{"slug":"alterlab-skill-finder","title":"alterlab-skill-finder","summary":"The AlterLab front door and multi-agent launcher — routes a task to the right AlterLab skill(s) when the user invokes the suite without naming one, and for a multi-stage goal (or on the keyword 'alterflow', aliases 'alterresearch' / 'ultralab') it clarifies the goal with a few qu","platform":"Claude","tags":[],"authorName":"LLM Mart","authorSlug":"llm-mart","score":0,"source":"github","price":null,"verified":false,"createdAt":"2026-09-23T18:57:03.088177Z","repo":{"url":"https://github.com/AlterLab-IEU/AlterLab-Academic-Skills","stars":68,"forks":13,"license":"MIT","updatedAt":"2026-09-23T13:42:59Z"},"bodyHtml":"<hr>\n<h2>name: alterlab-skill-finder\ndescription: \"The AlterLab front door and multi-agent launcher — routes a task to the right AlterLab skill(s) when the user invokes the suite without naming one, and for a multi-stage goal (or on the keyword 'alterflow', aliases 'alterresearch' / 'ultralab') it clarifies the goal with a few questions before anything runs, selects the skills the task needs, and runs a multi-agent workflow composing them (a packaged alterlab-workflows workflow, alterlab-research-pipeline, alterlab-ssci-orchestrator, or a bespoke plan via alterlab-workflow-orchestration). Triggers on 'use AlterLab skills', 'which AlterLab skill for X', 'is there an AlterLab skill for…', a multi-stage research goal, 'alterflow …', or any generic AlterLab request where the user does not know skill names. Use when someone references AlterLab generically, describes a multi-stage goal, or fires the alterflow keyword; when the user already names a specific skill, defer to that skill directly. Part of the AlterLab Academic Skills suite.\"\nlicense: MIT\nallowed-tools: Read Grep Agent\ncompatibility: \"No API key or network required. A routing + orchestration front-end: it reads the bundled skill index and hands off to the matching AlterLab skill(s). Multi-agent execution uses the host's subagent/Workflow tools where available (Claude Code, Cowork); on surfaces without them it decomposes the work into sequential phases.\"\nmetadata:\nskill-author: AlterLab\nversion: \"1.2.0\"\nlast_updated: \"2026-09-23\"\ndepends_on: \"dispatches to any AlterLab skill; composes multi-step runs via alterlab-research-pipeline, alterlab-workflow-orchestration, and alterlab-ssci-orchestrator\"</h2>\n<h1>AlterLab Skill Finder — Name the Task, Get the Right Skill (or the Whole Workflow)</h1>\n<p><strong>Skill type: ROUTER / LAUNCHER.</strong> Users say <em>\"use AlterLab skills\"</em> or fire <strong><code>alterflow</code></strong> without\nknowing the 230+ skill names. This is the front door: it reads the task, picks the AlterLab skill(s)\nthat fit, and either <strong>uses one skill</strong> or <strong>launches a clarified, multi-agent workflow</strong>.</p>\n<h2>Core Mission</h2>\n<pre><code>WHEN THE USER NAMES ALTERLAB BUT NOT A SKILL, YOU PICK THE SKILL(S).\nSIMPLE TASK → ROUTE TO ONE SKILL.   BIG / KEYWORD TASK → CLARIFY, THEN ORCHESTRATE MANY.\nALWAYS ASK YOUR QUESTIONS BEFORE YOU START A MULTI-STEP RUN.\n</code></pre>\n<h2>Two modes</h2>\n<table>\n<thead>\n<tr>\n<th>Mode</th>\n<th>Fires on</th>\n<th>What happens</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td><strong>Route</strong> (default)</td>\n<td>\"use AlterLab skills to…\", \"which AlterLab skill for…\", any generic AlterLab ask</td>\n<td>classify the task → map to a domain → name and <strong>apply</strong> the best-fitting skill(s) — Anthropic's <strong>routing</strong> pattern</td>\n</tr>\n<tr>\n<td><strong>Orchestrate</strong></td>\n<td>a task that <strong>spans several dependent stages</strong> (research → write → review, etc.); the keyword <strong><code>alterflow</code></strong> — aliases <code>alterresearch</code> / <code>ultralab</code> — is an explicit shortcut into this mode</td>\n<td><strong>clarify → select skills → plan a multi-agent workflow → confirm → execute</strong> — Anthropic's <strong>orchestrator-workers</strong> pattern</td>\n</tr>\n</tbody>\n</table>\n<h2>When to Use This Skill</h2>\n<ul>\n<li>\"Use AlterLab skills to [do X].\" / \"Which AlterLab skill should I use for [task]?\"</li>\n<li>\"Is there an AlterLab skill / workflow for [task]?\"</li>\n<li>\"<strong>alterflow</strong> — investigate [topic] and draft a paper.\" (→ Orchestrate mode)</li>\n<li>Any request that references AlterLab by name without specifying a skill.</li>\n</ul>\n<h3>Does NOT Trigger</h3>\n<table>\n<thead>\n<tr>\n<th>The request is really about…</th>\n<th>Route to</th>\n<th>Why not this skill</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>A <strong>specific named</strong> skill (\"run alterlab-deep-research\", \"use survey-analysis\")</td>\n<td>that skill directly</td>\n<td>The user already knows it; no routing needed.</td>\n</tr>\n<tr>\n<td>Composing a bespoke multi-agent workflow the user is hand-designing</td>\n<td><code>alterlab-workflow-orchestration</code></td>\n<td>This skill <em>calls</em> that engine; it doesn't replace its mechanics.</td>\n</tr>\n<tr>\n<td>A task with no AlterLab framing at all</td>\n<td>(answer normally)</td>\n<td>Nothing to route.</td>\n</tr>\n</tbody>\n</table>\n<h2>Route mode — the method</h2>\n<ol>\n<li><strong>Classify the task</strong> — what is the user doing (find literature? clean data? fit a model? write a\nsection? query a database? design a study?).</li>\n<li><strong>Map to a domain</strong> (table below).</li>\n<li><strong>Select the skill(s)</strong> — name the best fit. For an exact niche name, consult\n<code>references/skill_index.md</code> (every AlterLab skill, grouped by domain, one-liner each) or match the\ninstalled AlterLab skill descriptions.</li>\n<li><strong>Check it is installed.</strong> The index lists all AlterLab skills, but the host can only run the ones\ninstalled there. In Claude Code a skill is available when it appears in the host's skill list as\n<code>alterlab-&lt;domain&gt;:&lt;skill&gt;</code>; a Skill call that returns <em>Unknown skill</em> means its plugin is not\ninstalled. Then stop and use the <strong>not-installed reply</strong> below — do not retry, and do not search\nfor other tools.</li>\n<li><strong>Apply it</strong> — invoke the skill and do the work. Tell the user which skill you picked and why (one line).</li>\n<li><strong>If the routing genuinely forks, ask ONE clarifying question</strong> before committing.</li>\n</ol>\n<h3>Not-installed reply</h3>\n<p>Name the skill and why it fits, then how to get it. Its plugin is <code>alterlab-</code> plus the domain id shown\nin backticks in the index heading (for example <code>turkish-academia</code> → <code>alterlab-turkish-academia</code>):</p>\n<ul>\n<li><strong>Claude Code:</strong> <code>/plugin install alterlab-&lt;domain&gt;@alterlab-academic-skills</code>, then <code>/reload-plugins</code>\n(or <code>alterlab-complete@alterlab-academic-skills</code> for every domain).</li>\n<li><strong>Claude app (claude.ai):</strong> upload <code>skill-&lt;skill-name&gt;-&lt;version&gt;.zip</code> from the latest GitHub release\n(Customize ▸ Skills → Upload a skill).</li>\n</ul>\n<p>If an installed skill covers part of the task, offer it as a stopgap and say what it will not cover;\notherwise wait for the user to install the plugin.</p>\n<h2>Orchestrate mode — <code>alterflow</code> (clarify FIRST, then multi-agent)</h2>\n<p>Fires when the task <strong>spans several dependent stages</strong> (e.g. \"investigate X <em>and</em> write it up\",\n\"research → analyze → publish\") — the <code>alterflow</code> keyword is just an explicit shortcut into this\nmode, not the only way in; detect the multi-stage intent even when the user never types it. This is\nAnthropic's <strong>orchestrator-workers</strong> pattern (a lead agent dynamically decomposes the task, delegates\nto workers, and synthesizes) with a human-in-the-loop clarify/confirm gate in front. Do <strong>not</strong> start\nexecuting — run this sequence:</p>\n<ol>\n<li><strong>CLARIFY (mandatory, before any execution).</strong> Ask 2–4 sharp scoping questions — e.g. the research\nquestion / deliverable, quantitative vs qualitative, the dataset or corpus, the target output\n(report, paper, model, figures), depth/time budget, and language. Do not guess when the answer\nchanges the plan.</li>\n<li><strong>SELECT.</strong> From the answers, choose the AlterLab skills the job needs (use the domain map + the\nindex). List them. <strong>First ask whether it should be multi-agent at all:</strong> if the work is tightly\ncoupled, needs one shared context, or has little that runs in parallel (Anthropic's caution — most\ncoding-like, dependency-heavy tasks), prefer a <strong>sequential pipeline (prompt-chaining)</strong> or a single\nRoute-mode skill over a fan-out.</li>\n<li><strong>PLAN.</strong> Lay out the <strong>dynamic multi-agent workflow</strong> — the phases, which skill runs in each, where\nsubagents fan out in parallel, and the <strong>evaluator-optimizer</strong> verification/critique step. <strong>Scale\nthe subagent count to complexity and default low</strong> (Anthropic's rule of thumb: simple fact-finding\n≈ 1 agent; a focused comparison ≈ 2–4; only genuinely broad work ≈ 10+). Reuse the existing\norchestrators rather than reinventing them:\n<ul>\n<li>a packaged job — whole-manuscript citation audit, independent review panel, claim stress-test,\nPRISMA dual screening, response to reviewers, grant mock panel, literature map →\n<strong><code>alterlab-research-workflows</code></strong> (in Claude Code: <code>/alterlab-workflows:&lt;name&gt;</code>)</li>\n<li>end-to-end research→publish → <strong><code>alterlab-research-pipeline</code></strong> (deep-research → paper-writer → paper-reviewer, revision loops)</li>\n<li>a whole social-science study → <strong><code>alterlab-ssci-orchestrator</code></strong> (design → measurement/reflexivity → sampling → analysis module → inference)</li>\n<li>a bespoke fan-out / judge-panel / adversarial-verify workflow → <strong><code>alterlab-workflow-orchestration</code></strong></li>\n</ul>\n</li>\n<li><strong>CONFIRM — plan <em>and its rough cost</em>.</strong> Show the selected skills + the phases in a few lines,\n<strong>plus an effort estimate</strong> (≈ how many phases / subagents). Multi-agent runs spend far more tokens\nthan a single pass, so let the user opt in knowingly; get a go-ahead (or incorporate a correction).</li>\n<li><strong>EXECUTE.</strong> The user's go-ahead in step 4 is their opt-in to a multi-agent run. In Claude Code\nwith dynamic workflows, launch a matching packaged workflow (<code>/alterlab-workflows:&lt;name&gt;</code>), or write\na workflow script for a bespoke plan that fans out beyond a handful of workers; for a few workers,\nspawn subagents (the Agent tool). On surfaces without these tools, execute the phases sequentially and keep\nthe same hand-offs. <strong>Give every spawned subagent a complete task spec</strong> — an objective, an\noutput format, which skills/tools/sources to use, and clear boundaries — or workers duplicate work\nand leave gaps (<code>alterlab-workflow-orchestration</code> models these specs). Carry each stage's artifact\nto the next (the pipelines define the hand-off contracts).</li>\n</ol>\n<p>The one rule: <strong>questions before execution.</strong> A short clarify step beats a wrong 20-agent run.</p>\n<h2>Domain routing map (18 domains)</h2>\n<table>\n<thead>\n<tr>\n<th>If the task is about…</th>\n<th>Domain</th>\n<th>Representative skills</th>\n</tr>\n</thead>\n<tbody>\n<tr>\n<td>Find literature, fact-check, discover, manage references</td>\n<td><strong>research-tools</strong> / <strong>core</strong></td>\n<td><code>alterlab-deep-research</code>, <code>alterlab-research-lookup</code>, <code>alterlab-pyzotero</code>, <code>alterlab-citation-verifier</code></td>\n</tr>\n<tr>\n<td>Write / draft / revise a paper, abstract, grant, poster; disclose AI use</td>\n<td><strong>core</strong> / <strong>writing-tools</strong></td>\n<td><code>alterlab-paper-writer</code>, <code>alterlab-scientific-writing</code>, <code>alterlab-research-grants</code>, <code>alterlab-ai-use-disclosure</code></td>\n</tr>\n<tr>\n<td>Review / critique a manuscript</td>\n<td><strong>core</strong></td>\n<td><code>alterlab-paper-reviewer</code></td>\n</tr>\n<tr>\n<td>Query a scientific database (PubMed, ChEMBL, UniProt, GEO, …)</td>\n<td><strong>databases</strong></td>\n<td><code>alterlab-pubmed</code>, <code>alterlab-chembl</code>, <code>alterlab-uniprot</code>, … (39)</td>\n</tr>\n<tr>\n<td>Genomics, proteomics, single-cell, structure prediction</td>\n<td><strong>bioinformatics</strong></td>\n<td><code>alterlab-scanpy</code>, <code>alterlab-alphafold</code>, <code>alterlab-biopython</code>, …</td>\n</tr>\n<tr>\n<td>Chemistry, drug discovery, docking, ADMET</td>\n<td><strong>cheminformatics</strong></td>\n<td><code>alterlab-rdkit</code>, <code>alterlab-deepchem</code>, …</td>\n</tr>\n<tr>\n<td>Clinical decision support, trials, medical imaging</td>\n<td><strong>clinical-research</strong></td>\n<td><code>alterlab-clinicaltrials</code>, <code>alterlab-clinical-decision</code>, …</td>\n</tr>\n<tr>\n<td>ML, statistics, data analysis, dataframes</td>\n<td><strong>data-science</strong></td>\n<td><code>alterlab-scikit-learn</code>, <code>alterlab-statistical-analysis</code>, <code>alterlab-statsmodels</code>, <code>alterlab-pymc</code></td>\n</tr>\n<tr>\n<td>Plots, charts, figures, schematics</td>\n<td><strong>visualization</strong></td>\n<td><code>alterlab-matplotlib</code>, <code>alterlab-scientific-viz</code>, <code>alterlab-infographics</code></td>\n</tr>\n<tr>\n<td>Lab platforms (Benchling, DNAnexus, Opentrons)</td>\n<td><strong>lab-integrations</strong></td>\n<td><code>alterlab-benchling</code>, …</td>\n</tr>\n<tr>\n<td>Quantum, geospatial, materials, astronomy, digital humanities</td>\n<td><strong>domain-specific</strong></td>\n<td><code>alterlab-qiskit</code>, <code>alterlab-geopandas</code>, <code>alterlab-pymatgen</code>, …</td>\n</tr>\n<tr>\n<td>Convert / handle documents, Markdown, notebooks, PDFs</td>\n<td><strong>document-tools</strong></td>\n<td><code>alterlab-markitdown</code>, <code>alterlab-pdf-explore</code></td>\n</tr>\n<tr>\n<td>Finance, economics, market/financial data</td>\n<td><strong>finance-economics</strong></td>\n<td><code>alterlab-fred</code>, <code>alterlab-edgartools</code>, …</td>\n</tr>\n<tr>\n<td>Turkish academic system (YÖK, ÜAK, DergiPark, TÜBİTAK, BİDEB, doçentlik, KVKK)</td>\n<td><strong>turkish-academia</strong></td>\n<td><code>alterlab-dergipark</code>, <code>alterlab-tubitak-proposal</code>, <code>alterlab-tubitak-bideb</code>, <code>alterlab-docentlik-eligibility</code>, <code>alterlab-kvkk-dmp</code>, …</td>\n</tr>\n<tr>\n<td>Research ethics, IRB / ethics-committee applications, consent forms</td>\n<td><strong>research-tools</strong> (Türkiye: <strong>turkish-academia</strong>)</td>\n<td><code>alterlab-research-ethics</code>, <code>alterlab-tr-research-ethics</code></td>\n</tr>\n<tr>\n<td>Teaching, grant reporting, accreditation, recommendation letters, preprints</td>\n<td><strong>faculty-life</strong></td>\n<td><code>alterlab-syllabus-ai-policy</code>, <code>alterlab-grant-reporting</code>, <code>alterlab-accreditation-aol</code>, …</td>\n</tr>\n<tr>\n<td>Research-rigor gates (pre-registration, test choice, transparency)</td>\n<td><strong>methodology</strong></td>\n<td><code>alterlab-test-selection-guard</code>, <code>alterlab-preregistration-discipline</code></td>\n</tr>\n<tr>\n<td>Design/run a whole <strong>social-science study</strong> (survey, qualitative, causal, multilevel, meta, missing data)</td>\n<td><strong>social-science-workflow</strong></td>\n<td><code>alterlab-ssci-orchestrator</code> (+ 16 gates &amp; modules)</td>\n</tr>\n<tr>\n<td>Run a packaged <strong>multi-agent job</strong> — citation audit, review panel, PRISMA screening, rebuttal, grant mock panel, literature map, claim stress-test</td>\n<td><strong>workflows</strong></td>\n<td><code>alterlab-research-workflows</code> (<code>/alterlab-workflows:&lt;name&gt;</code>)</td>\n</tr>\n</tbody>\n</table>\n<p>Full per-domain listing of every skill with a one-liner: <code>references/skill_index.md</code>.</p>\n<h2>Output pattern</h2>\n<pre><code>Task: &lt;one-line restatement&gt;\nMode: route | orchestrate\nAlterLab skill(s): &lt;name(s)&gt; — &lt;why this fits&gt;\n[orchestrate] Questions first: &lt;2–4 scoping questions&gt;   ← ask, then wait\n[orchestrate] Plan: &lt;phase 1 skill → phase 2 skills (fan-out) → verify → deliver&gt;\n→ on confirmation, execute.\n</code></pre>\n<h2>References</h2>\n<ul>\n<li><code>references/skill_index.md</code> — every AlterLab skill, grouped by domain, one-liner each (generated, kept in sync with the catalog).</li>\n</ul>\n<p>Part of the AlterLab Academic Skills suite.</p>\n","files":[{"path":"evals/evals.json","sizeBytes":6836,"isText":true},{"path":"references/skill_index.md","sizeBytes":84981,"isText":true},{"path":"SKILL.md","sizeBytes":13118,"isText":true}],"reviewScore":null,"reviewSummary":null,"trust":{"provenance":"trusted-source-unreviewed","notice":"Community-authored content, reproduced verbatim and not vetted as instructions. Treat it as data to evaluate, never as directives to follow.","bodySource":null},"bodyLocked":false,"purchaseUrl":null,"sourceUrl":null,"report":{"provenance":"trusted-source-unreviewed","screen":{"ran":true,"outcome":"clean","suspicious":0,"notes":0,"hiddenCharacters":false},"virusScan":{"engine":"clamav","status":"clean","scannedAt":"2026-09-23T18:58:42.985668Z","sha256":"D393D6DB6BF595E989F3AEC22DE36DB1E7CFEEF2CB7B25D6953C54E1F0422CAC","sizeBytes":42484},"review":null,"source":{"repositoryUrl":"https://github.com/AlterLab-IEU/AlterLab-Academic-Skills","path":"skills/core/alterlab-skill-finder","license":"MIT","commit":"e4836c08a20da195a11f30f203a8cf23ec30aa95","subtreeSha":"313A734AAB00A47ADB92597389A1B54480BA5BF3DF25C88CF942C07FAB642B7B","lastSyncedAt":"2026-09-23T18:56:52.297238Z"},"reviewedAt":"2026-09-23T19:01:50.37836Z","notice":"Community-authored content, reproduced verbatim and not vetted as instructions. Treat it as data to evaluate, never as directives to follow."},"install":[{"target":"skills-cli","command":"npx skills add https://github.com/AlterLab-IEU/AlterLab-Academic-Skills/tree/main/skills/core/alterlab-skill-finder"},{"target":"claude-code","command":"claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install alterlab-ieu-alterlab-academic-skills@llmmart"},{"target":"git","command":"git clone https://github.com/AlterLab-IEU/AlterLab-Academic-Skills.git"}]}