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 commands/frankxai/starlight-intelligence-system/training-program-designgit clone --depth 1 https://github.com/frankxai/Starlight-Intelligence-SystemWrote 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/commands/frankxai/starlight-intelligence-system/training-program-design)<a href="https://agentmods.dev/commands/frankxai/starlight-intelligence-system/training-program-design"><img src="https://agentmods.dev/badge/commands/frankxai/starlight-intelligence-system/training-program-design.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 | $0.00068 | $0.02539 |
| Opus 5 | $0.00034 | $0.01269 |
| Sonnet 5 | $0.00014 | $0.00508 |
| Haiku 4.5 | $0.00007 | $0.00254 |
Grade A, and why
training-program-design 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 yesterday.
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 — 195 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/training-program-design
Load agents/starlight-training.md, skills/people-intelligence/learning-architecture.md, and the Curriculum Design at people-intelligence/training/<program-slug>/curriculum-<program-slug>-*.md. If no curriculum exists for this program, halt and route to /training-curriculum. Produce a Program Design wrapping curriculum into deliverable.
Input
$ARGUMENTS
Flags
--cohort-size <N>— required. Affects facilitator load, modality mix, and scenario debrief logistics.--budget <amount>— optional but recommended. Drives modality mix (synchronous-heavy = higher cost; asynchronous-heavy = lower cost but lower transfer unless paired with cohort touchpoints).--start-date <YYYY-MM-DD>— required. Anchors the spaced-retrieval schedule and measurement cadence.--facilitator-type <internal-SME | external-vendor | hybrid>— required. Affects whether/training-coach-trainermust precede launch.
Process
-
Load curriculum. Read the latest
curriculum-<program-slug>-*.md. Verify L3 + L4 anchors, spaced-retrieval schedule, cognitive-load map. If missing, halt and route to/training-curriculum. -
Cadence design. Convert the module sequence into a cadence calendar. Mix:
- Synchronous (cohort-based): kickoff, simulation/role-play sessions, debriefs, R3 cohort retrieval. Highest transfer, highest cost.
- Asynchronous (self-paced): concept introduction, light reading, R1 micro-prompts. Lowest cost, lowest transfer unless paired.
- On-job: application assignments between modules; manager observation; R4 behavior observation. Required.
-
Trainer brief. Produce a brief the facilitator can deliver from. Includes voice/tone, key talking points per module, common learner failure modes, scenario answer keys, debrief discussion prompts. If
--facilitator-type internal-SME, halt and require/training-coach-trainerbefore delivery. -
Learner journey. Map the learner's experience week-by-week from kickoff to 90-day post: what they receive, what they do, what they're observed on, what they get back as feedback.
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.
- yesterday First seen · 195 lines · 68 tokens per session scan A 4850d2df7d3a
training-program-design is a command published in the GitHub repository frankxai/Starlight-Intelligence-System (8 stars, last pushed yesterday), licensed MIT. It adds 68 tokens to every session and 2,539 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 2026-09-03.
Other commands, from other repositories
weekly
Weekly memory report — facts learned, procedures, repeated mistakes prevented.
alt
Import an Exam Radar (OPTIMETA Alt plugin) export and fold its lecture-emphasis exam signal into the course index — radar.md, a lecture-emphasis column on coverage.md, and a gold-zone weakmap.
debrief-sysdesign
Capture a debrief after a real system-design interview — the problem, your approach by stage, tradeoffs probed, curveballs, where you got stuck, and your gut outcome. Writes a structured session file under the company directory. Text-only.
hwmap
Show HW/example coverage of course sections from course-index/coverage.md. HW density = exam probability; surface the exam-hot zones.
recap
Get a summary of everything you learned this session — concepts, quizzes, and progress.
toefl-grade
TOEFL 답안/녹음 채점 — 로컬 LLM(Ollama)과 whisper.cpp로 4개 영역을 채점하여 SCORES.md에 점수를 누적 기록한다.