Published Wednesday, September 16, 2026 · 2026-W38

shadow-reviewer-meshRun a second, independent AI agent as a concurrent read-only reviewer while your primary agent keeps working, so a second set of eyes finds gaps in real time — treating its output as an advisory lead, never as proof, until you personally reproduce the finding.Use it for: Any nontrivial change you're about to ship without another human's eyes on it, or auditing an existing 'ask another model' habit that's started treating a second opinion as proof instead of a lead to verify.Read the full skillView on GitHub

Like (0)

Add a commentPost

preplan-interrogation-gateBefore locking a plan for autonomous or walk-away execution, run one bounded round of clarifying questions covering objective, constraints, and objectively-testable acceptance criteria — stopping once another answer wouldn't change the plan — so ambiguity gets resolved before there's no one left to ask.Use it for: Handing off a defined-end-state task to unattended or long-running autonomous execution, where you won't be present to answer a question mid-run.Read the full skillView on GitHub

Like (0)

Add a commentPost

Published Tuesday, September 15, 2026 · 2026-W38

cross-agent-behavior-observationSafely learn and adapt another AI agent or assistant's working style into your own tool's workflow, using a privacy-ranked source order, a redaction pass, and verified/inferred/unverified tagging, instead of ingesting its raw transcripts or leaking its prompts.Use it for: Porting good execution habits from one AI assistant/CLI to another (autonomy boundaries, verification discipline, delegation style) without copying leaked prompts or private session content.Read the full skillView on GitHub

Like (0)

Add a commentPost

brev-cliManage GPU and CPU cloud instances from the command line with the Brev CLI — search, create, SSH/exec, editor integration, file copy, port-forwarding, and safe teardown, for ML training, fine-tuning, and inference workloads.Use it for: Provisioning or tearing down cloud GPU/CPU instances for ML training, fine-tuning, or inference work from the command line.Read the full skillView on GitHub

Like (0)

Add a commentPost

Published Monday, September 14, 2026 · 2026-W38

self-pruning-knowledge-ledgerKeep an always-loaded rules/lessons file from growing without bound by having a pipeline append one-line distilled entries, snapshot the full file before every prune, and tier older entries into on-demand deep-reference files instead of re-loading the whole history every session.Use it for: An assistant's always-loaded system prompt, rules file, or memory index that keeps growing because something keeps appending to it, and you want lessons to accumulate indefinitely without a forced choice between keeping everything inline or deleting old lessons.Read the full skillView on GitHub

Like (0)

Add a commentPost

signal-to-ticket-pipelineTurn an ambient stream of unstructured signal (a chat channel, an inbox, voice notes) into scored, classified, ticketed work items automatically — with a cost cap, per-run rate limit, hash-based dedup, a sanitize pass on every external write, and a VERIFIED / CODE-SHIPPED-NOT-VERIFIED probe before any item is called done.Use it for: A channel, inbox, or recording stream that generates real signal irregularly, where you want it triaged into tracked work without a human reading every message, and without the noise or silent-drop failure modes of an unstructured 'just watch the channel' agent.Read the full skillView on GitHub

Like (0)

Add a commentPost

Published Saturday, September 12, 2026 · 2026-W37

safeguard-fallback-rerouteRecover full-quality output when a safety classifier silently downgrades your session to a weaker fallback model, instead of accepting the fallback's degraded work as final.Use it for: Noticing a silent model downgrade mid-session and working up a reframe/decompose/invert/escalate recovery ladder instead of retrying the same prompt or accepting the fallback's output as equivalent.Read the full skillView on GitHub

Like (0)

Add a commentPost

pre-push-ship-checklist-templateA six-gate checklist for shipping a change to any stateful service — ownership map, multi-source rule audit, structural gate, regression replay, safe restart, and one real end-to-end smoke test — run before every push.Use it for: Shipping a change to a long-running stateful service where a broken deploy is expensive to notice after the fact, and building a ship checklist for a project that doesn't have one yet.Read the full skillView on GitHub

Like (0)

Add a commentPost

cas-approval-gateAn atomic compare-and-swap two-phase state machine that makes an autonomous agent's risky action provably impossible to execute without an explicit, allow-listed human's approval — closing the race-condition gap a plain "ask permission" flag leaves open.Use it for: Pipelines where an autonomous agent proposes code-writing, spending, or state-mutating actions, and reviewing an existing approval flow for a race condition between checking and acting.Read the full skillView on GitHub

Like (0)

Add a commentPost

verified-destination-routing-gateBefore posting or mutating a single external destination, classify the payload into exactly one destination class and verify that destination against a freshly-read allowlist — never a cached ID — refusing on any ambiguity.Use it for: Any bot or automation that posts to one of several possible external destinations, where sending to the wrong one would be embarrassing, unsafe, or hard to undo.Read the full skillView on GitHub

Like (0)

Add a commentPost

cross-cli-delegation-preflightOperational hygiene rules for delegating work across multiple local AI CLIs or backends — a secretless read-only probe before any cross-tool call, a free-tier-first default, opaque aliases instead of real identifiers, and scoped, idempotent grants for any mutation triggered from a secondary surface.Use it for: Any setup that routes tasks across more than one AI CLI or backend and needs consistent safety hygiene regardless of which one ends up doing the work.Read the full skillView on GitHub

Like (0)

Add a commentPost

product-design-factoryA design pipeline that grounds a product surface in one job statement, generates exactly three comparable design directions before picking one, compiles a token/component contract, builds one real vertical slice, and requires an independent hostile review before calling it done.Use it for: Designing or redesigning a product surface where you want a genuine comparison between options and a real quality bar before shipping, not just the first workable idea.Read the full skillView on GitHub

Like (0)

Add a commentPost

cross-vendor-verify-pipelineA three-role coding pipeline — one model plans, a second (different-vendor) model executes, and a third independently redoes the build/test run itself rather than reading the executor's diff — so "it passed" is never taken on the executor's word alone.Use it for: Coding tasks important enough to want a genuine second opinion on execution quality, especially work you'll hand off and not watch step-by-step.Read the full skillView on GitHub

Like (0)

Add a commentPost

Published Friday, September 11, 2026 · 2026-W37

hash-locked-plan-releaseGate execution of any nontrivial plan behind a recorded consultation, an explicit user agreement, and a content-hash lock, so a plan can never run under an agreement that was given to a different version of it.Use it for: Plan-then-execute agent workflows where a plan edit must always revoke a stale go-ahead, and pre-authorizing a proven low-risk revision class without loosening hard stops on destructive or scope-expanding changes.Read the full skillView on GitHub

Like (0)

Add a commentPost

swarm-eligibility-gateDecide whether a task actually warrants fanning out multiple parallel agents before doing it, route eligible work across three generic roles with required-return contracts, and downgrade gracefully instead of silently duplicating work when a role is unavailable.Use it for: Deciding whether a substantial task actually benefits from parallel agents before reaching for a fan-out by reflex, and choosing a safe downgrade when a swarm node times out or is unavailable.Read the full skillView on GitHub

Like (0)

Add a commentPost

Published Thursday, September 10, 2026 · 2026-W37

production-promotionMerge and verify an exact-SHA delivery candidate all the way into production — binding the merge commit, the deployed production SHA, and a live probe together — instead of treating a successful merge as a successful release.Use it for: Promoting an already-verified PR into production and proving the live site is actually serving that exact commit, not just that the merge button worked.Read the full skillView on GitHub

Like (0)

Add a commentPost

headless-agent-launch-verificationLaunch a headless/background CLI coding-agent run so it can't silently hang or lie about completing, and verify what it actually did instead of trusting its self-report.Use it for: Delegating substantial work to a background CLI coding agent, and debugging a run that came back suspiciously empty.Read the full skillView on GitHub

Like (0)

Add a commentPost

sandboxed-cli-worker-controllerRun other AI coding CLIs as least-privilege, evidence-producing workers under one controller — stripped credentials and config, pinned binaries, content-free immutable receipts, and safe process-group cancellation.Use it for: Getting an independent second CLI agent's review or fan-out work without handing it full local permissions, credentials, or config.Read the full skillView on GitHub

Like (0)

Add a commentPost

code-graph-blast-radiusBefore editing a shared symbol, class, file, or module, query a code-dependency-graph tool for its actual upstream callers instead of relying on grep alone, then compare what you actually changed against the predicted blast radius afterward.Use it for: Refactors, renames, or edits to shared utilities where you need to know what could break before you touch it.Read the full skillView on GitHub

Like (0)

Add a commentPost

delegation-brief-contractA fixed structure for the handoff brief when an orchestrating loop delegates implementation work to an agent — the mechanism, an exclusive working path, hard prohibitions, exact gates, the proof standard, the stop line, and the report format.Use it for: Writing a delegation brief that doesn't turn into rework, and verifying what a delegated agent reports back.Read the full skillView on GitHub

Like (0)

Add a commentPost

backup-recoverability-canaryBefore treating an encrypted off-site backup as grounds to delete local originals, prove round-trip recoverability with a synthetic canary and independently escrow the recovery key — instead of trusting 'upload succeeded.'Use it for: Deleting local originals after backing them up to an encrypted remote, and periodically re-checking an existing backup's recoverability.Read the full skillView on GitHub

Like (0)

Add a commentPost

interface-quality-gateA mechanical-plus-visual quality gate for any UI — an AI-tells lint pass, multi-viewport screenshot checks for overlap/clipping/responsive states, and a frontend-standards checklist — run before calling an interface done.Use it for: Reviewing any generated or edited web page, app screen, dashboard, or mobile layout for AI-slop and responsive-state defects.Read the full skillView on GitHub

Like (0)

Add a commentPost

agent-prompt-architectureA repeatable method for designing and auditing an AI agent's system prompt, always-loaded rules file, tool-use guidance, and memory/compaction instructions — identify the surface, define one behavior target, remove obsolete scaffolding, verify with the smallest probe.Use it for: Improving a system prompt or rules file that isn't producing the behavior you want, or auditing one for bloat and contradictions before adding to it.Read the full skillView on GitHub

Like (0)

Add a commentPost

Published Wednesday, September 9, 2026 · 2026-W37

terminal-ui-quality-guardrailReview terminal and text-interface changes for clear hierarchy, readable contrast, keyboard-visible state, and useful status information before treating a visual change as complete.Use it for: Reviewing a terminal theme, status line, prompt, pane layout, or text interface after a visual change.Read the full skillView on GitHub

Like (0)

Add a commentPost

safe-cross-tool-configuration-transferTransfer a useful workflow pattern between developer tools by inspecting configuration safely, translating intent into the target tool native surface, and verifying the smallest reversible change.Use it for: Adapting a rule, command, hook, skill, or integration from one developer tool to another without copying private configuration.Read the full skillView on GitHub

Like (0)

Add a commentPost

coherent-product-design-workflowTurn a product change into a coherent interface by defining the user job and states first, compiling reusable visual and interaction rules, then verifying the primary journey across real conditions.Use it for: Designing or redesigning a product surface so its states, components, accessibility, and visual rules remain coherent.Read the full skillView on GitHub

Like (0)

Add a commentPost

Published Sunday, September 6, 2026 · 2026-W37

agent-history-recallBuild a queryable local catalog of every past AI-agent session, decision, and note, then answer "what did we decide about X / why does Y work this way" via hybrid search over the catalog instead of re-reading raw transcript trees or guessing from model memory.Use it for: Answering "what did we decide about X" from your own project history instead of guessing or re-reading old transcripts.Read the full skillView on GitHub

Like (0)

Add a commentPost

ai-eval-rigor-checklistA checklist of ~20 evidence-grounded rules for evaluating AI models, agents, and RAG pipelines rigorously — eval design, retrieval-grounding tests, agent-trajectory grading, and precise language for deletion/unlearning claims.Use it for: Reviewing an eval harness, benchmark, or RAG/agent grading setup before trusting its numbers.Read the full skillView on GitHub

Like (0)

Add a commentPost

workspace-leaseA file-leasing protocol that lets multiple concurrent AI-agent sessions — across separate git worktrees or clones of the same repository — claim exclusive write ownership of exact paths before editing, with atomic scoped commits and cross-worktree overlap detection.Use it for: Running multiple AI-agent sessions against worktrees or clones of the same repo without them clobbering each other's edits.Read the full skillView on GitHub

Like (0)

Add a commentPost

graph-engineeringDesign multi-agent work as a real directed graph instead of a chain — bounded nodes with explicit input/output contracts, edges only for genuine data dependencies, a fan-out/reduce/synthesize pattern, and isolated verifiers that never share context with the work they're checking.Use it for: Designing or auditing a multi-agent fan-out pipeline for hidden fake edges and shared-context verifiers.Read the full skillView on GitHub

Like (0)

Add a commentPost

cloud-deliveryTreat "delivered" as an exact-commit-SHA state machine — bind a pushed branch to a verified CI pass and a matching preview deployment at that same SHA, and refuse to call anything shipped from a stale, mismatched, or inferred signal.Use it for: Proving a pushed branch is actually good — CI and preview both verified at the exact same commit SHA — before calling it shipped.Read the full skillView on GitHub

Like (0)

Add a commentPost

scoping-cuttingAn MVP-scoping framework combining Shape Up appetite-setting, "cut the list in half twice," an end-to-end "scooter not axle" value test, Wizard-of-Oz testing before automating, and an upfront kill-on-time commitment.Use it for: Defining MVP scope or fighting scope creep on a feature that keeps growing.Read the full skillView on GitHub

Like (0)

Add a commentPost

changelog-generatorGenerate consistent, auditable release notes directly from Conventional Commits — auto-detects the correct MAJOR/MINOR/PATCH semver bump, renders a Keep a Changelog-style document, and pairs with a CI commit-format linter and a mandatory human-approval gate.Use it for: Auto-drafting release notes and the correct semver bump straight from Conventional Commits.Read the full skillView on GitHub

Like (0)

Add a commentPost

api-design-reviewerA static REST API design review — lints naming/method/status-code/error-format conventions, diffs two spec versions to classify breaking vs. non-breaking changes, audits the versioning strategy, and produces a weighted design scorecard.Use it for: Reviewing a REST API's design consistency before it ships, or diffing two spec versions for breaking changes.Read the full skillView on GitHub

Like (0)

Add a commentPost

architecture-decision-recordsCapture architectural decisions as structured ADRs (Michael Nygard format) the moment they're made — auto-detects decision moments, extracts context/alternatives/consequences, and only ever writes a file after explicit user approval.Use it for: Capturing why a significant architecture choice was made, in a structured, approved, indexed ADR.Read the full skillView on GitHub

Like (0)

Add a commentPost

runbook-generatorGenerate a standardized operational runbook for a service — start/stop/health-check/maintenance/incident-response/rollback/escalation sections, each critical step paired with an expected-output verification check.Use it for: Writing an operational runbook for a service whose every critical step is paired with a verification check.Read the full skillView on GitHub

Like (0)

Add a commentPost

Published Thursday, September 3, 2026 · 2026-W36

observable-shareBefore anything gets shared outward, require three written answers (what's the one thing this says, what live source backs every number, what watches it and fixes it when it breaks), plus a hard word budget and a 'stale must look stale' rule.Use it for: Any report, dashboard, or status post with numbers in it, especially catching staleness that looks healthy.Read the full skillView on GitHub

Like (0)

Add a commentPost

overnight-product-factoryAn unattended build-and-deploy pipeline that only ships on a hard gate — a unique canary token must round-trip through both the QA probe and the live health probe, and a failed deploy triggers an independently-verified rollback.Use it for: Handing a defined product idea to an agent overnight and getting back a trustworthy SHIPPED/BUILT/STOPPED_AT_GATE report instead of an unverified 'done' claim.Read the full skillView on GitHub

Like (0)

Add a commentPost

second-machine-offloadTurn a second machine you own into a safe batch worker for CPU-bound jobs — a health check, a secrets-excluding file sync, and a single round-trip command with a verifiable receipt.Use it for: Offloading compiles, transcodes, test sweeps, or bulk data jobs off your primary machine without ever putting its credentials on the second one.Read the full skillView on GitHub

Like (0)

Add a commentPost

ci-triageTriage every red CI run across your repos by classifying each failure's root signature — infrastructure 'never started' vs. a genuine 'step failed' — so a rerun is only attempted on the classes it can actually fix.Use it for: Stopping a billing outage or runner regression from being misdiagnosed as a wave of unrelated flaky tests.Read the full skillView on GitHub

Like (0)

Add a commentPost

deletion-safety-gateBefore deleting a worktree, clone, or cache, run a fail-closed reference check across scheduled jobs, config files, running processes, and shared git object stores — SAFE only if every screen completes and finds zero references.Use it for: Any bulk cleanup or 'reclaim disk space' pass, so you never delete a directory a scheduled job is still executing out of.Read the full skillView on GitHub

Like (0)

Add a commentPost

gate-erosion-auditAudit an existing safety gate or verifier for five specific ways enforcement silently stops working — fail-open on error, opt-in-by-default, self-supplied evidence, a single unenforced entry point, and asymmetric normalization.Use it for: Reviewing a gate you're about to trust more, or after a 'the gate should have caught this' incident.Read the full skillView on GitHub

Like (0)

Add a commentPost

skill-suite-routerOnce a personal AI-agent skill library grows past a few dozen entries, group related skills behind thin 'suite' router skills that only route, so the top-level skill list stays short and trigger phrases stop colliding.Use it for: A skill/command library that's gotten long enough that two skills compete for the same trigger phrasing.Read the full skillView on GitHub

Like (0)

Add a commentPost

Published Monday, August 10, 2026 · 2026-W33

verification-labelsA four-label vocabulary (VERIFIED, CODE-SHIPPED-NOT-VERIFIED, BLOCKED, INCONCLUSIVE) that every completion claim from an AI agent must carry, each backed by the exact command or probe that justifies it.Use it for: Making 'done' impossible to fake in agent status reports, and turning 'are you sure?' arguments into evidence lookups.Read the full skillView on GitHub

Like (0)

Add a commentPost

compaction-contractA fixed five-section template for summarizing long AI-agent sessions that quotes exact strings (paths, errors, flags) instead of paraphrasing them.Use it for: Long agentic sessions that get compacted or handed off, so the next session resumes without re-deriving state.Read the full skillView on GitHub

Like (0)

Add a commentPost

adversarial-planningA deep-planning protocol: parallel research agents map current state and blast radius, competing designs are drafted, then a mandatory adversarial pass attacks the plan before any implementation.Use it for: Non-trivial features, refactors, migrations, and anything handed to an autonomous agent to execute unattended.Read the full skillView on GitHub

Like (0)

Add a commentPost

model-lane-routingStable role-to-model-tier lanes — the strongest model plans and verifies, a cheaper capable model executes, a different-vendor long-context model judges — with live-pricing re-verification before cost-based decisions.Use it for: Designing multi-model pipelines, monthly cost reviews, and re-mapping lanes when a new model generation ships.Read the full skillView on GitHub

Like (0)

Add a commentPost

fanout-cost-pinningExplicitly pin every spawned subagent to the cheapest adequate model tier, closing the silent-inheritance leak where parallel readers run on the apex-priced main-loop model.Use it for: Anything that fans out parallel subagents, and investigating a surprising AI bill.Read the full skillView on GitHub

Like (0)

Add a commentPost

multi-cli-constitutionKeep multiple AI coding CLIs behaviorally consistent with one shared constitution file, a capability routing table with polyfill recipes, and a compiled policy block injected identically into each tool's config.Use it for: The day you add a second AI coding CLI, or after two tools behave differently on the same rule.Read the full skillView on GitHub

Like (0)

Add a commentPost

background-agent-rosterThe twelve standing background agents worth running on any machine hosting unattended AI automation, plus the receipt-file and whitelisted-autofix discipline that keeps the roster honest.Use it for: Growing a scheduled-automation estate without silent multi-week outages.Read the full skillView on GitHub

Like (0)

Add a commentPost

agent-memory-syncA git-backed persistent memory layout for AI agents — one fact per file with typed frontmatter, an always-loaded index, and a scheduled commit/rebase/push sync job.Use it for: Making what an agent learned survive session boundaries, machine switches, and agent-tool switches.Read the full skillView on GitHub

Like (0)

Add a commentPost

Published Sunday, July 26, 2026 · 2026-W31

chat-platform-agent-bridgeWire an external chat platform (Telegram, Discord, SMS, etc.) into an AI agent so you can send it a prompt from your phone and get the reply back in the same chat, with untrusted-input quarantining and sender allow-listing built in.Use it for: Sending your agent a prompt from your phone while away from your desk, or building a lightweight remote control for an unattended agent without exposing a web endpoint.Read the full skillView on GitHub

Like (0)

Add a commentPost

production-quality-autopatch-loopContinuously score live production conversation/interaction transcripts against a quality rubric, auto-draft a minimal prompt-only fix as a pull request, and auto-merge it only once it clears a strict multi-reviewer safety gate.Use it for: A deployed conversational or text-generating product where you want quality regressions caught from real traffic and small prompt fixes shipped automatically, never auto-merging an unreviewed or oversized change.Read the full skillView on GitHub

Like (0)

Add a commentPost

agent-burst-resource-prepA pre-flight routine that frees RAM and drops memory pressure on your machine so it can safely run a larger batch of parallel AI-agent sessions than usual, then restores normal state afterward.Use it for: Right before spawning a larger-than-normal parallel AI-agent fan-out, or after noticing a big parallel run swapping/thrashing instead of speeding up.Read the full skillView on GitHub

Like (0)

Add a commentPost

Published Sunday, July 19, 2026 · 2026-W30

architecture-invariant-auditA read-only, dependency-free script that checks a layered codebase against documented architecture rules (layer crossings, legacy reuse, anti-pattern regressions, doc drift) and fails CI when they're violated.Use it for: Enforcing a layered architecture boundary, tracking a strangler-pattern legacy file's shrink, or catching a previously-fixed anti-pattern before it regresses elsewhere.Read the full skillView on GitHub

Like (0)

Add a commentPost

policy-prompt-drift-checkCompares a code-level policy module against the natural-language prompt it governs, flags missing anchors and direct ALLOW/FORBID contradictions, and emits a per-policy ALIGNED/WARN/DRIFT verdict without editing anything.Use it for: Before merging a PR that touches a policy/constants module shared with a prompt file, or after any prompt rewrite.Read the full skillView on GitHub

Like (0)

Add a commentPost

git-identity-guardPre-push check that verifies the active git author email, authenticated gh CLI account, and target remote all match what you declared for this repo, and refuses to push on any mismatch or missing declaration.Use it for: Working across a personal and a work GitHub account (or several client accounts) on the same machine.Read the full skillView on GitHub

Like (0)

Add a commentPost

safe-project-retirementEnumerate every surface tied to a dead project (repos, scheduled jobs, dashboards, notes), get human confirmation, then archive everything reversibly with an undo manifest instead of deleting it.Use it for: Cleaning up a finished or abandoned project without hunting down every cron job and dashboard entry by hand.Read the full skillView on GitHub

Like (0)

Add a commentPost

cross-repo-work-recapPull a structured, evidence-backed recap (shipped / in-flight / at-risk) of work across multiple local git repos over a time window, cross-referenced with personal notes.Use it for: Answering 'what did you actually get done this week' across several repos without reconstructing it from memory.Read the full skillView on GitHub

Like (0)

Add a commentPost

agent-observability-bundleBuild a fully-local, zero-cost observability bundle for an AI coding agent — a status line, a hallucination-blocking turn-end hook, and an optional periodic quality grader.Use it for: Getting visibility into context usage, cost, and unverified 'done' claims from your AI coding agent without a paid monitoring service.Read the full skillView on GitHub

Like (0)

Add a commentPost

text-to-image-bakeoffWraps OpenAI gpt-image-1, Fal's Flux Pro, and Google Imagen text-to-image APIs behind one small script with a bake-off comparison mode, backend health tracking, and a reusable prompt template.Use it for: Generating hero art or concept backdrops when no image-generation MCP/tool is available, or comparing providers before picking a default.Read the full skillView on GitHub

Like (0)

Add a commentPost

evidence-gated-delegationBefore letting a secondary agent, worker fleet, or automation pipeline act on your behalf, require a proven track record plus deterministic post-hoc verification — never trust the delegate's own self-report of success.Use it for: Wiring up a second agent or automation pipeline that will act without your eyes on every step.Read the full skillView on GitHub

Like (0)

Add a commentPost

legitimate-work-rerouteRecover full-quality output when a safety classifier misreads legitimate, authorized defensive-security work as an attack because of how the request is worded, without ever using this to evade a genuine restriction.Use it for: An authorized defensive-security task comes back visibly weaker or refused because it was worded in attack-shaped language.Read the full skillView on GitHub

Like (0)

Add a commentPost

environment-drift-hunterRuns a scheduled and on-demand battery of deterministic, no-LLM checks over a local dev/automation setup and surfaces a ranked, evidence-backed finding list with concrete fixes and an autofix-safe flag.Use it for: Catching doc-vs-code drift, orphan jobs, security-default drift, and committed secrets before they turn into a real incident.Read the full skillView on GitHub

Like (0)

Add a commentPost

Published Sunday, July 12, 2026 · 2026-W29

negative-prompt-leakageWhen writing FORBIDDEN/NEVER/✗ example blocks in an LLM prompt, never spell out the literal forbidden phrase — models pattern-match into the bad example and reproduce it; use a short abstract marker instead.Use it for: Prompts where the model keeps producing an output you explicitly told it never to produce.Read the full skillView on GitHub

Like (0)

Add a commentPost

fail-closed-guardsWhen a guard/gate/precondition depends on data that might be missing, make the missing-data fallback REFUSE the action, not allow it — and confirm every write-site actually populates the field the guard reads.Use it for: Writing or reviewing any permission check, gate, or precondition with a fallback default.Read the full skillView on GitHub

Like (0)

Add a commentPost

twilio-a2p-10dlc-complianceGet a Twilio 10DLC A2P SMS campaign approved, or un-stick a rejected one — decodes the common rejection codes and covers the non-obvious resubmission mechanics.Use it for: A FAILED/rejected Twilio A2P campaign, or building an SMS opt-in flow before submitting one.Read the full skillView on GitHub

Like (0)

Add a commentPost

skill-estate-guardianContinuously lint your own AI-agent tooling (skills, subagent definitions, workflow scripts, hooks) for the defect classes that quietly rot an agent estate — leaked tokens, retired model IDs, unpinned models, gate docs with no real enforcement.Use it for: A pre-flight check after writing/editing a skill or agent, or a periodic cleanup pass over your whole skills directory.Read the full skillView on GitHub

Like (0)

Add a commentPost

roast-meMulti-round adversarial questioning that forces you to defend a design BEFORE you build it — generates hard questions across failure modes, edge cases, security, and performance, then iterates on your answers.Use it for: Stress-testing a plan or architecture decision before you commit to building it.Read the full skillView on GitHub

Like (0)

Add a commentPost

ai-tells-checkA pre-flight linter for the highest-signal 'AI-slop' tells in generated frontend/marketing copy — em-dashes, numbered eyebrows, BETA labels, generic filler phrases, decorative status dots.Use it for: A fast last check after generating any landing page, hero section, or marketing UI.Read the full skillView on GitHub

Like (0)

Add a commentPost

data-illustrationTurn a set of scores/metrics into an accurate AND illustrative graphic — the data literally drives the shape of the artwork, so it reads as a designed illustration instead of a generic chart.Use it for: Visualizing scores/metrics/dimensions when a plain bar/line/pie chart would be boring or off-brand.Read the full skillView on GitHub

Like (0)

Add a commentPost

agent-engineering-checklistA curated checklist of ~18 evidence-grounded rules for building reliable AI agents — tool-use gating, verification-before-trust, context engineering, and safety/trust boundaries.Use it for: Designing a new agent/workflow, or reviewing an existing one for reliability gaps before more autonomous use.Read the full skillView on GitHub

Like (0)

Add a commentPost

agent-quality-loopIteratively grade and improve a conversational AI agent's prompt quality without real end-users — synthetic dialogs, hard checks + an LLM judge, targeted human-approved prompt edits until quality plateaus.Use it for: After any meaningful change to a conversational agent's system prompt, or as a regression gate before a demo/release.Read the full skillView on GitHub

Like (0)

Add a commentPost

ai-ugc-video-pipelineA hardened, gated pipeline for producing short-form AI-generated 'UGC-style' marketing video — hard-won fixes for the media-validation, audio-sync, and judge-hallucination traps that silently ship broken output.Use it for: Producing or re-cutting a short-form AI-presenter product/marketing video.Read the full skillView on GitHub

Like (0)

Add a commentPost

Published Monday, July 6, 2026 · 2026-W28

expert-fanoutSpawn N domain-expert subagents in parallel to audit a codebase, product, or design from different angles, then synthesize convergent vs. divergent recommendations.Use it for: Breaking out of a surgical-patch loop, getting a second opinion before a risky change, comparing two architecture options.Read the full skillView on GitHub

Like (0)

Add a commentPost

doc-iterateProcess inline review-comment markers left in shared Markdown docs, apply edits in place, preserve history, and flag contradictions instead of guessing.Use it for: Async doc review between collaborators without a full comment/suggestion system.Read the full skillView on GitHub

Like (0)

Add a commentPost

bug-replayTurn every fixed bug into a permanent regression fixture and replay the whole set before you ship anything.Use it for: Catching regressions on bugs you already fixed, building a growing regression suite from real incidents.Read the full skillView on GitHub

Like (0)

Add a commentPost

ship-as-artifactRender any finished deliverable (report, brief, plan, design, mockup) as a self-contained hosted page and hand over the link instead of a terminal dump.Use it for: Reports, briefs, audits, roadmaps, and design/mockup previews meant to be read, reviewed, or shared.Read the full skillView on GitHub

Like (0)

Add a commentPost

offloadRoute self-contained work off your primary AI assistant's metered quota onto a secondary backend you already have, then verify the result yourself.Use it for: Being low on quota, parallelizing independent work, or getting an independent second opinion.Read the full skillView on GitHub

Like (0)

Add a commentPost

quality-gate-registryMap every 'done' claim to the specific proof required before you say it out loud.Use it for: Preventing confidently reporting unverified tests, builds, deploys, or fixes as complete.Read the full skillView on GitHub

Like (0)

Add a commentPost

staff-code-reviewA maximum-depth, multi-agent code review that auto-locates the diff, fans out a cross-functional panel, adversarially verifies findings, and synthesizes a prioritized risk register.Use it for: Pre-merge high-stakes diffs and production-readiness audits.Read the full skillView on GitHub

Like (0)

Add a commentPost

production-gauntletA looped, dual-verification stress-test and KPI gate to run before calling any non-trivial change 'done'.Use it for: Wanting genuine walk-away quality on anything non-trivial you've built, fixed, or deployed.Read the full skillView on GitHub

Like (0)

Add a commentPost

autobuild-loopAn end-to-end, self-verifying build pipeline (plan, swarm, docs, QA, code quality, canary, summary) gated stage-by-stage on verification artifacts.Use it for: A defined-end-state build you want to walk away from and trust came back correct.Read the full skillView on GitHub

Like (0)

Add a commentPost

skill-minerMine your own AI-assistant session transcripts for repeated workflow patterns and draft candidate reusable skills, queued for human approval.Use it for: Discovering which of your repeated multi-step workflows would make good reusable skills.Read the full skillView on GitHub

Like (0)

Add a commentPost

Archive: older Builder's Log posts