start

A command that opens an autoresearch experiment loop: a dedicated Git branch, saved experiment state and a measured baseline. Autoresearch means repeatedly testing changes against a fixed measurement process.

In plain words
What is it for?
Use it when explicitly starting an autoresearch run to create the experiment setup and measure its baseline before later iterations.
Why use it?
It separates real improvements from ordinary run-to-run variation and preserves a starting point for reverting experiments.

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/emaballarin/ccplugins/start
Any agent
npx skills add emaballarin/ccplugins --skill start
Clone the repo
git clone --depth 1 https://github.com/emaballarin/ccplugins

Made for: Claude Code, Codex.

Per session 166 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,248 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 $0.00166 $0.01248
Opus 5 $0.00083 $0.00624
Sonnet 5 $0.00033 $0.00250
Haiku 4.5 $0.00017 $0.00125

Measured yesterday against content hash 35f11f03ce66, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

start 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.

plugins/autoresearch/skills/start/SKILL.md · 89 lines

How it starts

The opening of the file, as written. The whole thing — 89 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/ar:start — begin an autoresearch run

Stand up the loop: dedicated branch, ./.ar/ state, a real harness, and a measured noise floor. Starting does not iterate — /ar:resume does that.

First action, always

Check for existing state before anything else:

ls -la ./.ar/ar.jsonl 2>/dev/null && git branch --show-current

If ar.jsonl already exists, this is not a fresh run. Print the status block and say so — offer /ar:resume to continue or an explicit new-segment start over. Never silently overwrite a run in progress.

Hard rules

  1. Confirm before creating. If this fired on inferred rather than stated intent, describe what opening a loop would do and confirm it is wanted before step 2. Ordinary optimisation — making something faster, tuning a few hyperparameters, running a benchmark — is handled directly. A loop is for many measured iterations under a locked harness, and it is not free: it branches and commits.
  2. Never act on a branch this run did not create. Create ar/<slug>-<DD-MM-YYYY>, carrying the current working tree across — uncommitted and untracked files included — then commit it immediately as the revert floor. Full sequence in ${CLAUDE_PLUGIN_ROOT}/references/protocol.md §0.
  3. Never fabricate a benchmark. The harness is filled in by the operator, and the loop does not start until it emits a real number. An invented benchmark makes every number downstream of it fiction.
  4. Locked harness. benchmark.sh, checks.sh, evaluator.py and the code emitting the metric are never edited, for the whole run. A score is only meaningful while the thing measuring it holds still.
  5. State lives only under ./.ar/. Nothing is written outside the target repo, and no global or shared store is touched.
  6. Defer execution. Print exact commands for long runs; do not launch training inside the session.
  7. ./.ar/ is gitignored as .ar/* plus !.ar/final_report.md — which is also what keeps git clean -fd from eating the loop's own state.

Read the full file on GitHub · 89 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. yesterday First seen · 89 lines · 166 tokens per session scan A 35f11f03ce66

Subscribe to this mod's changes

start is a skill published in the GitHub repository emaballarin/ccplugins (3 stars, last pushed 26d ago), licensed MIT. It adds 166 tokens to every session and 1,248 once invoked, about $0.0008 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.

Related

Other skills, from other repositories

systematic-debugging

Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.

obra/superpowers · 21 tokens

next-cache-components-adoption

Turn on Cache Components in a Next.js app and resolve the blocking routes it surfaces. Use when the user wants to enable, adopt, or migrate to Cache Components, flip the cacheComponents flag, work through a flood of blocking-prerender / instant validation errors, run the cache-components-instant-false codemod, or…

vercel/next.js · 95 tokens

babysit-pr

Babysit a GitHub pull request after creation by continuously polling review comments, CI checks/workflow runs, and mergeability state until the PR is merged/closed or user help is required. Diagnose failures, retry likely flaky failures up to 3 times, auto-fix/push branch-related issues when appropriate, and keep…

openai/codex · 114 tokens

imagegen

Generate or edit raster images when the task benefits from AI-created bitmap visuals such as photos, illustrations, textures, sprites, mockups, or transparent-background cutouts. Use when Codex should create a brand-new image, transform an existing image, or derive visual variants from references, and the output…

openai/codex · 113 tokens

cpu-profile-analysis

Analyze V8/Chrome CPU profiles (.cpuprofile) and DevTools trace files (Trace-.json). Use when: profiling performance, investigating slow functions, comparing code paths, finding bottlenecks, analyzing timeToRequest, understanding call trees from sampling profiler data, analyzing layout/paint/rendering, investigating…

microsoft/vscode · 71 tokens

next-cache-components-optimizer

Drive a Next.js route to instant navigation by setting up an agentic loop, under Cache Components / PPR, on initial load (hard navigation) and client-side navigation (soft navigation). Encode the goal as a failing @next/playwright instant() e2e and work it to green, one verified route at a time; the shipped test then…

vercel/next.js · 170 tokens