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/bengous/claude-code-plugins/prompt-healthgit clone --depth 1 https://github.com/bengous/claude-code-pluginsWrote 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/bengous/claude-code-plugins/prompt-health)<a href="https://agentmods.dev/commands/bengous/claude-code-plugins/prompt-health"><img src="https://agentmods.dev/badge/commands/bengous/claude-code-plugins/prompt-health.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 | $0.00034 | $0.01297 |
| Opus 5 | $0.00017 | $0.00648 |
| Sonnet 5 | $0.00007 | $0.00259 |
| Haiku 4.5 | $0.00003 | $0.00130 |
Grade A, and why
prompt-health 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 4d 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 — 91 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Prompt Health
Two-layer health check of a prompt artifact: a deterministic harness-staleness pass (reproducible run to run), then the vendored prompt-audit methodology for dated prompting patterns. This command never edits files — findings and the proposed diff are for the caller to apply. It is also non-interactive: state assumptions in the report; never stop to ask.
Step 1: Load the artifact
Treat the input as a file reference only when it starts with /, ./, or ~, contains a
glob metacharacter (*, ?, [), or ends in .md — resolve it with Glob and Read every
match. Otherwise treat it as inline prompt text; prose containing slashes is still prose.
Step 2: Harness staleness pass (deterministic)
Check the artifact against each row. These are Claude Code harness facts the native audit does not cover (it audits the API surface, not the harness). An artifact asserting any of these is factually wrong, not merely dated.
| Assertion in the artifact | Reality |
|---|---|
think / think hard / think harder / think more allocate thinking budget |
Not keywords. Passed through as ordinary prompt text. |
ultrathink sets maximum thinking budget |
Recognized, but it adds an in-context instruction and the effort level sent to the API is unchanged. Not a budget dial. |
budget_tokens caps thinking |
Returns 400 on Claude 4.7+. Deprecated on 4.6. Use adaptive thinking plus effort. |
| Prefill the assistant turn to steer format | Returns 400 on Claude 4.6+. Use structured outputs. |
| Workarounds that avoid or replace the word "think" | Premised on thinking being disabled. Fable 5 and Mythos 5 are thinking-always-on. |
The subagent-spawning tool is named Task |
It is Agent. Note that TaskCreate / TaskGet / TaskList / TaskUpdate are real and unrelated — do not match a bare "Task". |
| Pinned prior-generation model IDs in self-knowledge boilerplate | Current IDs: claude-fable-5, claude-mythos-5, claude-opus-5, claude-sonnet-5, claude-haiku-4-5-20251001. Prefer abstract aliases in frontmatter. |
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.
- 4d ago First seen · 91 lines · 34 tokens per session scan A b2e64a3cb0d2
prompt-health is a command published in the GitHub repository bengous/claude-code-plugins (4 stars, last pushed 2d ago), licensed MIT. It adds 34 tokens to every session and 1,297 once invoked, about $0.0002 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
pr-address
Address PR review comments on current branch.
enforce_standards
CRITICAL: Before running the code standards enforcer, we must prepare the stack properly and abort if there are merge conflicts.
OPSX: Verify
Verify implementation matches change artifacts before archiving.
OPSX: Continue
Continue working on a change - create the next artifact (Experimental).
OPSX: Sync
Sync delta specs from a change to main specs.
context
项目上下文管理：初始化 .context 目录、记录决策日志、压缩归档、查看历史.