anti-slop-engineering
AGENTS.md4 starsChanged 31 days agoMITView on GitHub
Operating rules and behavioral guidelines for AI coding agents working on this repository. Anti-Slop Engineering is an open-source playbook, agent skill, and lightweight triage guard for preventing code bloat, test-oracle erosion, security shortcuts, and uncontrolled complexity in AI-assisted software development.
- Installs packages
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 33,315 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository47 other files it gives its agents.
- anti-slop-engineeringSkill
- anti-slop-engineeringSkill
- api-ergonomics-and-contract-governorSkill
- apple-craft-and-frame-budgetSkill
- apple-craft-and-interaction-physicsSkill
- asset-provenance-guardSkill
- backprop-guardSkill
- chaos-and-degradation-choreographerSkill
- chromatic-drift-guardSkill
- cognitive-load-and-attention-governorSkill
- comment-budget-guardSkill
- context-window-and-token-economics-architectSkill
- cryptographic-agility-and-secret-lifecycle-gateSkill
- decision-guardSkill
- demock-guardSkill
- design-system-drift-guardSkill
- deterministic-simulation-and-property-testingSkill
- distributed-consensus-and-concurrency-gateSkill
- finops-and-unit-economics-governorSkill
- future-scale-gateSkill
- grumpy-senior-staff-engineerSkill
- honeypot-trap-guardSkill
- house-tell-guardSkill
- install-anti-slopSkill
- llm-eval-and-prompt-injection-fortressSkill
- mandela-guardSkill
- mechanical-sympathy-and-perf-gateSkill
- mechanical-sympathy-profilerSkill
- motion-runtime-guardSkill
- observability-and-slo-architectSkill
- observability-and-slo-gateSkill
- paranoid-principal-architectSkill
- pr-boundary-guardSkill
- privacy-and-data-lineage-gateSkill
- re0-handoff-guardSkill
- reduction-guardSkill
- repo-memory-and-invariant-stewardSkill
- root-fix-gateSkill
- safe-evolution-and-rollout-choreographerSkill
- scientific-fault-isolationSkill
- styleseed-guardSkill
- supply-chain-and-sbom-sentinelSkill
- tagore-guardSkill
- type-level-and-ast-slop-gateSkill
- ui-and-copy-disciplineSkill
- verification-oracle-architectSkill
- zero-downtime-migration-gateSkill
The file
We split this AGENTS.md into the rules it gives an agent
12 rules. About 0 are common, 3 are distinctive, and 9 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
9
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 14
Run tests using python3 -m unittest discover -s tests -v.
- Distinctive: line 15
Never weaken assertions, delete tests, or add skips to bypass failures.
- Very specific: line 16
All tests in tests/test_diff_guard.py and tests/test_sre_guard.py must remain green.
- Very specific: line 17
Self-Diff Verification: Run python3 scripts/diff_guard.py --base HEAD and python3 scripts/sre_guard.py --base HEAD before concluding any …
- Distinctive: line 18
No Hallucinated Pass Badges: Only report tests and checks that were executed and verified against real terminal outputs.
- Very specific: line 35
Name the Built-in: Never hand-roll primitives provided by the platform/runtime (crypto.randomUUID(), URLSearchParams, structuredClone(), …
- Very specific: line 37
Polyglot Test Discipline: All tiers (python3 -m unittest discover -s tests -v, cargo test in services/anti-slop-rust, go test -v ./...
- Very specific: line 56
The Mandela Test-Oracle & Leakage Mandate: When authoring or evaluating test suites, benchmarks, or eval rubrics, read …
- Very specific: line 57
The Musk 5-Step Reduction & Add-Back Ledger Mandate: When reviewing PRs, simplifying architectures, or pruning dependencies, read …
- Very specific: line 59
The Upstream Invariant & Rule Backpropagation Mandate: When diagnosing test failures, PR rejections, or code smells, read …
- Distinctive: line 70
Follow Anti-Slop Engineering standards, Google SRE Discipline (KISS, DRY, YAGNI, Fix Issues at Scale), and clean-room gates:
- Very specific: line 110
Before concluding tasks, verify diff triage: bunx anti-slop-engineering diff --base HEAD (or check --base HEAD).
≈ “Common” is an estimate: at least 1 in 5 of the 42,283 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
AGENTS.md
Operating rules and behavioral guidelines for AI coding agents working on this repository.
Repository Mission
Anti-Slop Engineering is an open-source playbook, agent skill, and lightweight triage guard for preventing code bloat, test-oracle erosion, security shortcuts, and uncontrolled complexity in AI-assisted software development.
Directives for AI Agents
- Follow the Seven Gates: You MUST review and adhere to
SKILL.mdbefore generating or modifying any code in this repository. - Preserve Zero-Dependency Mandate:
scripts/diff_guard.pyandscripts/sre_guard.pyMUST remain compatible with standard library Python (>=3.9) without requiringpip installor external packages. - Protect the Test Suite:
- Run tests using
python3 -m unittest discover -s tests -v. - Never weaken assertions, delete tests, or add skips to bypass failures.
- All tests in
tests/test_diff_guard.pyandtests/test_sre_guard.pymust remain green.
- Run tests using
- Self-Diff Verification: Run
python3 scripts/diff_guard.py --base HEADandpython3 scripts/sre_guard.py --base HEADbefore concluding any work session. For multi-guard verification, runbunx anti-slop-engineering audit --base HEAD; for full repository test & lint verification, runmake check-all(orbunx anti-slop-engineering check-all). - No Hallucinated Pass Badges: Only report tests and checks that were executed and verified against real terminal outputs.
- The Canonical Grounding & Authority Invariant (Official Boilerplate & Generators First): Never improvise project scaffolding, directory layouts, configuration files, or dependencies from LLM statistical memory when an official CLI installer (
uv init,cargo init,bun create) or package registry exists. Always ground against official upstream tooling. - The Dependency Determinism Invariant (Strict Latest Stable Pinning): When adding new libraries, frameworks, or SDKs, strictly query the registry and pin the latest stable release. Never guess obsolete versions based on training cutoff dates.
- The Audience Boundary & Zero Meta-Leakage Invariant (Zero Dev-Notes Leakage in UI): Enforce complete physical separation between internal engineering meta-context (agent commentary, developer notes, TODOs, mock data banners, system instructions) and external user-facing surfaces (UI/DOM, API contracts, production logs, public documentation).
- The Visual Restraint & Single-Purpose Density Invariant (Layout Restraint & Noise Reduction): Every UI component, navbar item, button, icon, and card must justify its existence through a distinct, non-redundant primary user objective. Ban cosmetic clutter masquerading as functionality (overcrowded headers/navbars, multi-row icon spam, repetitive buttons, nested cards, ornamental badges). Prioritize purposeful white space, hierarchy, and quiet functional utility.
- The Anti-Logorrhea & Functional Economy Invariant (Zero Fluff in Code & UI): Strip away conversational chatter, redundant text blocks, over-explanatory tooltips, and obvious code narration. Let clean interfaces and self-explanatory code speak for themselves.
- The Speculative Artifact Prohibition Invariant (Zero Hallucinated Scaffolding & File Sprawl): Never generate unrequested documentation files (e.g., submission dossiers, third-party platform publishing guides, fake reviewer packets like
docs/OPENAI_SUBMISSION.md) or invent distribution workflows. Document only what is real, requested, and verifiable in the repository. Keep READMEs and docs lean, high-signal, and factual—no marketing filler or ceremonial note callouts. - The Zero-Obviousness & Anti-Truism Doctrine (Silence Over Noise): High engineering discipline values silence over noise. Never emit code comments that narrate obvious syntax (e.g.
// increment count,// return result,// check if null), tautological docstrings that merely restate symbol names, patronizing UI microcopy under self-evident controls (e.g."Please enter your email"under an Email field), or ceremonial documentation truisms (e.g. prerequisite inflation like "requires computer, internet, and text editor"). If a concept is obvious to a competent engineer or user, do not write it. - The Direct Payload & Zero Affordance Narration Mandate (Direct Technical Transition): Enforce the Direct Payload Invariant across all documentation, markdown files, PR descriptions, and UI copy. Transition directly from domain headers to payloads (tables, code blocks, media, schemas) with zero meta-preambles ("The following table outlines...", "In this section we will explore..."), zero affordance instructions ("click to play/expand/copy/scroll", "watch here without leaving the page"), zero mechanical hand-holding ("open your terminal", "save the file"), zero bilingual title-stacking (
English · عربي), and zero buzzword inflation (calling screencasts "Live Interactive Demos"). If an affordance is standard (player, chevron, copy button, table, link), never explain how to use it. - The Google Site Reliability Engineering (SRE) Mandate: Strictly enforce KISS, DRY, YAGNI, and Fix Issues at Scale across all changes:
- KISS: Maintain control flow nesting <=4 levels. Decompose functions >80 lines into cohesive, single-responsibility units. Remove trivial wrapper hops.
- DRY: Consolidate duplicate code blocks (>=4 lines) and duplicate validation patterns into canonical modules. Single source of truth is mandatory.
- YAGNI: Ban speculative parameters (
_unused_param,future_kwargs), dead branches (if False:), and placeholder comments (TODO(future)). Build for current verified contracts only. - Fix Issues at Scale: Eliminate operational toil through automation. Never use flat retries or blind
sleep()synchronization band-aids; implement exponential backoff with full jitter. Bound concurrency with pools or semaphores. Ban silent exception swallowing (except Exception: pass). Enforce idempotency on state-changing operations.
- The Clean-Room Senior Dev Ladder & SHIT Interceptor Mandate (Ponytail, STS, Boring Engineering):
- Intercept the four SHIT hazards: Scope creep (
S), Hypothetical hardening/unread companion hashes (H), Intent violation (I), and Task thrashing (T). - Climb the Senior Dev Ladder before generating code: verify necessity, reuse codebase helpers, prioritize standard library built-ins, prefer native platform features, use existing dependencies, and emit minimal sufficient code.
- Name the Built-in: Never hand-roll primitives provided by the platform/runtime (
crypto.randomUUID(),URLSearchParams,structuredClone(), etc.). - Practical DRY 3-Point Barrier: Abstract only with >=2 real existing call-sites, identical mutation requirements, and clean non-conjunction naming.
- Polyglot Test Discipline: All tiers (
python3 -m unittest discover -s tests -v,cargo testinservices/anti-slop-rust,go test -v ./...inservices/toil-guard-go) must remain 100% green.
- Intercept the four SHIT hazards: Scope creep (
- The Decision Guard Mandate: Before committing to a consequential architecture, vendor, database, or derived-metric approach, read
skills/decision-guard/SKILL.md. Search for an existing direct answer, distinguish the user objective from the proposed method, compare independent alternatives and operational horizons, and interrupt obsolete work when new evidence changes its premise. Do not apply its full decision workflow to narrow bug fixes or routine changes. - The Root-Fix Gate Mandate: When an apparently local symptom prompts repeated changes across controllers, views, APIs or layers, read
skills/root-fix-gate/SKILL.mdbefore editing. Trace the earliest incorrect invariant and responsible boundary; compare a root correction with distributed patches, protect auth/data consumers, and reconcile actual changed files against a causal plan. Do not assume every fix belongs in the ORM model or reject a necessary wide diff solely by file count. Skip this extra gate for localized fixes with known ownership. - The Future-Scale Gate Mandate: Before a hard-to-reverse runtime, public contract, schema or platform decision, read
skills/future-scale-gate/SKILL.md. Rank credible future scenarios and reversal costs; demonstrate an evolution path with a bounded experiment before calling a foundation future-ready. Preserve cheap architecture options early without implementing speculative clusters, services or migrations. Do not use today's small workload as proof that future design constraints are irrelevant, and skip this gate for routine reversible changes. - The Paranoid Principal Architect Mandate: When designing system architectures, generating backend/frontend code, refactoring critical paths, conducting adversarial security reviews, or diagnosing scale bottlenecks, race conditions, memory leaks, and EDoS vulnerabilities, read
skills/paranoid-principal-architect/SKILL.md. Enforce the 10,000,000 Concurrent Users Axiom and Active Battlefield Zero-Trust mindset: require boundary schema parsing, data-layer authorization, distributed idempotency, cursor pagination, timeouts viaAbortSignal.timeout(), exponential backoff with full jitter, and constant-time crypto. Audit assessment records withparanoid audit <file>. Do not apply to trivial UI styling or non-consequential scripts. - The Safe Evolution & Rollout Choreographer Mandate (Meta & Google SRE Coexistence Invariant): When modifying network contracts, serialization schemas (Protobuf/Thrift/JSON), database models, high-throughput execution paths, or decomposing legacy services, read
skills/safe-evolution-and-rollout-choreographer/SKILL.md. Reject the "Atomic Cutover Fantasy": enforce bidirectional wire compatibility (N-1 <-> N+1) with Tolerant Readers and additive-only changes; mandate asynchronous dark launches and shadow reads with strict side-effect fencing (zero unmocked storage mutations or third-party API calls) and statistical parity gates; govern all feature flags with sub-second kill-switches, hard TTL (<= 90 days), and automated demolition tickets; and decompose legacy monoliths exclusively via the Strangler Fig Protocol (edge facade, vertical slices, canary ramp). Audit rollout records withnpx anti-slop-engineering rollout audit <record.json> --json. - The Deterministic Simulation & Property Testing Mandate (Microsoft Research & AWS Rigor): When verifying distributed consensus, stateful lifecycles, concurrency protocols, serialization codecs, or untrusted input boundaries, read
skills/deterministic-simulation-and-property-testing/SKILL.md. Reject example-based test complacency: mandate Property-Based Testing (fast-check, Hypothesis, proptest) generating >= 1,000 inputs with automated shrinking to discover minimal failing counterexamples; mandate Deterministic Simulation Testing (DST) with a virtual clock and 64-bit reproducible seed (simulating packet drops, clock skews, and message reordering; zero unseededMath.random()or wall-clock sleeps); enforce explicit State Transition Tables with impossible states made unrepresentable in the type system; and enforce Structure-Aware Fuzzing on all parsers and deserializers asserting crash freedom, bounded recursion (<= 16), and payload byte caps (<= 10MB). Audit simulation records withnpx anti-slop-engineering sim audit <record.json> --json. - The Privacy & Data Lineage Gate Mandate (Apple Privacy & Meta Data Infra Discipline): When designing user data architectures, processing PII/sensitive customer data, implementing GDPR/CCPA deletion lifecycles, or designing analytics/telemetry pipelines, read
skills/privacy-and-data-lineage-gate/SKILL.md. Protect user data from the system itself: wrap all sensitive personal attributes in compiler-enforced branded Tainted PII types (Redacted<T>) with unredacted logging and serialization banned; mandate Cryptographic Erasure (Crypto-Shredding via KMS key-per-subject envelope encryption) ensuring account deletion instantaneously destroys the encryption key rendering distributed cold backups, append-only WALs, and multi-tier caches permanently unreadable; enforce Data Minimization & Mandatory Retention TTL by rejecting any table, document schema, or cache storing user data without an explicit purpose binding and automated TTL garbage collection; and maintain an audited Data Lineage DAG with telemetry egress strictly fenced against an allowlist and differential privacy ($\epsilon \le 1.0$). Audit privacy records withnpx anti-slop-engineering privacy audit <record.json> --json. - The Type-Level & AST Slop Gate Mandate (Oxlint & Strict AST Discipline): When writing or auditing TypeScript/JavaScript code, refactoring contracts, or configuring linters, read
skills/type-level-and-ast-slop-gate/SKILL.md. Enforce the 18 generic and 5 Effect AST rules: eliminate chained type assertions (as unknown as T), ungrounded assertions (as Twithout nearby// SAFETY:invariant justification), quadratic reducer accumulator copies, eager array.filter().map()chaining, conditional empty spreads...(c ? { x } : {}), and test-oracle erosion via module mocking (vi.mock()). Audit diffs withbunx anti-slop-engineering ast --base HEADand vendor Oxlint plugins viabunx anti-slop-engineering install-oxlint. - The API Ergonomics & Contract Governor Mandate (Microsoft TypeScript & Stripe DX): When designing public SDKs, library interfaces, CLI commands, or internal module boundaries, read
skills/api-ergonomics-and-contract-governor/SKILL.md. Enforce Pit-of-Success defaults (safe timeouts, automatic idempotent retries with exponential backoff and full jitter, bounded pagination limits); render illegal states unrepresentable via tagged discriminated unions and nominal branded types (Brand<T, B>) eliminating boolean parameter traps (fn(true, false)) and parameter transposition; structure APIs via Progressive Disclosure (1-liner common facade -> typed options bag -> raw transport/streaming power engine); and standardize all machine and network errors on RFC 9457 Problem Details with deterministic machine codes, invalid parameter paths, and automated remediation actions. Audit contract records withnpx anti-slop-engineering contract audit <record.json> --json. - The UI Craftsmanship, Copywriting & WCAG 2.2 Contrast Mandate: When generating or modifying frontend user interfaces, styling, components, or marketing/product copy, read
skills/ui-and-copy-discipline/SKILL.md. Enforce the 5 Craftsmanship Standards (C-1 Intentionality, C-2 Functional Completeness, C-3 Content-Driven Composition, C-4 Resilience & WCAG 2.2 Contrast, C-5 Performance & Motion). All visual techniques must pass the Purpose Test. Low-contrast text (<4.5:1 for normal, <3:1 for large), uniform pill buttons on inputs, decorative comment banners (// =====), generic AI gradient cliches (from-blue-500 via-purple-500 to-pink-500), and AI marketing hyperbole ("Unleash the power of AI") are strictly flagged. Verify contrast withbunx anti-slop-engineering contrast "#FFFFFF" "#1E293B"orbunx anti-slop-engineering contrast --scan src/. - The Apple Craft & Interaction Physics Mandate (Apple HIG & Design Engineering): When building, refactoring, or auditing user interfaces, transitions, animations, or client state flows, read
skills/apple-craft-and-interaction-physics/SKILL.md. Enforce the Four Physical Pillars: (1) Exhaustive UI State Matrix covering loading_skeleton, empty, partial_error, offline, and optimistic_rollback via Discriminated Unions with compile-time exhaustiveness; (2) Zero Cumulative Layout Shift (CLS = 0.000) via pre-reserved geometry for skeletons, images, dynamic error slots, and scrollbar-gutter: stable; (3) Interruptible Physics with velocity continuity ($F = -kx - cv$), immediate cancellation on user intent change, compositor-only transform/opacity, and epsilon-terminated rAF ($\epsilon \le 10^{-4}$); and (4) Keyboard & Focus Mastery with high-contrast :focus-visible rings (never unstyled outline: none), modal focus trap and restoration, roving tabindex for composite controls, and ARIA live regions. Audit records withnpx anti-slop-engineering apple-craft audit <record.json> --json. - The Observability & SLO Architect Mandate (Google Monarch & Dapper Invariant): When designing telemetry architectures, instrumenting APIs, background queues, outbox workers, or database queries, read
skills/observability-and-slo-architect/SKILL.md. Transform observability into proactive engineering: enforce OpenTelemetry by default with unbroken W3Ctraceparent(00-<trace_id>-<span_id>-<flags>) andcorrelation_idacross every synchronous RPC, message broker hop (Kafka, SQS, RabbitMQ, BullMQ), Transactional Outbox, and SQLcommenter database span; mandate Google Monarch stream pre-aggregation discipline by strictly banning dynamic identifiers (user_id,request_id,order_id,email, raw URLs) in metric labels ($\prod |L_i| \le 1,000$) and collapsing unmatched 404 routes to"__unmatched__"; measure latencies exclusively via explicit-bucket Histograms with Trace Exemplars (never Gauge averages); emit automated RED metrics (Rate, Errors, Duration) for endpoints and USE metrics (Utilization, Saturation, Errors) for resources; and ban non-impact threshold alerts (e.g. CPU > 80%), gating production alerts strictly on Google SRE Multi-Window Multi-Burn-Rate Error Budget policies (14.4x 1h/5m for paging; 6.0x 6h/30m for ticketing). Audit telemetry records withnpx anti-slop-engineering observability audit <record.json> [--source <file>] --json. - The PR Boundary & Git Topology Mandate: When authoring, reviewing, or triaging pull requests and git branches, read
skills/pr-boundary-guard/SKILL.md. Enforce branch hygiene (never open PRs directly from unbranchedmain/masterforks), bounded commit subject lines (<= 100 chars), radio-button template contracts (exactly 1 checkbox checked in mutually exclusive groups), and zero metadata logorrhea (bounded emoji and code token density). Audit records withbunx anti-slop-engineering pr-guard --base HEAD --json. - The Honeypot Trap & Prompt Canary Mandate: When configuring repository templates or triaging incoming automated AI contributions, read
skills/honeypot-trap-guard/SKILL.md. Embed deterministic HMAC-SHA256 canary tokens (<!-- anti-slop-canary:canary-xxx -->), decoy issues, and decoy checkboxes in issue/PR templates to catch ungrounded LLM crawlers with zero false positives. Template maintenance diffs under.github/are strictly exempt from triggering honeypots. Audit records withbunx anti-slop-engineering honeypot --json. - The Comment Budget & Obviousness Mandate (Zero-Obviousness & CCR Invariant): When writing or reviewing code comments across any polyglot language, read
skills/comment-budget-guard/SKILL.md. Enforce the Zero-Obviousness & Anti-Truism Doctrine: calculate Comment-to-Code Ratio ($\text{CCR} \le 0.35$ with floor $\ge 20$ comments) and strictly ban syntax play-by-play narration comments (// increment,// return result,// loop through). Formal// SAFETY:invariants, JSDoc/TSDoc docstrings,///Rust doc comments, and license headers are strictly protected. Audit diffs withbunx anti-slop-engineering comment-guard --base HEAD --json. - The Repo Memory & Invariant Steward Mandate (Architectural Memory & Zero-Drift Defense): When designing architectures, conducting PR reviews, starting tasks at Gate 0, or evolving contracts, read
skills/repo-memory-and-invariant-steward/SKILL.md. Prevent architectural amnesia across sessions: sync all historical decision records into the compact Invariant Ledger (.anti-slop/invariant-ledger.jsonand.anti-slop/invariant-ledger.toon), enforce Gate 0 ingestion of active invariants before modifying governed paths, audit PRs for architectural drift (bunx anti-slop-engineering steward drift --base HEAD), and require formal Reassessment Triggers (reassessment-trigger.json) before modifying or superseding established invariants. Audit ledger records withbunx anti-slop-engineering steward audit <file> --json. - The Cognitive Bandwidth & Attention Economy Mandate (Human & Agent Attention Governance): When communicating with users, drafting pull request descriptions, generating standalone deliverables, authoring documentation, or orchestrating multi-agent handoffs, read
skills/cognitive-load-and-attention-governor/SKILL.md. Enforce the 5 Cognitive Invariants: (1) Terminal Blocking Placement ($\mathcal{B}{pos}$) — position blocking decisions in the last non-empty block with zero trailing prose, and flag Line 1 when length $\ge 40$ words; (2) Deliverable Purity ($\mathcal{P}{deliv} = 1.0$) — strip conversational opener wrappers ('Here is...'), closer offers ('Hope this helps!'), and transitional lead-ins; (3) Text Shape Scannability ($\mathcal{S}$) — enforce Time-to-Point $TTP \le 20$ words, maximum wall length $W_{max} \le 60$ words, visual anchor density $\ge 3.0$ anchors per 100 words, and answer-first on Line 1; (4) Working Memory Bounding — cap flat lists at $\le 5$ items with progressive disclosure; and (5) Non-Regression Discipline — never drop or round off numeric thresholds, scoped conditions, or safety warnings during brevity compression. Audit records withbunx anti-slop-engineering attention --file <file> --json. - The Demock & Phantom-Fallback Mandate: When implementing or refactoring runtime logic, service clients, or auth guards, read
skills/demock-guard/SKILL.md. Never use test doubles or mocks on production paths, eliminate swallowed exceptions returning dummy defaults, fail closed on DB errors (no SQLite bypasses in Supabase apps), and ban fail-open security gates. Audit records withbunx anti-slop-engineering demock --base HEAD --json. - The Mandela Test-Oracle & Leakage Mandate: When authoring or evaluating test suites, benchmarks, or eval rubrics, read
skills/mandela-guard/SKILL.md. Eliminate self-grading tests, tautological assertions (expect(true).toBe(true)), swallowedAssertionErrors, circular mock confirmations, and frame-injected evals. Audit test suites withbunx anti-slop-engineering mandela tests/ --json. - The Musk 5-Step Reduction & Add-Back Ledger Mandate: When reviewing PRs, simplifying architectures, or pruning dependencies, read
skills/reduction-guard/SKILL.md. Question requirements, delete parts/processes before optimizing, and maintain.anti-slop/add-back-ledger.jsonlwith an add-back rate target between 10% and 25%. Audit withbunx anti-slop-engineering reduction status --json. - The Re0 State-Note & Cold-Read Shower Test Mandate: When handing off tasks across sessions, subagents, or multi-step agentic loops, read
skills/re0-handoff-guard/SKILL.md. Enforce NVIDIA's 1,500-token context reset discipline (Belcak et al. 2026). State notes must carryOBJECTIVE,PROVEN_INVARIANTS,NEGATIVE_CORPUS, and a discreteNEXT_SINGLE_ACTIONwithout ungrounded pronouns, passing the Cold-Read Shower Test before long-horizon handoffs. Audit withbunx anti-slop-engineering re0-handoff shower-test --json. - The Upstream Invariant & Rule Backpropagation Mandate: When diagnosing test failures, PR rejections, or code smells, read
skills/backprop-guard/SKILL.md. Downstream code failures and test rejections must be backpropagated to upstream rules (AGENTS.md,.anti-slop/invariant-ledger.json) rather than merely patching downstream symptoms. Audit withbunx anti-slop-engineering backprop --finding <type> --file <file> --apply. - The Design System Drift & Variant-Budget Mandate: When authoring or modifying multi-surface UI applications, components, or styles, read
skills/design-system-drift-guard/SKILL.md. Enforce component variant budgets (button=4, select=2, card=4, badge=3), calculate Design System Entropy Index ($DSEI$), eliminate orphaned single-route tokens, and strictly prohibit outline focus ring suppression (outline: none,focus:outline-none) without an explicit:focus-visiblereplacement. Audit withbunx anti-slop-engineering system-drift --scan src/ --json. - The Chromatic Saturation Drift & OKLCH Category Reflex Mandate: When specifying or auditing colors, themes, or gradients, read
skills/chromatic-drift-guard/SKILL.md. Work strictly in CIE OKLCH color space: prevent cheerful saturation drift ($\Delta C \le 0.03$), enforce committed background lightness ($|L_{\text{actual}} - L_{\text{target}}| \le 0.12$), ban the warm-cream AI default band ($L \in [0.84, 0.97], C < 0.06, H \in [40, 100]$), ban purple-blue gradient clichés ($H \in [250, 290]$), and reject unmotivated dark-mode sliding ($L < 0.15$). Audit withbunx anti-slop-engineering chroma-drift --scan src/ --target-l <L> --json. - The Gate 0 Physical Commit-Sheet & House-Tell Inversion Mandate: Before generating code for any feature, landing, or UI screen, read
skills/house-tell-guard/SKILL.md. Enforce the physical pre-commit contract (.anti-slop/commit-sheet.jsonordesign/COMMIT-SHEET.md): commit the Signature Peak, Color Tier & Lightness, Typographic Contrast Axis (rejecting Inter/Space Grotesk defaults), Grid Break, Motion Budget ($\le 3$ patterns), Category Reflex Check (1st, 2nd, and chosen deviation), and contractually break at least two habits of the agent itself (House Tells). Audit withbunx anti-slop-engineering house-tell --file .anti-slop/commit-sheet.json --json. - The DPR-2 Runtime Motion & Fillrate Sentry Mandate: When implementing or profiling client-side animations, WebGL contexts, canvas flights, or scroll state machines, read
skills/motion-runtime-guard/SKILL.md. Fullscreen post-processing costs per pixel, not per object: measure motion performance exclusively at $\text{DPR} \ge 2.0$ (retina 5.2MP) under 4x CPU throttle ($\ge 50$ FPS, LongTasks $\le 50\text{ms}$); require user gestures for unmuted audio autoplay; enforce alternative art-direction parity forprefers-reduced-motion; and filter dev-server false failures. Audit withbunx anti-slop-engineering motion-runtime --file <report.json> --json. - The Asset Sourcing & License Provenance Ledger Mandate: When sourcing or embedding static media, 3D meshes, fonts, audio, video, or icons, read
skills/asset-provenance-guard/SKILL.md. Maintain cryptographic SHA-256 asset integrity and permissible open-license provenance (CC0,MIT,OFL-1.1,Apache-2.0,Public Domain,Hand-Authored) in.anti-slop/asset-ledger.jsonl. Unverified external CDN hotlinks and unrecorded binary media are strictly prohibited. Audit withbunx anti-slop-engineering asset-provenance --scan assets/ --json. - The StyleSeed Cognitive Grammar, Token Canon DAG & Spatial Rhythm Mandate: When designing or modifying UI components, tokens, or layouts, read
skills/styleseed-guard/SKILL.md. Classify the surface into its Cognitive Output Grammar (operations-console,technical-instrument,editorial-reading,commerce-conversion,consumer-service,institutional-service,expressive-marketing); enforce the 4-Tier Token Layering DAG (Primitive -> Semantic -> Spatial/Type -> Component) and eliminate primitive color leakage; enforce 11 pairwise mathematical contrast gates; adhere to the 6-Role Spatial Rhythm Hierarchy (pageInset,sectionGap,groupGap,stackGap,inlineGap,componentInset) eliminating spatial slop and inversion anomalies; bound longitudinal multi-screen drift ($S_1 \le 0.25$); and enforce that decisions outrank defaults while exceptions can NEVER waive invariants. Audit withbunx anti-slop-engineering styleseed --scan src/ --json. - The Tagore Engineering Discourse & Substance Mandate: When drafting, editing, or auditing pull request descriptions, git commit messages, Architectural Decision Records (ADRs), documentation, state notes, or agent reasoning traces, read
skills/tagore-guard/SKILL.md. Enforce Rabindranath Tagore's literary restraint, empirical specificity, and authentic technical voice over statistical predictability and mechanical puffery: eliminate the 29 cataloged AI tells (significance inflation, copula avoidance, superficial -ing chains, rule-of-three, em-dash addiction, and promotional hype); mandate active engineering agency and empirical measurements; enforce the 8-Dimension Scoring Matrix with non-negotiable Mechanics ($\ge 35/50$) and Substance ($\ge 21/30$) subtotal floors ($\text{Total} \ge 56/80$); calibrate prose to repository ground truth via.anti-slop/voice-profile.json; and run the self-audit reflection loop before delivering. Audit withbunx anti-slop-engineering tagore --file <file> --json.
<!-- anti-slop-engineering:start -->anti-slop-engineering
Follow Anti-Slop Engineering standards, Google SRE Discipline (KISS, DRY, YAGNI, Fix Issues at Scale), and clean-room gates:
anti-slop-engineering:skills/anti-slop-engineering/SKILL.mdapi-ergonomics-and-contract-governor:skills/api-ergonomics-and-contract-governor/SKILL.mdapple-craft-and-frame-budget:skills/apple-craft-and-frame-budget/SKILL.mdapple-craft-and-interaction-physics:skills/apple-craft-and-interaction-physics/SKILL.mdasset-provenance-guard:skills/asset-provenance-guard/SKILL.mdbackprop-guard:skills/backprop-guard/SKILL.mdchromatic-drift-guard:skills/chromatic-drift-guard/SKILL.mdcognitive-load-and-attention-governor:skills/cognitive-load-and-attention-governor/SKILL.mdcomment-budget-guard:skills/comment-budget-guard/SKILL.mddecision-guard:skills/decision-guard/SKILL.mddemock-guard:skills/demock-guard/SKILL.mddesign-system-drift-guard:skills/design-system-drift-guard/SKILL.mddeterministic-simulation-and-property-testing:skills/deterministic-simulation-and-property-testing/SKILL.mdfuture-scale-gate:skills/future-scale-gate/SKILL.mdgrumpy-senior-staff-engineer:skills/grumpy-senior-staff-engineer/SKILL.mdhoneypot-trap-guard:skills/honeypot-trap-guard/SKILL.mdhouse-tell-guard:skills/house-tell-guard/SKILL.mdinstall-anti-slop:skills/install-anti-slop/SKILL.mdmandela-guard:skills/mandela-guard/SKILL.mdmechanical-sympathy-and-perf-gate:skills/mechanical-sympathy-and-perf-gate/SKILL.mdmechanical-sympathy-profiler:skills/mechanical-sympathy-profiler/SKILL.mdmotion-runtime-guard:skills/motion-runtime-guard/SKILL.mdobservability-and-slo-architect:skills/observability-and-slo-architect/SKILL.mdobservability-and-slo-gate:skills/observability-and-slo-gate/SKILL.mdparanoid-principal-architect:skills/paranoid-principal-architect/SKILL.mdpr-boundary-guard:skills/pr-boundary-guard/SKILL.mdprivacy-and-data-lineage-gate:skills/privacy-and-data-lineage-gate/SKILL.mdre0-handoff-guard:skills/re0-handoff-guard/SKILL.mdreduction-guard:skills/reduction-guard/SKILL.mdrepo-memory-and-invariant-steward:skills/repo-memory-and-invariant-steward/SKILL.mdroot-fix-gate:skills/root-fix-gate/SKILL.mdsafe-evolution-and-rollout-choreographer:skills/safe-evolution-and-rollout-choreographer/SKILL.mdscientific-fault-isolation:skills/scientific-fault-isolation/SKILL.mdtype-level-and-ast-slop-gate:skills/type-level-and-ast-slop-gate/SKILL.mdui-and-copy-discipline:skills/ui-and-copy-discipline/SKILL.mdstyleseed-guard:skills/styleseed-guard/SKILL.mdtagore-guard:skills/tagore-guard/SKILL.mdverification-oracle-architect:skills/verification-oracle-architect/SKILL.mdzero-downtime-migration-gate:skills/zero-downtime-migration-gate/SKILL.mdBefore concluding tasks, verify diff triage:bunx anti-slop-engineering diff --base HEAD(orcheck --base HEAD). For multi-guard verification, run:bunx anti-slop-engineering audit --base HEAD. To update anti-slop-engineering across all agents:bunx anti-slop-engineering update.
<!-- anti-slop-engineering:end -->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.

