capability-guard-must-assert-presence
Skill2 starsChanged 55 days agoNo licenceView on GitHub
에이전트/샌드박스의 능력을 제한하는 가드는 "금지된 것이 없음"만 검사하면 정작 자기가 존재하는 이유인 실패(능력이 0개)를 못 잡는다. 툴이 없는 에이전트는 실패하지 않고 '지어낸다'. 트리거: 툴/권한 allowlist·denylist 가드, "leaked tools" 검사, 샌드박스 격리 검증, 에이전트가 그럴듯한 거짓 결과를 success로 리턴.
In this repository188 other files it gives its agents.
- absence-guard-listing-spellings-proves-only-imaginationSkill
- absence-measurement-validity-checkSkill
- a-cached-config-layer-keeps-the-first-credential-it-sawSkill
- acceptance-gate-must-measure-delta-not-stateSkill
- a-check-that-cannot-flip-is-not-measuring-anythingSkill
- a-config-slot-cannot-hold-two-legitimate-producersSkill
- active-passive-dual-dispatch-trapSkill
- a-cut-that-does-not-compile-is-not-a-wire-cutSkill
- a-dead-entity-can-be-the-subject-of-a-live-claimSkill
- a-failed-read-degraded-to-empty-becomes-a-measurementSkill
- agent-degraded-output-suspect-execution-environmentSkill
- agent-executor-host-cwd-leakSkill
- agentic-cli-as-llm-transportSkill
- a-guard-on-the-shared-function-proves-nothing-about-its-call-sitesSkill
- ai-slop-cleanerSkill
- a-layering-contracts-refusal-names-the-right-homeSkill
- alembic-enum-create-type-trapSkill
- alembic-fresh-pg-smoke-testSkill
- alembic-phantom-revision-from-unpushed-branchSkill
- alembic-postgres-enum-migrationSkill
- alembic-revision-id-32-chars-and-pinned-headSkill
- a-living-doc-repeats-a-fact-and-updates-only-one-copySkill
- analysis-cost-reportSkill
- a-never-walked-checklist-item-may-be-unbuildableSkill
- an-isolation-flag-covers-a-mechanism-not-its-goalSkill
- a-question-back-instead-of-a-choice-means-the-menu-is-wrongSkill
- a-rejection-records-a-verdict-but-the-reason-is-what-expiresSkill
- ascii-fixture-cannot-catch-byte-vs-char-confusionSkill
- asyncio-lock-non-reentrant-deadlockSkill
- asyncpg-testing-patternsSkill
- a-timeout-names-the-waiter-not-the-cause-read-the-other-endSkill
- auth-jwt-patternsSkill
- autoplanSkill
- benchmarkSkill
- blocking-defect-masks-the-ones-behind-itSkill
- boundary-test-must-not-supply-the-answerSkill
- bsage-runtime-config-vs-settingsSkill
- bsvibe-auth-centralizationSkill
- bsvibe-llm-wrapper-not-raw-litellmSkill
- bsvibe-per-run-sandbox-isolationSkill
- build-the-missing-link-not-the-missing-systemSkill
- bulk-batch-partial-write-discardSkill
- bulk-code-transform-costs-more-than-it-savesSkill
- canarySkill
- cannot-do-x-must-be-narrowed-by-a-control-you-ownSkill
- carefulSkill
- checkpointSkill
- cli-test-stub-bypasses-url-resolutionSkill
- coding-generalSkill
- collapsing-per-item-dirs-merges-state-keyed-on-the-pathSkill
- completion-detector-fires-on-an-earlier-success-stringSkill
- compose-down-leaks-network-silentlySkill
- config-menu-offers-options-nothing-implementsSkill
- confirm-the-premise-before-offering-the-choiceSkill
- content-blog-postSkill
- csoSkill
- cwd-resets-across-a-background-task-boundarySkill
- deep-interviewSkill
- deleting-inert-config-row-disables-hidden-featureSkill
- deletion-pr-needs-an-absence-guard-and-a-controlSkill
- The first 60 shown.
The file
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.