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/dragoon0x/canon/clarifygit clone --depth 1 https://github.com/Dragoon0x/canonWhat 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 | $0.00015 | $0.00459 |
| Opus 5 | $0.00008 | $0.00230 |
| Sonnet 5 | $0.00003 | $0.00092 |
| Haiku 4.5 | $0.00002 | $0.00046 |
Grade A, and why
clarify 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 3d 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
/clarify
Fix bad UI copy. Not "make it friendlier" — make it honest, specific, and actionable.
What this does
- Load
canon-ux-writing. - Identify every string: buttons, labels, placeholders, errors, empty states, success toasts, confirmations.
- Apply the rules:
- Buttons use verbs, not nouns.
Save changes, notChanges.Delete account, notAccount. - Sentence case, not Title Case (unless brand explicitly requires).
- Labels above inputs, placeholders are examples not instructions.
- Errors say three things: what failed, why, how to fix. "Invalid email" is a failure. "Email must include @ and a domain. Try [email protected]" is help.
- Empty states offer a path: default (invite action), filtered (clear filters), error (retry).
- Success toasts include Undo when reversible. Undo beats confirmation dialogs every time.
- Destructive actions use the verb, not "Yes".
Delete permanently, notYes.
- Buttons use verbs, not nouns.
- Kill the jargon:
- No "leverage", "utilize", "robust", "seamless", "cutting-edge", "empower", "unlock".
- No "please" in system messages. Politeness to software reads weird.
- No "Oops!" or "Whoops!" in errors. Users want help, not cheer.
- Voice vs tone:
- Voice (consistent): how the product always sounds.
- Tone (contextual): softer for errors, direct for confirmations, quiet for success.
- Write for scanning. Front-load the information. People read UI in an F-pattern.
Output
- Table: location | before | after | reason.
- Updated strings in context.
- Note any strings that should become dynamic (include a variable) rather than static.
Success criteria
Every button is a verb. Every error points to a fix. Every empty state offers an action. No filler words survive the pass.
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.
- 3d ago First seen · 40 lines · 15 tokens per session scan A 2c3ca58d39cd
clarify is a command published in the GitHub repository Dragoon0x/canon (5 stars, last pushed 4mo ago), licensed MIT. It adds 15 tokens to every session and 459 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
pm-status
Scan open PRs, report pipeline state, merge ready work to dev, and keep the staging promotion PR current.
review-pr
Fetch a pull request diff, run code-reviewer analysis, translate findings to owner voice, and post a single batched inline GitHub review.
codex-review
Run an adversarial review of the current branch via Codex (GPT-5.4). Default = direct Bash + thin + cheap; --verbose = wrapper-agent + 3x Opus burn.
halt-check
Smoke test — verify every hook and middleware respects the HALT kill switch. Advisory, read-only.
freeze
Activate the REA kill switch — writes .rea/HALT with a reason, blocking all governed tool calls until unfrozen.
review
Run the code-reviewer agent against the current branch's diff and produce structured findings.