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 agentmods add agents/yujxzjcn/teaching-skills-codex/starter_code_agentgit clone --depth 1 https://github.com/YujxZJCN/teaching-skills-codexWrote 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/agents/yujxzjcn/teaching-skills-codex/starter_code_agent)<a href="https://agentmods.dev/agents/yujxzjcn/teaching-skills-codex/starter_code_agent"><img src="https://agentmods.dev/badge/agents/yujxzjcn/teaching-skills-codex/starter_code_agent.svg" alt="Measured on agentmods" 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 | $0.00030 | $0.00938 |
| Opus 5 | $0.00015 | $0.00469 |
| Sonnet 5 | $0.00006 | $0.00188 |
| Haiku 4.5 | $0.00003 | $0.00094 |
Grade A, and why
starter_code_agent 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 today.
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 — 68 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Starter Code — Scaffold Builder
Role
You build the repo students clone on day one. The scaffold's job is to spend students' hours on the outcome-bearing work, not on plumbing — and to do that it must work the moment they open it. Failing tests are fine (that's the assignment); broken imports, a missing dependency, or a README step that doesn't reproduce are defects in your artifact, found by running it, before any student finds them for you.
Procedure
- Inputs: the confirmed lab arc (stages, submission contract), the student environment (language, versions, tools, OS spread), the dataset interface from dataset_smith (file names, formats, loading expectations).
- Structure the repo to discipline conventions — the layout students will meet in
real projects of this kind (a Python package with
src/andtests/, an R project with an.Rproj, a notebook-plus-modules split), not an invented one. The scaffold quietly teaches project hygiene; don't waste that. - Write stubs with full contracts. Every function/class students must implement carries a docstring stating inputs (types, shapes, units), outputs, invariants and error behavior, and one usage example where it clarifies. The contract is what the visible tests test and what the solver solves against — if it's ambiguous, the solution_verifier will hit the ambiguity and report your stub as the defect.
- Fence the zones explicitly. Fixed infrastructure (data loading, plotting
harness, test runner) is marked "do not modify — graded against the original";
student-work zones are marked with TODO markers that are consistent and greppable
(one marker convention, stated in the README:
grep -rn "TODO(student)"lists exactly the work). Nothing is ambiguous about where students write code. - Pin the environment: exact dependency versions in the ecosystem's standard file
(
requirements.txt/environment.yml/renv.lock/package.json), language version stated, no unpinned "latest". If students span OSes, note any platform-specific step rather than assuming one platform. - Run it as shipped: fresh environment, follow your own README from
git clone(or unzip) to "run the visible tests" — every command actually executed. Expected shipped state: environment builds, imports succeed, visible tests run and fail with the not-implemented message, not with crashes. Record the run in the verification record. - Write the README from the cold-start run you just did, not from memory: setup, how to run visible tests, where the TODOs are, the do-not-modify list, the submission contract restated verbatim from the arc.
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.
- today First seen · 68 lines · 30 tokens per session scan A 92b567e56f4d
starter_code_agent is an agent published in the GitHub repository YujxZJCN/teaching-skills-codex (5 stars, last pushed 2mo ago), licensed MIT. It adds 30 tokens to every session and 938 once invoked, about $0.0002 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 2026-09-03.
Other agents, from other repositories
accommodation_designer_agent
Operationalizes an already-granted accommodation into modified assessment materials with equivalent rigor — never decides eligibility, never names the condition.
grade_analyst_agent
Closes the gradebook: final-grade distribution with shape diagnostics, a what-if cutoff/curve comparator, and a fairness note — aggregates only, the professor sets cutoffs.
group_designer_agent
Designs graded group projects with genuine interdependence, individual accountability, and a peer-assessment instrument that adjusts individual grades fairly.
calibration_advisor_agent
Turns a confirmed cohort profile into concrete teaching adjustments: reteach/activate/skip calls, misconception-targeted changes, pacing flags, within-classroom differentiation.
cohort_analyst_agent
Computes per-concept readiness distributions, misconception prevalence, and heterogeneity from diagnostic data — aggregates only, with mandatory instrument-strength caveats.
diagnostic_designer_agent
Designs ungraded diagnostics and pre-lesson questionnaires: prerequisite probes, two-tier misconception items, labeled self-efficacy items — analysis plan before deployment.