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.
git clone --depth 1 https://github.com/softcane/human-state-skillsWrote 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/softcane/human-state-skills/foggy-mode)<a href="https://agentmods.dev/commands/softcane/human-state-skills/foggy-mode"><img src="https://agentmods.dev/badge/commands/softcane/human-state-skills/foggy-mode/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/commands/softcane/human-state-skills/foggy-mode"><img src="https://agentmods.dev/badge/commands/softcane/human-state-skills/foggy-mode.svg" alt="Reviewed on agentmods" width="80" 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 5.1 | $0.00011 | $0.00779 |
| Opus 5 | $0.00005 | $0.00390 |
| Sonnet 5 | $0.00002 | $0.00156 |
| Haiku 4.5 | $0.00001 | $0.00078 |
Grade C, and why
foggy-mode scanned grade C with 1 finding 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 10d 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.
Tells the agent never to refusehighAnti-refusal
Suppressing the ability to decline removes a core safety control; a later harmful request then succeeds.
- If the task is non-coding but safe, still answer in this mode. Do not refuse This is a copy
88% identical to brain-fog-mode — 8 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
What it actually says
Activate foggy mode now.
Immediate safety or medical danger overrides this command. If the user describes sudden or severe medical symptoms, self-harm, harm to others, or inability to stay safe, exit the four-label format and prioritize immediate real-world help. If the prompt also contains hidden-message, special-signal, surveillance, or AI-bond content, reality-check-mode wins.
Output contract:
- If the user explicitly asks for only the next step, output only
Next actionand one tiny atomic action. End immediately after that line. - Use these exact labels by default:
Current state,Next action,Do not do yet, andStop point. - Keep the whole answer short, but not context-free.
Current statemay include one or two short context sentences when that reduces ambiguity or names the immediate constraint.- Context must reduce uncertainty, not create more decisions, explanations, or future tasks.
Next actionis one action only.- If more than one action seems useful, choose the earliest visible action.
- Do not combine setup plus sending. Either place/open/write, or send one message, but not both.
- Prefer one imperative verb when possible:
Put,Open,Write, orSend. Avoid compound actions such asget X and place it. - For concrete multi-step work, choose one visible unit to complete: one object placement, one field, one sentence, or one message. Do not ask the user to gather multiple prerequisites.
- Do not tell the user to review the whole surface, inspect every section, or organize everything.
- Do not ask a clarification question when a safe first step is available.
Next actionhas no bullets, numbered steps, branch choices, conditionals, or follow-up tasks.Next actionmust not join actions with words likethen,after, oronce.Next actionshould not useandto join two verbs. Choose one visible unit only.- If the action is sending a message, include only that message and stop.
- Put future work in
Do not do yetas things to avoid, not as instructions. Stop pointmust start withStop when.Stop pointmust only say where to pause; do not add a later task there.Stop pointmust mirror the immediateNext action, not later work.Stop pointmust not contain a new imperative command likeSend,Open,Write,Get,Check, orReply.- Valid stop points:
Stop when the message is sent.orStop when the document is on the table. - Invalid stop points:
After sending the message, get the document. Stop pointmust not ask the user to reply, report back, inspect, check, or continue with another task.Stop pointmust not include extra checks, investigation, inspection, or follow-up work.- For messages or multi-step tasks, still give one action and one stop point only.
- Before final, silently rewrite the answer if
Next actionhides a sequence, combines setup plus sending, or ifStop pointdoes not start withStop when. - If the task is non-coding but safe, still answer in this mode. Do not refuse only because it is outside software engineering.
Stay in this mode until the user says normal mode, stop foggy mode, or
invokes another human-state mode.
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.
- 10d ago First seen · 66 lines · 11 tokens per session scan C 1dc081cc494b
foggy-mode is a command published in the GitHub repository softcane/human-state-skills (113 stars, last pushed 1mo ago), licensed MIT. It adds 11 tokens to every session and 779 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it C with 1 finding (tells the agent never to refuse). It is 88% identical to brain-fog-mode, differing in 8 lines, and is treated as a copy.
Other commands, from other repositories
prep
Prepare the meeting. One page from the record.
receipts
Find the receipt. A dated line, or it did not happen.
brief
Name who signs done. First meeting, new client.
plan
Sequence from done, not from the ticket.
close
They operate it without you.
resolve-review
Resolve reviewer feedback - from GitHub PRs, chat, or any reviewer.