dev-loop

dev-loop is a skill for Claude Code, Codex from TimothyVang/verdict-dfir. It costs 204 tokens per session (3,573 once invoked), scanned A, original, Apache-2.0.

Drive a coding task to a verified, 5-star finish with a bounded agentic OODA loop (observe, orient, plan, act) gated by TDD, an explicit Definition of Done, and a code-quality bar, then commit and open a PR to the dev remote. Runs in an isolated git worktree and keeps a durable plan/checklist on disk so an interrupted…

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/timothyvang/verdict-dfir/dev-loop
Any agent
npx skills add TimothyVang/verdict-dfir --skill dev-loop
Clone the repo
git clone --depth 1 https://github.com/TimothyVang/verdict-dfir

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.

agentmods badge for dev-loop

README.md
[![agentmods](https://agentmods.dev/badge/skills/timothyvang/verdict-dfir/dev-loop.svg)](https://agentmods.dev/skills/timothyvang/verdict-dfir/dev-loop)
Your own site
<a href="https://agentmods.dev/skills/timothyvang/verdict-dfir/dev-loop"><img src="https://agentmods.dev/badge/skills/timothyvang/verdict-dfir/dev-loop.svg" alt="Measured on agentmods" height="20"></a>
Per session 204 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 3,573 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 $0.00204 $0.03573
Opus 5 $0.00102 $0.01786
Sonnet 5 $0.00041 $0.00715
Haiku 4.5 $0.00020 $0.00357

Measured today against content hash 929408d8c221, 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 today.

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/skills/dev-loop/SKILL.md · 234 lines

How it starts

The opening of the file, as written. The whole thing — 234 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Dev Loop

Take one coding task from "asked" to "verifiably done, clean, committed, and proposed as a PR" through a short feedback loop you can defend. The point is not autonomy — it is a loop with a real acceptance gate, a quality bar, and named stopping conditions, so the work ends in a green build or an honest "blocked", never in a confident claim that isn't true.

Treat this like the project's own evidence ethic: no claim of done without a receipt you can re-run. A passing test you executed is a receipt. "It should pass" is not.

The contract (read once, then hold it the whole loop)

  • Isolate the work in a git worktree so the main checkout is never disturbed and the diff stays clean (next section).
  • Keep a resumable plan on disk and check it off as you go, so an interrupted run picks up where it left off instead of starting over.
  • Define done before you act. Write the Definition of Done as an observable, reproducible checklist first. If you cannot state the command whose exit code decides "done", you are not ready to act — go back and make the goal checkable.
  • Clear the quality bar, not just the test. Done means clean code and sound logic too — see references/quality-bar.md. A green test over sloppy code is not done.
  • One bounded change per cycle. Smallest reversible step toward the DoD, then verify.
  • Re-read fresh state before consequential actions. Don't ship against stale code.
  • Preserve unrelated work. Keep the diff surgical. Never "clean up" code the task didn't ask about.
  • Never report an error, a skipped step, or an exhausted budget as success. A red test is red. Say so, with the output.
  • Stop on a named terminal state, not on vibes (see Terminal states).

Work in an isolated worktree

Run the whole loop in a dedicated git worktree, not the main checkout. This keeps unrelated uncommitted work safe, keeps your diff clean, and lets a run be abandoned without leaving a mess. Create it from the branch you'll ship:

Read the full file on GitHub · 234 lines

Files

What ships with it

2 files 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. today First seen · 234 lines · 204 tokens per session scan A 929408d8c221

Subscribe to this mod's changes

dev-loop is a skill published in the GitHub repository TimothyVang/verdict-dfir (12 stars, last pushed yesterday), licensed Apache-2.0. It adds 204 tokens to every session and 3,573 once invoked, about $0.0010 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 2026-09-04.

Related

Other skills, from other repositories

xways-xtension-authoring

This skill should be used when the user asks to "create/scaffold a new X-Tension", "wrap a CLI tool in an X-Tension", "port a convention into an X-Tension" (helper-exe verification, Ctrl-to-save, output-dir), "audit/modernize an X-Tension", "build/compile an X-Tension", "prep an X-Tension for public release", or asks…

kev365/xways-xtension-builder-skill · 165 tokens

investigation-workflow

REQUIRED when the user says "start investigation", "investigate", "analyze case", "Read case-templates/manifest.json", references a manifest.json, or provides a SAVVYDFIR-MCP caseid. Defines the 5-phase DFIR methodology from evidence mounting through report generation, with mandatory tools, decision points, and…

kismatkunwar89/SAVVYDFIR-MCP · 84 tokens

pivot-methodology

Load when you have an initial finding and need to determine what to investigate next. Defines universal pivot chains from each artifact type to related evidence, enabling systematic investigation expansion.

kismatkunwar89/SAVVYDFIR-MCP · 37 tokens

tools-reference

Load when you need exact command syntax for SIFT Workstation tools. Covers Volatility 3, Sleuth Kit, EZ Tools, Plaso, YARA, and Regripper with actual invocation examples and output parsing guidance.

kismatkunwar89/SAVVYDFIR-MCP · 49 tokens

artifact-routing

Load when you need to determine which forensic tool to use for a specific Windows artifact type. Maps artifact categories to exact MCP tools and Bash commands using the Practical Windows Forensics taxonomy.

kismatkunwar89/SAVVYDFIR-MCP · 39 tokens

sigma-detection

Load when running sigmascan(), interpreting anomaly results, or mapping findings to MITRE ATT&CK techniques. Covers all 5 universal detectors, severity levels, and pivot patterns from each detection type.

kismatkunwar89/SAVVYDFIR-MCP · 43 tokens