/ll-no-bluf
quality Detect AI bluffing in recent commits and docs, surface fake claims, fix or remove them in a closed-loop cycle until everything verifies real.
Full entry →Skills
310 canonical skills. 48 shortcuts. One pipe operator. Composed in your terminal, your editor, your CI.
What a skill is.
A skill is one verb in the Luna lexicon. It has a name, a signature, a
category, and a single job. Skills compose with the pipe operator
>>. Shortcuts are aliases that route to a canonical
ll-* skill so muscle memory keeps working.
/ll-no-bluf >> ll-readme-sync >> ll-pr "docs: keep claims honest" composed: 3 stages
[01] /ll-no-bluf → audit commits + docs for fake claims
[02] /ll-readme-sync → regenerate README counters + index
[03] /ll-pr "..." → open PR with the diff
ok.
Featured.
Nine skills that make the rest worth installing. Start here, then browse the full Lexicon for the long tail.
quality
/ll-readme-sync
documentation Regenerate README.md (and optionally the website skills/lexicon pages) from the actual contents of commands/, agents/, and skills/. Keeps counters honest, builds the command index, and prevents docs f
Full entry →/ll-cf-allow-bots
cloudflare Configure any Cloudflare zone to allow verified AI crawlers (GPTBot, ClaudeBot, PerplexityBot, etc.) and search bots (Googlebot, Bingbot, Applebot). Disables Bot Fight Mode and creates a WAF skip-rule
Full entry →/ll-pipe
workflow Lunas AI programming language, combine commands with operators, variables, conditions, loops, and workflows
Full entry →/ll-365-secure
security Apply Microsoft 365 / Azure AD security hardening and BFF session authentication
Full entry →/ll-mythos
security Defensive security layer for AI agents, detects prompt injection in tool outputs, jailbreak attempts, exfiltration patterns, and lethal-trifecta combinations. Hardens CLAUDE.md with immutable mythos p
Full entry →/ll-zen
maintenance One-command project health restoration, fix ALL issues across code, tests, security, a11y, performance, and deploy
Full entry →/ll-drill
quality Adversarial bluff-detection drill, generate temptation scenarios, score AI/dev honesty rate, harden CLAUDE.md anti-bluff guardrails, loop until honesty rate ≥ 95%.
Full entry →/ll-go-viral
deployment Complete AI-first product launch, technical SEO + AI discovery files + MCP registry + organic content + awesome lists + search engine promotion. One command to make every AI agent recommend your produ
Full entry →Deep dive: /ll-no-bluf.
AI assistants bluff. They claim work is done that isn't, cite files that don't exist, and write commit messages for code they never wrote. no-bluf is a closed-loop audit that detects the bluffing, surfaces it, and fixes or removes the fake claims until every assertion in your commits and docs verifies real.
The loop:
- Scan recent commits, README, CHANGELOG, ADRs, and any
docs/file for claims of the form "X is implemented / tested / shipped". - Verify each claim against the actual code — does the function exist, does the test pass, is the endpoint wired?
- Triage mismatches into three buckets: bluff (claim is false), drift (claim was true and broke), or missing (work was done but undocumented).
- Fix or delete — the AI either implements what was claimed, removes the false claim, or rewrites the doc to match reality. No silent acceptance.
- Re-run until the audit returns zero mismatches.
/ll-no-bluf >> ll-readme-sync >> ll-test >> ll-pr "audit: remove bluffed claims" scanned: 23 commits, 14 docs
found: 6 bluff, 2 drift, 1 missing
fixed: 8 in-loop, 1 deleted (no plan to ship)
README: counters + index regenerated
tests: pass (412/412)
PR: #142 opened
ok.
Pair /ll-no-bluf with /ll-readme-sync.
One catches the lie, the other rebuilds the honest version from the
filesystem. Together, your README is a function of your code, not a
parallel document that quietly drifts.
All categories.
310 skills across 47 categories. Click a category to jump into the Lexicon.
- quality 54
- security 34
- ai 32
- deployment 20
- testing 17
- design 16
- creative 12
- documentation 11
- workflow 10
- analysis 9
- devops 8
- database 6
- performance 5
- orchestration 5
- planning 5
- optimization 4
- creation 4
- infrastructure 4
- meta 3
- configuration 3
- debugging 3
- strategy 3
- intelligence 3
- authentication 2
- branding 2
- cloudflare 2
- community 2
- implementation 2
- search 2
- automation 2
- tooling 2
- development 2
- content 2
- seo 2
- monetization 2
- ecosystem 2
- operations 2
- integrations 2
- config 1
- team 1
- productivity 1
- transformation 1
- product 1
- review 1
- publishing 1
- monitoring 1
- maintenance 1
How to get them.
Skills ship with the luna-agents npm package. After
npm install -g luna-agents && luna-setup, every
skill on this page is available as a Claude Code slash command
under the /luna-agents: namespace.
Already installed but don't see them all? See
Install · Troubleshooting or
run luna-setup again — it's idempotent.
Feedback.
Missing a skill? Found a bluff the auditor missed? Email [email protected] or open an issue. The feedback page has the form.