Arbor is an autonomous research agent that proposes hypotheses, edits code, runs experiments, and retains improvements that succeed on held-out data in a growing hypothesis tree. Researchers use it to investigate problems and iteratively optimize solutions with real experiments. The catalogue skills and plugin expose Arbor's research-agent workflow to coding agents.
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.
npx skills add RUC-NLPIR/Arbor --skill arbor-agent-merge-evalgit clone --depth 1 https://github.com/RUC-NLPIR/ArborWrote 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.
[](https://agentmods.dev/skills/ruc-nlpir/arbor/arbor-agent-merge-eval)<a href="https://agentmods.dev/skills/ruc-nlpir/arbor/arbor-agent-merge-eval"><img src="https://agentmods.dev/badge/skills/ruc-nlpir/arbor/arbor-agent-merge-eval/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/ruc-nlpir/arbor/arbor-agent-merge-eval"><img src="https://agentmods.dev/badge/skills/ruc-nlpir/arbor/arbor-agent-merge-eval.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 5.1 | $0.00063 | $0.01143 |
| Opus 5 | $0.00032 | $0.00571 |
| Sonnet 5 | $0.00013 | $0.00229 |
| Haiku 4.5 | $0.00006 | $0.00114 |
Grade A, and why
arbor-agent-merge-eval 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 11d 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.
How it starts
The opening of the file, as written. The whole thing — 129 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Arbor Merge And Eval
Use this whenever scores, metadata, merge decisions, or final validation are involved.
Dataset Discipline
- B_dev is for routine iteration, executor experiments, score tracking, and idea selection.
- B_test is for milestone checks only: before merging a branch and at final report time.
- If B_test diverges from B_dev, do not hand-wave it. Investigate overfitting, noise, data split mismatch, or eval contamination.
Tree Metadata
Persist evaluation metadata early and update it after merges:
baseline_score: unmodified B_dev score.trunk_score: current trunk B_dev score.test_baseline_score: unmodified B_test score.test_trunk_score: current trunk B_test score.eval_cmd: B_dev command.eval_cmd_test: B_test command.eval_timeout,eval_retries,eval_retry_base_delay,eval_retry_max_delay.dataset_info: paths and split descriptions.metric_direction:maximizeorminimize.trunk_branch: non-protected branch that receives verified merges.submission_path,sample_submission_path.
Use {cwd} and {node_id} placeholders. Example:
cd {cwd} && uv run python run_eval.py --split dev --run-name {node_id}
Score Semantics
- Tree node
scoreis an absolute B_dev metric value. - Merge verification uses B_test.
metric_directioncontrols improvement:- maximize: higher is better.
- minimize: lower is better.
- Do not compare deltas with absolutes.
- If output has JSON with
score, prefer it. Otherwise extract the primary metric from text (primary_score,score,accuracy,acc, etc.).
Merge Procedure
Native GitMergeBranch:
- Refuses target
mainormaster. - Resolves target to configured
trunk_branch.main/masterare base branches, not merge targets. - Creates an isolated worktree at
source_branch. - Runs
eval_cmd_testwith{cwd}and{node_id}substituted. - Retries transient failures if configured.
- Extracts verified B_test score.
- Rejects the merge if B_test does not improve over
test_trunk_scoreortest_baseline_score. - Checks plugin protected paths and required outputs.
- Merges source into trunk with
--no-ff. - Reports the verified test score and instructs the coordinator to update tree metadata and node status.
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.
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.
- 11d ago First seen · 129 lines · 63 tokens per session scan A d0ba4902fd6f
arbor-agent-merge-eval is a skill published in the GitHub repository RUC-NLPIR/Arbor (1,060 stars, last pushed 2d ago), licensed Apache-2.0. It adds 63 tokens to every session and 1,143 once invoked, about $0.0003 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 2026-08-30.
Other skills, from other repositories
codex-autoresearch
Triage improvement work and run or resume accepted measured loops in a local project. Architecture, documentation, UX, product study, open research, taste, and one-shot fixes stay direct unless the user explicitly requests repeated measurement with a complete experiment contract.
skill-forge
Autonome Verbesserung nach dem Autoresearch-Paradigma (Karpathy). Zwei Modi: (1) Skill-Modus — optimiert eine SKILL.md durch iterative Mutation und Evaluation. (2) Generic-Modus — optimiert beliebige Dateien gegen jede mechanische Metrik (Testabdeckung, Bundle-Size, Lighthouse-Score, Docker-Image-Größe, etc.). Zwei…
foundry-hosted-agent-validation
Step-by-step process for validating a Python Foundry hosted agent sample (under python/samples/04-hosting/foundry-hosted-agents/) end to end — running it locally (native runtime and azd ai agent run) and after deploying it to an Azure AI Foundry project with azd. Use this when asked to validate a hosted agent sample.
build-and-test
How to build and test .NET projects in the Agent Framework repository. Use this when verifying or testing changes.
skill-release-gate
Evaluate an Agent Skill bundle for structural integrity, trigger quality, artifact improvement, script correctness, safety, installed-tree integrity, and target-host portability before release.
workflow-patterns
Use this skill when implementing tasks according to Conductor's TDD workflow, handling phase checkpoints, managing git commits for tasks, or understanding the verification protocol.