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/robconery/cassini-workshop/exploregit clone --depth 1 https://github.com/robconery/cassini-workshopWhat 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 | $0.00019 | $0.00602 |
| Opus 5 | $0.00010 | $0.00301 |
| Sonnet 5 | $0.00004 | $0.00120 |
| Haiku 4.5 | $0.00002 | $0.00060 |
Grade A, and why
explore 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 — 61 lines — stays where its author put it; the contents beside it link to each section on GitHub.
explore
Think through an idea with me. This is a problem-space interview, not a
solution. By the end, docs/PROJECT.md says what we're building and why,
honestly including what we don't know yet.
IMPORTANT: ask if there's an existing spec written anywhere. If so, ask for it and improve on it, otherwise proceed as if nothing existed.
Scope
- IN: the problem, the user, the why-now, in/out scope, success, unknowns, light research where a fact is needed to proceed.
- OUT: architecture, schema, platform, tech choices — all
/design. If a solution idea comes up, park it under "Open questions", don't decide it.
Preflight
- Read
docs/PROJECT.mdif it exists. Summarize current state in 1–2 lines ("scope defined, no success metric yet"). Re-entrant: refine, don't restart. - Read
CLAUDE.mdanddocs/MEMORY.mdfor prior context.
Interview
Up to ~10 questions, adaptive — only ask what's missing or unresolved. Batch them (AskUserQuestion, 4 at a time → 2–3 rounds, not an interrogation). Question bank, in priority order:
- What problem, concretely? What happens today without it?
- Who is it for? Who is not?
- Why now / why worth doing?
- What's explicitly out of scope?
- What does success look like — observable, ideally measurable?
- Hard constraints (time, money, platform, must/can't use)?
- What's the riskiest unknown?
- What's the smallest version that's still worth shipping?
- Any prior art or existing thing this replaces?
- What would make you kill this?
Do targeted research only when an answer hinges on a fact you can check.
Produce
- docs/PROJECT.md (owned here): Problem, Who it's for, Goals, Success,
Scope (in/out), Constraints, Open questions, Riskiest unknowns. Mark
unresolved items
TODOrather than guessing. - CLAUDE.md: fill the basics that surfaced (domain, one-line purpose).
Update only —
/initowns the file. - docs/MEMORY.md: append a dated entry for each decision actually made here (scope cut, killed alternative, etc.). One line each, with the why.
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 · 61 lines · 19 tokens per session scan A 3b1a72a79c87
explore is a command published in the GitHub repository robconery/cassini-workshop (1 stars, last pushed 2mo ago), licensed MIT. It adds 19 tokens to every session and 602 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 2026-08-31.
Other commands, from other repositories
git-remote
Create the GitHub remote for this project and make it look sharp — name, license, real README, contributing, security, issue templates.
git-merge
Merge the current branch into main safely — commit-check, remote sync, confirm, merge.
git-commit
Stage and commit current changes using Conventional Commits, branch-or-trunk by size. Uses the github skill.
init
Scaffold the project files I like — CLAUDE.md and the /docs skeleton.
git-pr
Open a pull request with a clear summary, verification plan, and risk note via gh pr create. Uses the github skill.
git-issue
Open a GitHub issue with a clear, detailed body via gh issue create. Uses the github skill.