A development workflow that repeats research and planning, implementation, and two independent reviews. The reviews check code quality and whether the original task was fully completed.

Use it for code changes that need a documented plan, implementation, quality review, and task-completion review. It also records the research, review results, and fixes in a .dev-loop folder.

It helps catch correctness, security, maintainability, and completeness problems before work is finished. The loop continues by addressing review findings until both reviews pass.

dev-loop — Skill by swDomass

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/swdomass/ai_orchestrator/dev-loop

Any agent

npx skills add swDomass/AI_orchestrator --skill dev-loop

Clone the repo

git clone --depth 1 https://github.com/swDomass/AI_orchestrator

Made for: Claude Code, Codex.

Per session 22 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 396 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.

Model	Per session	Once invoked
Fable 5	$0.00022	$0.00396
Opus 5	$0.00011	$0.00198
Sonnet 5	$0.00004	$0.00079
Haiku 4.5	$0.00002	$0.00040

Measured 2d ago against content hash 9faa7bbc674d, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

dev-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 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.

skills/dev-loop/SKILL.md · 38 lines

What it actually says

System Prompt Addition

Implement tasks in a structured 3-phase loop:

Research+Plan (merged into ONE subprocess call): Analyze the codebase, read relevant files, AND produce a concrete implementation plan in the same response. Search the web only when local code and docs are insufficient. Output: .dev-loop/research-and-plan.md.
Execute: Implement the solution based on the research+plan output. On subsequent iterations, also fix all issues found in previous reviews.
Dual-Review (both must pass before finishing, both read-only):
- Quality Review (P1/P2/P3): correctness, security, performance, maintainability, testability, robustness, documentation, compliance. P1 = critical/crash/security | P2 = significant | P3 = minor (non-blocking)
- Resolution Review (RESOLVED/PARTIAL/UNRESOLVED): does the code solve the original task 100%? Quality is irrelevant here — only task completion matters. Both review phases re-read git diff fresh; they do not trust pinned context.

Loop repeats until both reviews pass. No auto-commit or push. Output written to .dev-loop/.

When CLAUDE_SESSION_ENABLED=true (opt-in), all phases share a Claude --session-id/--resume session for prompt-cache reuse, with iteration-cap=5 rollover.

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.

2d ago First seen · 38 lines · 22 tokens per session scan A 9faa7bbc674d

Subscribe to this mod's changes

dev-loop is a skill published in the GitHub repository swDomass/AI_orchestrator (5 stars, last pushed 16d ago), licensed MIT. It adds 22 tokens to every session and 396 once invoked, about $0.0001 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.

Code review LLM Planning automation claude-code codex coding dev feature gemini issue obsidian quality review

dev-loop

Getting it into your agent

What it costs to keep this loaded

Grade A, and why

Nothing flagged

What it actually says

System Prompt Addition

What this file has done since we first saw it

Other skills, from other repositories

code-reviewer

code-review

claude_code_delegation

plugin-review

fin-review-loop

plugin-quality