ask-bigin
Skill1 starsChanged 40 days agoMITView on GitHub
Explicit only — type /ask-bigin <what you want>. Names the one bigin-skills skill that fits and hands off, reading the shared triage ladder for build work and the named subject otherwise. Use when you don't know which skill applies, or want the inventory.
In this repository23 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- harness-auditSkill
- skill-benchSkill
- bigin-harness-setupSkill
- contract-syncSkill
- debug-workflowSkill
- discovery-workflowSkill
- epic-workflowSkill
- flutter-figma-handoffSkill
- go-scaffoldSkill
- knowledge-distillSkill
- model-routerSkill
- napkinSkill
- next-scaffoldSkill
- nodejs-scaffoldSkill
- nuxt-marketing-scaffoldSkill
- nuxt-scaffoldSkill
- nuxt-ui-figma-handoffSkill
- project-scaffoldSkill
- session-handoffSkill
- sprint-distillSkill
- task-workflowSkill
- write-testsSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 34
Read the request for what it names: Not what it implies — what it names.
- Distinctive: line 43
Do not re-derive the bar from memory, and do not paraphrase its triggers here — rung 2 is a three-way disjunction and a paraphrase that …
- Distinctive: line 63
If nothing matches, say so plainly and answer the question directly instead of forcing a skill.
- Distinctive: line 68
Don't dump them all with equal weight: lead with the ladder, since that's what most sessions need.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill 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
ask-bigin
You name the skill that fits and hand off. You never do the work yourself — not the scoping,
not the spec, not the diff. One line of reasoning, then invoke. A router that starts implementing
is just a slower task-workflow.
The two things you must not restate
This skill exists because there are many doors, not because routing needs a new definition. Both inputs already have a home:
- The triage ladder —
${CLAUDE_PLUGIN_ROOT}/skills/discovery-workflow/references/triage-ladder.md. Three rungs, already read bytask-workflow,epic-workflowanddiscovery-workflow. Read it; never paraphrase it here. You are its fourth reader, not a fourth bar. If your answer for build work disagrees with the ladder, the ladder is right and you are wrong. - The inventory — every skill's
description:frontmatter is already in your context on every turn. That is the list. Derive any listing from what you can see loaded, never from a table written here: a hardcoded inventory drifts the moment a skill is added, and the generated tables in the plugin'sREADME.mdare the human-facing copy that stays current mechanically. One gap: adisable-model-invocation: trueskill is never in that listing (this one included). The only other one today is/napkin— a topic explained as a picture — so route a request to see something as a diagram there by name.
Procedure
- Read the request for what it names. Not what it implies — what it names. "Write tests for the parser" names tests. "Why is this flaky" names a bug. "We want a client portal" names a product nobody has scoped.
Is it build work? — a change to this codebase, at any size. If yes, read the triage ladder and route by rung. Say the rung and the one discriminator that decided it:
Rung 2 — a contract and its consumers is two surfaces, so
epic-workflowfirst.Do not re-derive the bar from memory, and do not paraphrase its triggers here — rung 2 is a three-way disjunction and a paraphrase that drops one of them silently lowers the bar. Read the file and quote the trigger that actually fired.
Otherwise route on the named subject. These are the non-ladder doors — each is a skill whose own
description:you can already see, so check that description rather than trusting this list to be complete:The request names Route to tests for a named unit, or a PRD acceptance criterion write-testsa bug that needs diagnosis before it can be scoped debug-workflow(then rung 1)a third-party library at a version knowledge-distillthe end of a sprint, or what the team learned sprint-distilla new repo or app from nothing the matching *-scaffoldAI rules, gates, or Cursor support for an existing repo bigin-harness-setupa Figma handoff in a Nuxt UI app nuxt-ui-figma-handoffa context limit, or saving state to resume session-handoffwhich model or tier should run something model-routerIf nothing matches, say so plainly and answer the question directly instead of forcing a skill. "No skill covers this, here's the answer" is a correct outcome.
- No arguments given? List what's available — grouped as build work, knowledge, scaffolding, and setup — one line each, from the loaded descriptions. Then ask what they're trying to do. Don't dump them all with equal weight: lead with the ladder, since that's what most sessions need.
- Two candidates genuinely tied? One
AskUserQuestionwith those two as the options and the real trade-off in each description, then route. Genuinely tied means you can state a concrete reason for each. One question, not a survey — if you can pick, pick. - Hand off. Invoke the chosen skill via the Skill tool. If the user asked which skill rather than asking you to run it, name it and stop — answering "which" with an unrequested invocation is its own failure.
What this is not
- Not a gate. It approves nothing and blocks nothing. The spec gate still runs inside
task-workflowexactly as it would have. - Not a step in any workflow. Nothing invokes this skill; it's a door for a human who doesn't know which door. No skill should ever route through here.
- Not a wrapper. Routing to
task-workflowand routing toepic-workflowcost the same from here: one line and a hand-off. If you find yourself summarizing what the next skill will do, stop — it will do it, and its own first step says so.
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.

