.5 $0.00008 $0.00428

Measured yesterday against content hash dc7a325e8343, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

hyper-implement-loop 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 yesterday.

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.

skills/hyper-implement-loop/SKILL.md · 197 lines

How it starts

The opening of the file, as written. The whole thing — 197 lines — stays where its author put it; the contents beside it link to each section on GitHub.

hyper-implement-loop

Autonomous implement-hardening gate. Runs hyper-implement to completion first, invokes Codex code-review --base main through the bridge, and — on the FIRST round that carries blocking findings — spawns the fixer agent once to apply them, reusing that same agent via SendMessage on every later round until no blocking findings remain (judged semantically — see Step 4) or the cap is hit. A run Codex clears on its first review spawns no fixer at all. The reviewer is always the Codex bridge, never an agent — this preserves the "Claude builds, Codex reviews" invariant.

When to use

  • User typed /hyperclaude:hyper-implement-loop <plan path>.
  • User wants an autonomous implement → review → fix cycle in a single gesture.

Skip when:

  • The task is one step — use /hyperclaude:hyper-implement directly.
  • You want hands-on control over each implement / review round — use /hyperclaude:hyper-implement + /hyperclaude:hyper-code-review manually.

Failure & recovery protocol — read first

${CLAUDE_PLUGIN_ROOT}/references/loop-protocol.md carries the shared cross-loop protocol: Spawn contract, Reply transport, Correctives and transport failures, Shared anti-patterns. references/failure-protocol.md (sibling of this file) is the implement-loop binding layer: it names this loop's structured per-finding reply schema, the schema-gate accept rule, the semantic finding-map validation, the named reports, and what a transport failure preserves. Step 0 makes Reading BOTH mandatory before the loop starts.

Spawn & reply transport

See ${CLAUDE_PLUGIN_ROOT}/references/loop-protocol.mdSpawn contract for the Agent / SendMessage argument shapes, Reply transport for how each round's reply reaches the lead. Loop-specific bindings:

  • Lazy spawn: the fixer is spawned inside the first fix round (Step 5), not ahead of it. That spawn's prompt already carries round 1's blocking findings, so the spawn is itself a working round — it can mutate the tree.
  • Fixer-reply ownership: there is NO canonical output file — the fixer applies edits in place and its reply is the structured findings-map schema (finding: / status: / files-changed: / verification: / notes: per cited finding). The lead avoids reading full source bodies on the normal path, but MAY run scoped git status / git diff --stat / targeted file reads for validation and failure reporting.

Read the full file on GitHub · 197 lines

Files

What ships with it

1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

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. yesterday First seen · 197 lines · 78 tokens per session scan A dc7a325e8343

Subscribe to this mod's changes

hyper-implement-loop is a skill published in the GitHub repository zeikar/hyperclaude (3 stars, last pushed 14d ago), licensed MIT. It adds 78 tokens to every session and 4,280 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 2026-08-31.