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/phuryn/claude-usage/triagegit clone --depth 1 https://github.com/phuryn/claude-usageWhat 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 | $0.00053 | $0.03334 |
| Opus 5 | $0.00026 | $0.01667 |
| Sonnet 5 | $0.00011 | $0.00667 |
| Haiku 4.5 | $0.00005 | $0.00333 |
Grade A, and why
triage 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.
How it starts
The opening of the file, as written. The whole thing — 203 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/triage — weekly claude-usage triage
Designed to be run headless via Windows Task Scheduler (claude -p "/triage") once a week. Operates in the local working copy on the DEV branch only. Never pushes to main.
Identity & tone
- Sign every public-facing comment with
_— Claude Code & Codex collab_on its own italicized line. - Mention the version that ships next (the bumped patch version, e.g.
v1.1.1) in close messages so contributors know where to look. - Friendly, brief, honest. No emojis.
- Never dismiss a contributor's work as "wrong" when it duplicates a landed fix — they got there independently; thank them.
Hard safety rails (do not violate)
- Dirty-worktree guard. Before any
checkout/reset/merge, rungit status --porcelain. If output is non-empty, abort immediately — there is local maintainer work the routine would otherwise destroy. Do not stash, do not--force. Exit with a self-comment (step 8) explaining why the run was skipped. - Never push to
main. Final release is the maintainer's call. DEV gets pushed; main never does. - Never close a PR or issue opened by the repo owner (
gh repo view --json owner --jq .owner.login). Those are intentional, not triageable. - Never auto-merge a PR that:
- touches more than 8 files, OR
- has more than 200 added lines, OR
- mentions any of these keywords in title / body / diff (security or contract-sensitive — escalate):
auth,password,credential,cookie,token,secret,api key,oauth,bearer,login,permission,encrypt,session,csrf, OR - introduces new top-level dependencies (
requirements.txt,pyproject.toml,package.json), OR - modifies anything under
.github/workflows/,scripts/,.claude/, OR - includes deletions or renames of existing files (not just additions/edits), OR
- includes a database schema change (
init_dbbody, newCREATE TABLE, newALTER TABLE).
- Never push DEV if
python -m unittest discover -s tests -vfails onmainfirst. Baseline must be green before any work. - Never push DEV if the final test sweep after merges fails. Roll back, leave nothing on DEV.
- Stop if any external dependency is missing (
gh,codex,python) — exit cleanly with a noted error rather than partial state. - Codex sign-off is mechanical, not advisory. Before any
gh pr close/gh issue closefires, a file at/tmp/triage-codex-signoff.mdmust exist containing (a) every item on the close list, and (b) the exact phraseCodex sign-off: close list approvedon its own line. Codex generates this in step 2. No file → no closes. If Codex says "uncertain" on any item, that item stays open regardless. - If unclear, leave it open and comment. Don't guess. Surfacing as "needs maintainer review" is always preferable to a wrong close.
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 · 203 lines · 53 tokens per session scan A 8d6520f7ef1d
triage is a command published in the GitHub repository phuryn/claude-usage (2,185 stars, last pushed 1mo ago), licensed MIT. It adds 53 tokens to every session and 3,334 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 2026-08-30.
Other commands, from other repositories
git
Git operations with intelligent commit messages and workflow optimization.
checklist
Generate a custom checklist for the current feature based on user requirements.
clarify
Identify underspecified areas in the current feature spec by asking up to 5 highly targeted clarification questions and encoding answers back into the spec.
specify
Create or update the feature specification from a natural language feature description.
analyze
Perform a non-destructive cross-artifact consistency and quality analysis across spec.md, plan.md, and tasks.md after task generation.
converge
Assess the current codebase against the feature's spec, plan, and tasks, then append any remaining unbuilt work as new tasks to tasks.md so implement can complete it.