Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/ma-compbio-lab/SkillFoundrynpx agentmods add skills/ma-compbio-lab/skillfoundry/inspect-evaluation-harness-starterWrote 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/ma-compbio-lab/skillfoundry/inspect-evaluation-harness-starter)<a href="https://agentmods.dev/skills/ma-compbio-lab/skillfoundry/inspect-evaluation-harness-starter"><img src="https://agentmods.dev/badge/skills/ma-compbio-lab/skillfoundry/inspect-evaluation-harness-starter/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/ma-compbio-lab/skillfoundry/inspect-evaluation-harness-starter"><img src="https://agentmods.dev/badge/skills/ma-compbio-lab/skillfoundry/inspect-evaluation-harness-starter.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 | $0.00000 | $0.00281 |
| Opus 5 | $0.00000 | $0.00140 |
| Sonnet 5 | $0.00000 | $0.00056 |
| Haiku 4.5 | $0.00000 | $0.00028 |
Grade A, and why
inspect-evaluation-harness-starter 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 6d 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.
What it actually says
Inspect Evaluation Harness Starter
Use this skill to run a deterministic Inspect evaluation harness over toy scientific-agent cases and compare a candidate solver against a weaker baseline.
What This Skill Does
- defines a small local Inspect task set without external model credentials
- evaluates two deterministic solver variants on the same cases
- writes machine-readable accuracy summaries plus Inspect log files for both runs
When To Use It
- when you need a runnable
evaluation-harnesses-for-scientific-agentsstarter - when you want a local Inspect example before wiring in real agents or model-backed solvers
- when you need a stable comparison harness for repository tests
Run
./slurm/envs/agents/bin/python skills/scientific-agents-and-automation/inspect-evaluation-harness-starter/scripts/run_inspect_evaluation_harness.py \
--cases skills/scientific-agents-and-automation/inspect-evaluation-harness-starter/examples/toy_eval_cases.json \
--summary-out scratch/agents/inspect_evaluation_harness_summary.json \
--log-dir scratch/agents/inspect-eval-logs
Notes
- This starter intentionally avoids external model APIs so it can run in the repository sandbox.
- The candidate and baseline solvers are both deterministic; the purpose is to verify the harness and comparison surface, not to benchmark large models.
What ships with it
6 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.
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.
- 6d ago First seen · 30 lines · 0 tokens per session scan A 42c3a82c2d23
inspect-evaluation-harness-starter is a skill published in the GitHub repository ma-compbio-lab/SkillFoundry (38 stars, last pushed 4mo ago), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 281 tokens. 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 2026-09-03.
Other skills, from other repositories
arxiv-search
Skill "arxiv-search" from TaewoooPark/MagLab, covering category scoping, query families (3–6 per topic), version and doi resolution, tier classification and output contract.
revision-letter
Point-by-point peer-review response letter workflow for journal resubmission. Invokes RevisionLetterAgent to quote each reviewer comment verbatim, draft a response, and add a change-location marker. Outputs carry HUMAN REVIEW REQUIRED; no auto-send. A DOI or manuscript location is required for every factual response…
literature-search
Broad literature search for magnetism & spintronics — OpenAlex REST query strategy, query family generation, tier classification, and evidencematrix construction (§14.3·§14.7). Activated by the maglab lit search command and the research orchestration search-scout agent.
physics-oracle
Use when validating the dimensional, range, and conservation-law plausibility of magnetic physics quantities, or when performing deterministic physics formula calculations and unit conversions. Gilbert damping 0≤α≤1 check, M≤Ms, exchange length and domain wall width calculations, Oe↔A/m↔T·emu/cm³↔A/m·Jij meV↔K…
statistical-experimental-evaluation
Design and run statistical experiments that test the formal problem, proposed methods, theoretical predictions, baselines, and ablations.
batch-effect-correction
Use when correcting batch effects in merged bulk expression matrices with sample-level batch metadata while preserving biological group structure and generating before-and-after QC plots. NOT for: single-cell integration, raw FASTQ processing, differential expression without batch labels, or datasets without…