jacobian: Skill for Codex

.agents/skills/harbor-benchmarks/SKILL.md

harbor-benchmarks is a skill for Codex from morluto/jacobian. It costs 22 tokens per session (553 once invoked), scanned A, original, MIT.

A toolkit for building and testing Harbor benchmark datasets. Harbor is an evaluation system for tasks with repeatable checks, and a verifier decides whether a result is correct.

In plain words
What is it for?
Use it to create or change Harbor tasks, independent verifiers, Oracle jobs, workflow fixtures, task summaries, and evaluation handoffs.
Why use it?
It helps turn difficult mathematical capabilities into bounded experiments that can be replayed and evaluated consistently.

Skill for Codex

Written for Codex: agents/openai.yaml present. Also seen: installed under .agents/ (shared by several agents).

This is morluto/jacobian's own configuration. It tells Codex how to work on jacobian itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything jacobian configures →

Reuse

Borrowing it

Nothing to install: this file belongs to morluto/jacobian. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/morluto/jacobian/main/.agents/skills/harbor-benchmarks/SKILL.md
Clone the repo
git clone --depth 1 https://github.com/morluto/jacobian

Made for: 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 harbor-benchmarks

README.md
[![agentmods](https://agentmods.dev/badge/skills/morluto/jacobian/harbor-benchmarks/github.svg)](https://agentmods.dev/skills/morluto/jacobian/harbor-benchmarks)
Your own site
<a href="https://agentmods.dev/skills/morluto/jacobian/harbor-benchmarks"><img src="https://agentmods.dev/badge/skills/morluto/jacobian/harbor-benchmarks/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.

agentmods 80×15 button for harbor-benchmarks

Your own site · 80×15
<a href="https://agentmods.dev/skills/morluto/jacobian/harbor-benchmarks"><img src="https://agentmods.dev/badge/skills/morluto/jacobian/harbor-benchmarks.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 22 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 553 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe. Third-party audits
  • NVIDIA SkillSpector pass 7 Sept 2026
How audits are shown
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 5.1 $0.00022 $0.00553
Opus 5 $0.00011 $0.00277
Sonnet 5 $0.00004 $0.00111
Haiku 4.5 $0.00002 $0.00055

Measured 3d ago against content hash 88ab91954e00, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-08, from the pricing page.

Security

Grade A, and why

harbor-benchmarks 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 3d 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.

.agents/skills/harbor-benchmarks/SKILL.md · 62 lines

How it starts

The opening of the file, as written. The whole thing — 62 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Harbor Benchmarks

Build a bounded mathematical experiment with a hidden, replayable verifier. Jacobian availability is an experimental condition, not a prerequisite for task admission. Task correctness depends on the submitted mathematics, not the agent's tool choices or transcript.

Choose the artifact

  • Put meaningful agent-facing mathematical tasks with replayable outcomes in Harbor, including capability gaps and tool-free baselines.
  • Put provider/environment feasibility spikes under tests/fixtures/providers/.
  • Keep public known-answer regressions and human-reviewed proof corpora separate from held-out comparative scoring.

For submission design, mathematical predicates, scoring, or verifier repairs, use verifier-evaluations. Ordinary packaging and execution do not require that skill. A task's instruction and schema must expose its complete protocol while keeping solutions, Oracle fixtures, verifier code, and host caches hidden.

Package and validate

Each task is a direct dataset child with one member record:

benchmarks/datasets/<dataset>/<task-id>/
benchmarks/datasets/<dataset>/members/<task-id>.toml

Use frozen offline inputs, pinned images and dependencies, and an explicit execution profile. Verifier Dockerfiles build from tests/; avoid parent COPY, host paths, floating tags, and symlinks. For the affected contract, use benchmark contracts. For local commands and planner selection, use the Harbor section of CONTRIBUTING.md.

make harbor-plan BASE=origin/main
make harbor-prepare-task DATASET=<dataset> TASKS="<task-id>"
make harbor-validate-task DATASET=<dataset> TASKS="<task-id>"

Preparation formats selected task Python and refreshes contracts/checksums; validation runs the selected static, host, and Oracle checks without a model agent. Run the selected verifier's behavioral attacks when its contract changes. Report any deferred affected Oracle coverage.

Read the full file on GitHub · 62 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. 3d ago Changed · -135 lines · -26 tokens per session 88ab91954e00
  2. 9d ago First seen · 197 lines · 48 tokens per session scan A 78037c6ef7cb

Subscribe to this mod's changes

harbor-benchmarks is a skill published in the GitHub repository morluto/jacobian (188 stars, last pushed today), licensed MIT. It adds 22 tokens to every session and 553 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 2026-08-30.

Related

Other skills, from other repositories

witness

Independently witness that an Allium loop's convergence claim is true and was reached honestly. Use when the user wants to verify a loop's self-report, confirm tests really pass and no generated test was weakened, produce a convergence certificate or witness record, gate CI on a trustworthy signal, or check that an…

juxt/allium · 75 tokens

coding-agents-hooks-authoring

To author, register, and test Rosetta hooks, add a SemanticKind, or debug a hook that won't fire.

griddynamics/rosetta · 31 tokens

flutter-mcp-toolkit-control

Drive a running Flutter app — tap, scroll, type, fill forms, hot-reload, navigate. Use when you need to interact with the UI.

Arenukvern/mcp_flutter · 38 tokens

harness-engineering-lifecycle

Design, implement, and integrate generalized validation harnesses across a producer-consumer boundary after a local harness contract exists. Use when refactoring custom validation CLIs/MCPs for large polyrepos, extending Steward across sibling repos, or deploying a local tool to a consumer project for dogfooding and…

Arenukvern/mcp_flutter · 88 tokens

flutter-mcp-toolkit-dogfood-iterations

Runs and records fluttertestapp dogfood iterations (toolqualityrubric, rundogfoodeval.sh, dogfoodwebeval.yaml). Use when scoring MCP/intentcall quality, appending iteration N, comparing regressions, or CI static/weekly eval gates.

Arenukvern/mcp_flutter · 66 tokens

data-collection

To gather QA source artifacts from issue tracker, wiki, test management system.

griddynamics/rosetta · 18 tokens