{"slug":"no-attack-procedures","title":"No attack procedures","description":"The library does not hold procedures, commands, parameters, or payloads for compromise, credential theft, malware, or availability attacks. Naming a failure and teaching a defense are in scope. A working payload is not.","body":"# No attack procedures\n\nThis file is a policy: the limit on what the library will hold.\n\n## When to use\n\nApply this policy at intake and again at the publish gate. It is the limit on what the library will hold. A specimen that teaches a stranger to reuse a public practice can stay. A specimen that hands a stranger a working attack cannot.\n\nThe policy is public. A submitter can read it before they file, and a moderator can cite it when they refuse.\n\n## When not to\n\nDo not apply this policy to a specimen whose only offense is that it mentions a failure. Naming a failure is in scope. Teaching a defense a stranger can apply without a working payload is in scope.\n\nDo not apply it as a reason to accept a payload that has had its commentary softened. The test is whether a reader could use the text to compromise a system, steal credentials, write malware, or attack availability. Tone is irrelevant.\n\n## The specimen\n\nThe artifact is this policy.\n\n```text\npolicy: no attack procedures\napplies to: everything the public shelf holds\n\nthe library does not hold:\n  - procedures for compromise\n  - commands for compromise\n  - parameters for compromise\n  - payloads for compromise\n  - the same four shapes for credential theft\n  - the same four shapes for malware\n  - the same four shapes for availability attacks\n  - a script whose purpose is to bypass a safety check\n  - a shortened or \"educational\" copy of any of the above that a reader could still run\n\nin scope:\n  - naming a failure\n  - teaching a defense that does not include a working payload\n  - a check that goes red when a banned class appears\n  - a refusal recorded at intake\n\ndefense, which is the action this policy requires:\n  1. Do not place the procedure on the shelf.\n  2. Do not edit it down into a shorter procedure that still runs.\n  3. Record a refusal that names the class: compromise, credential theft, malware, or availability attack.\n  4. A person may publish a different specimen that names the failure and teaches the defense.\n  5. A sensor may flag the submission. The flag advises the person. It does not publish the payload by another route, and it does not excuse a person who publishes it anyway.\n\nnot a defense:\n  - wrapping the payload in a warning and publishing it\n  - moving the payload into a footnote\n  - describing the payload as fictional while leaving the steps intact\n```\n\nCredential theft here means steps that obtain a password, a key, or a session that belongs to someone else. Malware means steps that write or run software whose purpose is to harm a system the reader does not own. An availability attack means steps whose aim is to exhaust or take down a service. Compromise means steps that take over a system the reader is not allowed to control. This policy names those classes so a moderator can refuse them. It does not demonstrate them.\n\nThe two safety questions on the scorecard exist to advise that refusal. Their wording lives in the sensor question pack. A yes on the attack question can block a publish. The block supports this policy. It does not replace the person's duty to refuse, and a missing sensor reading does not make a payload legal.\n\n## How a stranger checks it\n\nA stranger checks the policy by reading it, and by applying it to two imaginary submissions without writing either attack down.\n\n1. In the fenced policy, the banned classes include compromise, credential theft, malware, and availability attacks, and they cover procedures, commands, parameters, and payloads.\n2. Naming a failure and teaching a defense are marked in scope. A working payload is marked out of scope, including a shortened copy that still runs.\n3. The defense steps are a refusal. They are not a recipe.\n4. Search this specimen for a command sequence that obtains a credential, deploys malware, exhausts a service, or takes over a system. There is no such sequence. If a later edit adds one, the policy specimen has violated itself and comes off the shelf.\n5. A submission that is only a warning plus a runnable procedure fails the policy. A submission that names \"a publish happened with no person\" and teaches the publish gate as the defense passes the policy.\n\nThe check fails closed when you cannot tell whether a block is a payload. If you cannot tell, it does not go on the shelf.\n","kind":"policy","stratum":"sense","license":"CC-BY-4.0","shot":"na","tags":["policy","safety","defense","scope"],"author_name":"Campaign Hub","author_url":"https://campaignhub.fyi/about","source_url":null,"provenance":"Original teaching specimen written for the Campaign Hub library. Not a transcript of a private run.","maturity":"practiced","audience_role":"any","safety_class":"defensive_only","harness":["generic"],"use_when":"Use this policy when deciding whether a submission may sit on the public shelf. It is the limit on attack material. A defense without a payload can stay. A working payload cannot.","avoid_when":"Do not use this policy to reject ordinary teaching, a named failure, or a defense that contains no runnable attack. Do not 'clean up' a payload into a shorter procedure that still runs.","eval_checks":[{"id":"POLICY-1","text":"The policy names the banned classes, allows a named failure and a defense without a payload, and refuses a working payload without rewriting it into a shorter procedure.","how_checked":"Read the fenced policy. Confirm the four banned classes, the in-scope defense, and the refusal. Confirm the policy text itself contains no command sequence that carries an attack out."}],"published_at":"2026-09-24T21:59:21.000Z","updated_at":"2026-09-24T21:59:21.000Z","sensor":null,"edges":[{"from":"four-scores-two-nouls","to":"no-attack-procedures","kind":"cites","title":"Four scores, two safety questions"},{"from":"human-publishes","to":"no-attack-procedures","kind":"cites","title":"A person publishes"}]}