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/train-executorgit 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/train-executor)<a href="https://agentmods.dev/commands/frankxai/starlight-intelligence-system/train-executor"><img src="https://agentmods.dev/badge/commands/frankxai/starlight-intelligence-system/train-executor.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 | $0.00052 | $0.03396 |
| Opus 5 | $0.00026 | $0.01698 |
| Sonnet 5 | $0.00010 | $0.00679 |
| Haiku 4.5 | $0.00005 | $0.00340 |
Grade A, and why
train-executor 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 — 219 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/train-executor
Load SIP.md, VOICES.md, genius/profile-<person-slug>.md, genius/freedom-path-<person-slug>.md, and genius/reclamation-<person-slug>.md (if present). Also load agents/starlight-genius.md and skills/intelligence/genius-excavation.md if they exist in this release. If any are missing, emit a one-line notice: genius agent/skill not yet loaded — proceeding with architect voice + Frank DNA.
Convert the DELEGATE bucket of a Freedom Path into a working playbook an executor can actually use on day one. SOPs carry the person's voice. Outdated material is flagged, not hidden. Training is phased across 12 weeks with explicit handover milestones.
Input
$ARGUMENTS
When this command fires
/discover-geniushas produced a Genius Profile + Freedom Path for the named person.- The DELEGATE bucket contains ≥3 items — enough to justify a hire.
- The person has identified an executor (named, even if hypothetical) and wants handover scaffolding.
- The person is non-technical by default. Output is readable in Claude Desktop or Notion without a terminal.
When this command does NOT fire
- Before Genius excavation — you cannot train an executor on work whose shape isn't mapped. Halt and route to
/discover-genius. - When the person's ask is really
/reclaim-knowledge— scattered material with no delegation plan. Route there first; the reclamation map is an input to this command. - When the DELEGATE bucket is empty or <3 items — either excavation is shallow (re-run
/discover-geniuswith a deeper corpus) or this person genuinely has no delegatable work (rare — route to/intakefor re-qualification). - When the ask is to automate, not delegate — AUTOMATE items belong to
/creator-pipelineor a dedicated automation command, not to a human executor playbook.
Process
- Validate inputs and the DELEGATE bucket.
- Resolve
<person-slug>from<person-name>(lowercase, kebab-case). - Read
genius/profile-<person-slug>.md. If missing → halt with:Genius Profile not found. Run /discover-genius <person-name> before /train-executor. - Read
genius/freedom-path-<person-slug>.md. If missing → halt with same message. - Count items in the DELEGATE bucket. If <3 → halt with:
DELEGATE bucket has <3 items. Either genius excavation is too shallow (re-run /discover-genius with a broader corpus), or this person has no meaningful delegatable work yet. Do not ship a playbook on thin ground. - If a focus domain was passed as third argument, filter DELEGATE items to that domain. If the filter yields zero items, halt and name the mismatch.
- Resolve
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 · 219 lines · 52 tokens per session scan A 63b8b4bc89e2
train-executor is a command published in the GitHub repository frankxai/Starlight-Intelligence-System (8 stars, last pushed yesterday), licensed MIT. It adds 52 tokens to every session and 3,396 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
issue-create
Create a structured GitHub issue from current context or description.
issue-implement
Implement a specific stage of an issue's implementation plan using feature-dev.
start
Get oriented with ArcKit — guided project onboarding, workflow selection, and command recommendations.
add-onboarding
Create user onboarding flow with guided tour. Implement step-by-step tutorial overlay using React Joyride or similar library. Add tooltips highlighting key features with descriptions. Create welcome modal introducing main functionality. Implement checklist for first-time users to complete setup tasks. Support…
at-risk-outreach
When an engagement signal trips a threshold, generate the right escalation chain — never a single coercive blast. The escalation defaults are autonomy-preserving and only escalate to manager/sponsor lanes after the lighter touches have had time to land.
close-issue
../../knowledge/procedures/close-issue-procedure.md.