mikko-help

The discoverability sidekick for the mikko-* namespace — for when you remember you have a skill for something but not its exact name. /mikko<Tab> already lists the names (free, built into the CLI); this adds the descriptions so you can pick by what each one does.

All the work — glob the skill dirs, parse each SKILL.md frontmatter, dedupe, sort, format — is deterministic, so it lives in a script rather than in LLM prose. One Bash call instead of ~8–11K tokens of glob+read+format per run.

How to run

Run the bundled script and print its output verbatim:

python3 ~/.claude/skills/mikko-help/mikko-help.py [--barney] [--detect]

(From the source repo: python3 skills/mikko-help/mikko-help.py.) It globs ~/.claude/skills/mikko-*/SKILL.md (user-wide) and .claude/skills/mikko-*/SKILL.md (project-local, wins on a name clash and gets a (project) tag), then prints a two-column name + description table. Nothing is written to disk.

Flags

• (none) — list every skill with its description. Default; cheapest.
• --barney — show each skill's plain-English barney: line instead (falls back to the truncated description, tagged (no barney), for skills without one).
• --detect — also fingerprint the current repo (reads a handful of root config files — package.json, tsconfig.json, pyproject.toml, Cargo.toml, go.mod, *.sln/*.csproj/global.json, framework configs) and print an ordered "which audit should I run here" recommendation. Does not walk the source tree and does not dispatch audits — it's a hallway sign, not a tour guide.

--detect and --barney are independent and combine.

Detection matrix (--detect)

The recommendation logic lives in skills_listing.recommend_audits. In short: React → react-anti-patterns-audit first, then the universal ai-codegen-smell-audit + audit; .NET / ASP.NET Core → dotnet-audit first, then ai-codegen-smell-audit + audit; other non-React → audit + ai-codegen-smell-audit; security-sensitive deps (auth/db/network/crypto libraries, or ASP.NET Core Identity / EF Core) lift security-audit from skip to suggest. React Native adds --force to bypass the web-shape pre-flight. A repo matching multiple stacks (e.g. React + .NET in one root) gets each stack's audit.

Failure modes

• No skills directory. Prints "no mikko-* skills installed yet" and a hint to run ./install-mikko.sh. Exits cleanly.
• Malformed frontmatter. Falls back to the directory name + (no description) rather than failing the run.
• --detect with no config files. Reports "language: unknown" and recommends audit only — honest about what it can't infer.

Why this skill exists

/mikko<Tab> gives names; the gap mikko-help fills is needing descriptions to choose between similarly-named skills, or to recall what a freshly-installed skill does. For a portfolio audience it's a small, real signal that the mikko-* prefix is a deliberately-designed namespace with its own "what's in here" command — and that the listing is a cheap script, not an LLM loop.

Freshness check

Staleness checks run by /mikko-skills-freshness on any change to this skill.

[[check]]
kind = "file_contains"
path = "SKILL.md"
pattern = "^## Flags"

[[check]]
kind = "file_contains"
path = "SKILL.md"
pattern = "`--detect`"

[[check]]
kind = "path_exists"
path = "mikko-help.py"