resume

An experiment-loop skill that repeatedly proposes one change, measures its effect, and keeps it only when the improvement is larger than normal measurement noise.

In plain words
What is it for?
Use it to resume autoresearch runs, continue testing changes, and stop when the target, time or experiment budget is reached.
Why use it?
It lets an optimisation run continue after a session ends or its context is lost, while recording progress and avoiding changes that may have succeeded by chance.

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

Made for: Claude Code, Codex.

Per session 107 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,023 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 $0.00107 $0.01023
Opus 5 $0.00053 $0.00511
Sonnet 5 $0.00021 $0.00205
Haiku 4.5 $0.00011 $0.00102

Measured 2d ago against content hash 8d92548fab6d, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

resume 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.

plugins/autoresearch/skills/resume/SKILL.md · 78 lines

How it starts

The opening of the file, as written. The whole thing — 78 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/ar:resume — advance the loop

Iterate until the target is met, the budget is spent, or the run is stopped.

First action, always (MANDATORY)

Read state from disk before saying or doing anything else. Context is never the carrier — this is what makes the loop survive compaction and session resets.

cat ./.ar/ar.jsonl 2>/dev/null | tail -50; git branch --show-current

No ar.jsonl means no run exists — say so and point at /ar:start. Otherwise reconstruct best-so-far, run count, plateau streak and budget per ${CLAUDE_PLUGIN_ROOT}/references/resume-loop.md §1, then print the status block before the first iteration.

Hard rules

  1. Never act on a branch this run did not create. If the current branch is not config.branch, do not switch onto it — create a fresh branch from the present state and open a new segment (references/protocol.md §0).
  2. One atomic change per iteration. No compound edits. Two changes give one number and no attribution.
  3. Commit before measuring, with a Result: pending trailer; amend it with the real result on keep.
  4. Revert with git checkout -- . && git clean -fd on discard or crash. Never -fdx — that flag deletes ./.ar/ and the run with it.
  5. Locked harness. Any diff touching benchmark.sh, checks.sh, evaluator.py or the metric-emitting code is rejected and the hypothesis abandoned. Optimising a number while free to redefine it is not optimisation.
  6. Keep only above the noise floor. Improvement must exceed noiseFloor × noiseFloorMultiple; borderline candidates get a multi-seed re-run before the verdict, not after.
  7. Respect the budgetmaxRuns, maxSeconds, targetMetric.
  8. Defer execution. Print the measurement command; do not launch long jobs.
  9. Do not stop to ask permission. Once looping, keep going until the target is met, the budget is exhausted, /ar:stop fires, or interruption.

The iteration

Pick one hypothesis → apply one change → commit pending → measure → gate on checks.sh → decide keep/discard/crash/checks_failed → amend or revert → append to ar.jsonl and results.tsv, update research.md, worklog.md, ideas.md → next. Three consecutive non-improvements switch strategy family; five propose a paradigm shift. Full detail in references/protocol.md §2–§3.

Read the full file on GitHub · 78 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. 2d ago First seen · 78 lines · 107 tokens per session scan A 8d92548fab6d

Subscribe to this mod's changes

resume is a skill published in the GitHub repository emaballarin/ccplugins (3 stars, last pushed 27d ago), licensed MIT. It adds 107 tokens to every session and 1,023 once invoked, about $0.0005 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

instrument-data-to-allotrope

Convert laboratory instrument output files (PDF, CSV, Excel, TXT) to Allotrope Simple Model (ASM) JSON format or flattened 2D CSV. Use this skill when scientists need to standardize instrument data for LIMS systems, data lakes, or downstream analysis. Supports auto-detection of instrument types. Outputs include full…

anthropics/knowledge-work-plugins · 123 tokens

exploratory-data-analysis

Perform bounded, local exploratory analysis of explicitly supported scientific files. Use for redacted CSV/TSV/JSON profiles; optional NumPy, HDF5, FASTA/FASTQ, and basic image metadata inspection; missingness/leakage audits; outlier and transformation sensitivity; and rigorous EDA report scaffolds. Other domain…

K-Dense-AI/scientific-agent-skills · 83 tokens

mapping-to-snomed

Maps clinical concept spans extracted by OpenMed to SNOMED CT concepts through a USER-SUPPLIED terminology server (the user's own Ontoserver, Snowstorm, or UMLS/UTS), never a bundled vocabulary. Use when the user wants to code findings, disorders, procedures, body structures, or substances to SNOMED CT, run an ECL…

maziyarpanahi/openmed · 205 tokens

auditing-subgroup-fairness

Audit an OpenMed NER or de-identification model for performance disparities across demographic subgroups (sex, age band, race/ethnicity when available) using openmed.eval.fairnessreport. Use when the user wants per-subgroup recall and leakage, wants to check whether de-identification under-protects a group, wants to…

maziyarpanahi/openmed · 148 tokens

overleaf-sync

Two-way sync between a local paper directory and an Overleaf project, so ARIS audit/edit workflows stay on the local copy while collaborators edit in the Overleaf web UI. Use when user says "同步 overleaf", "overleaf sync", "推送到 overleaf", "connect overleaf", "Overleaf 桥接", "pull overleaf", "push overleaf", or wants to…

wanshuiyin/Auto-claude-code-research-in-sleep · 97 tokens

mixed-precision

Use FP16/BF16 mixed precision to accelerate training and reduce memory. Use when optimizing GPU performance.

aiming-lab/AutoResearchClaw · 25 tokens