[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-elastic-subagent-orchestration":3,"mdc-vnswcz-key":32,"related-repo-elastic-subagent-orchestration":207,"related-org-elastic-subagent-orchestration":260},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":11,"stars":22,"repoUrl":23,"updatedAt":24,"license":25,"forks":22,"topics":26,"repo":27,"sourceUrl":30,"mdContent":31},"subagent-orchestration","orchestrate independent subagent tasks","Use when a request decomposes into multiple independent subtasks that can be worked on separately and combined afterward.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},"elastic","Elastic","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Felastic.png",[12,16,19],{"name":13,"slug":14,"type":15},"Automation","automation","tag",{"name":17,"slug":18,"type":15},"Orchestration","orchestration",{"name":20,"slug":21,"type":15},"Multi-Agent","multi-agent",2,"https:\u002F\u002Fgithub.com\u002Felastic\u002Fsourcerer","2026-08-13T05:04:19.431744",null,[],{"repoUrl":23,"stars":22,"forks":22,"topics":28,"description":29},[],"Sourcerer answers questions about your software from the source.","https:\u002F\u002Fgithub.com\u002Felastic\u002Fsourcerer\u002Ftree\u002FHEAD\u002Fsrc\u002Fsourcerer\u002Fskills\u002Fsubagent-orchestration","---\nname: \"subagent-orchestration\"\ndescription: \"Use when a request decomposes into multiple independent subtasks that can be worked on separately and combined afterward.\"\n---\n# Parallel subagent delegation\n\n## Default: delegate as one task\nUnless you already know of multiple distinct, already-located targets, delegate the whole investigation to a single subagent as one task. A subagent working serially through a multi-step investigation costs less than N subagents each paying their own fixed overhead (system prompt, skill instructions, tool schemas) - and can't redundantly re-search ground it's already covered, the way parallel subagents can.\n\n## Cost test (apply before fanning out to more than one)\nEach subagent pays its own token cost for every search it runs, plus its own fixed overhead. If two or more candidate tasks would plausibly search the same files or directories to get oriented, spawning them separately means paying that discovery cost once per subagent instead of once total. Shared search space is redundant spend, regardless of whether the tasks are otherwise independent.\n\nBefore writing more than one task, ask: **would these subagents likely re-derive the same location or content to do their job?** If yes:\n- **Locate first, fan out second.** Resolve the shared target yourself (or with one subagent call) - get the actual file paths, directories, or line ranges - before spawning anything else.\n- **Hand subagents targets, not search problems.** Once located, write each task around a concrete pointer (\"read `x-pack\u002F...\u002Fesql_tool.ts` lines 40-120 and explain X\") rather than a keyword strategy (\"search for terms like esql, execute_query, ... in agent builder directories\"). A subagent given a known path does one cheap read call; a subagent given a search problem burns tokens on multiple exploratory searches - and burns them again per subagent if several are all guessing at the same thing.\n- **Only fan out once discovery is done and targets have diverged.** If after locating things there's genuinely nothing left to parallelize (one file, one concern), stick with a single task.\n\n## How to use\n1. Locate the relevant files\u002Fdirectories first - directly, or with a single subagent call - until you have concrete targets, not just a search strategy.\n2. Decide whether more than one task is actually warranted using the cost test above. Default to one.\n3. If more than one, break the work into tasks each built around one already-known target (a specific path, directory, or narrow sub-question) - one per subagent. Count = number of distinct targets, not number of ways to search for the same thing.\n4. Call the `sourcerer.subagents.spawn` tool with `tasks` set to that list.\n5. The tool's response contains each subagent's raw result. Compose your final answer directly from them.\n\n## When NOT to use\n- Sequential tasks where step 2 depends on step 1's output.\n- Any task set where you don't yet have concrete targets - locate first; fanning out multiple open-ended searches over the same unknown location is the single most expensive pattern this skill can produce.\n- A single target\u002Fconcern - just answer directly or delegate as one task; the fan-out overhead isn't worth it.\n- Very large fan-outs (15-20+) - reconsider whether that many distinct targets actually exist.",{"data":33,"body":34},{"name":4,"description":6},{"type":35,"children":36},"root",[37,46,53,59,65,70,83,127,133,178,184],{"type":38,"tag":39,"props":40,"children":42},"element","h1",{"id":41},"parallel-subagent-delegation",[43],{"type":44,"value":45},"text","Parallel subagent delegation",{"type":38,"tag":47,"props":48,"children":50},"h2",{"id":49},"default-delegate-as-one-task",[51],{"type":44,"value":52},"Default: delegate as one task",{"type":38,"tag":54,"props":55,"children":56},"p",{},[57],{"type":44,"value":58},"Unless you already know of multiple distinct, already-located targets, delegate the whole investigation to a single subagent as one task. A subagent working serially through a multi-step investigation costs less than N subagents each paying their own fixed overhead (system prompt, skill instructions, tool schemas) - and can't redundantly re-search ground it's already covered, the way parallel subagents can.",{"type":38,"tag":47,"props":60,"children":62},{"id":61},"cost-test-apply-before-fanning-out-to-more-than-one",[63],{"type":44,"value":64},"Cost test (apply before fanning out to more than one)",{"type":38,"tag":54,"props":66,"children":67},{},[68],{"type":44,"value":69},"Each subagent pays its own token cost for every search it runs, plus its own fixed overhead. If two or more candidate tasks would plausibly search the same files or directories to get oriented, spawning them separately means paying that discovery cost once per subagent instead of once total. Shared search space is redundant spend, regardless of whether the tasks are otherwise independent.",{"type":38,"tag":54,"props":71,"children":72},{},[73,75,81],{"type":44,"value":74},"Before writing more than one task, ask: ",{"type":38,"tag":76,"props":77,"children":78},"strong",{},[79],{"type":44,"value":80},"would these subagents likely re-derive the same location or content to do their job?",{"type":44,"value":82}," If yes:",{"type":38,"tag":84,"props":85,"children":86},"ul",{},[87,98,117],{"type":38,"tag":88,"props":89,"children":90},"li",{},[91,96],{"type":38,"tag":76,"props":92,"children":93},{},[94],{"type":44,"value":95},"Locate first, fan out second.",{"type":44,"value":97}," Resolve the shared target yourself (or with one subagent call) - get the actual file paths, directories, or line ranges - before spawning anything else.",{"type":38,"tag":88,"props":99,"children":100},{},[101,106,108,115],{"type":38,"tag":76,"props":102,"children":103},{},[104],{"type":44,"value":105},"Hand subagents targets, not search problems.",{"type":44,"value":107}," Once located, write each task around a concrete pointer (\"read ",{"type":38,"tag":109,"props":110,"children":112},"code",{"className":111},[],[113],{"type":44,"value":114},"x-pack\u002F...\u002Fesql_tool.ts",{"type":44,"value":116}," lines 40-120 and explain X\") rather than a keyword strategy (\"search for terms like esql, execute_query, ... in agent builder directories\"). A subagent given a known path does one cheap read call; a subagent given a search problem burns tokens on multiple exploratory searches - and burns them again per subagent if several are all guessing at the same thing.",{"type":38,"tag":88,"props":118,"children":119},{},[120,125],{"type":38,"tag":76,"props":121,"children":122},{},[123],{"type":44,"value":124},"Only fan out once discovery is done and targets have diverged.",{"type":44,"value":126}," If after locating things there's genuinely nothing left to parallelize (one file, one concern), stick with a single task.",{"type":38,"tag":47,"props":128,"children":130},{"id":129},"how-to-use",[131],{"type":44,"value":132},"How to use",{"type":38,"tag":134,"props":135,"children":136},"ol",{},[137,142,147,152,173],{"type":38,"tag":88,"props":138,"children":139},{},[140],{"type":44,"value":141},"Locate the relevant files\u002Fdirectories first - directly, or with a single subagent call - until you have concrete targets, not just a search strategy.",{"type":38,"tag":88,"props":143,"children":144},{},[145],{"type":44,"value":146},"Decide whether more than one task is actually warranted using the cost test above. Default to one.",{"type":38,"tag":88,"props":148,"children":149},{},[150],{"type":44,"value":151},"If more than one, break the work into tasks each built around one already-known target (a specific path, directory, or narrow sub-question) - one per subagent. Count = number of distinct targets, not number of ways to search for the same thing.",{"type":38,"tag":88,"props":153,"children":154},{},[155,157,163,165,171],{"type":44,"value":156},"Call the ",{"type":38,"tag":109,"props":158,"children":160},{"className":159},[],[161],{"type":44,"value":162},"sourcerer.subagents.spawn",{"type":44,"value":164}," tool with ",{"type":38,"tag":109,"props":166,"children":168},{"className":167},[],[169],{"type":44,"value":170},"tasks",{"type":44,"value":172}," set to that list.",{"type":38,"tag":88,"props":174,"children":175},{},[176],{"type":44,"value":177},"The tool's response contains each subagent's raw result. Compose your final answer directly from them.",{"type":38,"tag":47,"props":179,"children":181},{"id":180},"when-not-to-use",[182],{"type":44,"value":183},"When NOT to use",{"type":38,"tag":84,"props":185,"children":186},{},[187,192,197,202],{"type":38,"tag":88,"props":188,"children":189},{},[190],{"type":44,"value":191},"Sequential tasks where step 2 depends on step 1's output.",{"type":38,"tag":88,"props":193,"children":194},{},[195],{"type":44,"value":196},"Any task set where you don't yet have concrete targets - locate first; fanning out multiple open-ended searches over the same unknown location is the single most expensive pattern this skill can produce.",{"type":38,"tag":88,"props":198,"children":199},{},[200],{"type":44,"value":201},"A single target\u002Fconcern - just answer directly or delegate as one task; the fan-out overhead isn't worth it.",{"type":38,"tag":88,"props":203,"children":204},{},[205],{"type":44,"value":206},"Very large fan-outs (15-20+) - reconsider whether that many distinct targets actually exist.",{"items":208,"total":259},[209,222,233,244,253],{"slug":210,"name":210,"fn":211,"description":212,"org":213,"tags":214,"stars":22,"repoUrl":23,"updatedAt":221},"code-citations","cite code symbols and repository claims","Use whenever an answer names a code symbol, class, function, config key, endpoint, file, directory, or file extension - or makes any checkable claim about repository behavior (a default value, a version, a recall\u002Flatency figure, a behavioral guarantee). This is nearly every substantive answer this agent produces, so treat it as the default path, not an edge case reserved for explicit \"cite your sources\" requests.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[215,218],{"name":216,"slug":217,"type":15},"Code Analysis","code-analysis",{"name":219,"slug":220,"type":15},"Documentation","documentation","2026-08-07T05:03:32.676248",{"slug":223,"name":223,"fn":224,"description":225,"org":226,"tags":227,"stars":22,"repoUrl":23,"updatedAt":232},"code-search","search and explore repository code","Use when you need to find where something is defined, understand how a feature is implemented, or explore an unfamiliar repo. Guides which tools to use and in what order.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[228,229],{"name":216,"slug":217,"type":15},{"name":230,"slug":231,"type":15},"Search","search","2026-08-13T04:31:28.616389",{"slug":234,"name":234,"fn":235,"description":236,"org":237,"tags":238,"stars":22,"repoUrl":23,"updatedAt":243},"ref-resolution","resolve ambiguous repository references","Use any time a branch, tag, or version in the user's question is ambiguous or unspecified - whether at the start of a question or mid-conversation (e.g. when comparing behavior across versions).",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[239,240],{"name":216,"slug":217,"type":15},{"name":241,"slug":242,"type":15},"Git","git","2026-08-10T04:16:18.220644",{"slug":245,"name":245,"fn":246,"description":247,"org":248,"tags":249,"stars":22,"repoUrl":23,"updatedAt":252},"repo-discovery","discover indexed repositories","Use at the start of most questions to identify which repos are indexed before querying content. Also use mid-conversation when an answer may require expanding into a related or upstream repo (e.g. tracing from an application repo into a lower-level dependency).",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[250,251],{"name":216,"slug":217,"type":15},{"name":241,"slug":242,"type":15},"2026-08-13T04:31:29.594165",{"slug":4,"name":4,"fn":5,"description":6,"org":254,"tags":255,"stars":22,"repoUrl":23,"updatedAt":24},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[256,257,258],{"name":13,"slug":14,"type":15},{"name":20,"slug":21,"type":15},{"name":17,"slug":18,"type":15},5,{"items":261,"total":433},[262,281,298,313,332,344,354,369,381,396,407,420],{"slug":263,"name":263,"fn":264,"description":265,"org":266,"tags":267,"stars":278,"repoUrl":279,"updatedAt":280},"accessing-benchmark-results","retrieve and analyze Rally benchmark results","Retrieve Rally benchmark results from an external Elasticsearch metrics store. Use to list past races, get a single race's overall (per-task) results, chart a metric's trend across multiple runs, compare two races, or check whether a run converged — e.g. \"show me recent geonames races\", \"what's the service_time trend for nyc_taxis over the last 30 days?\", \"compare these two race-ids\". Applies when datastore.type = elasticsearch is set in ~\u002F.rally\u002Frally.ini.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[268,271,274,275],{"name":269,"slug":270,"type":15},"Analytics","analytics",{"name":272,"slug":273,"type":15},"Data Analysis","data-analysis",{"name":9,"slug":8,"type":15},{"name":276,"slug":277,"type":15},"Performance","performance",2029,"https:\u002F\u002Fgithub.com\u002Felastic\u002Frally","2026-07-12T07:46:38.54144",{"slug":282,"name":282,"fn":283,"description":284,"org":285,"tags":286,"stars":278,"repoUrl":279,"updatedAt":297},"developing-rally","develop and debug Rally source code","Work on Rally's own codebase, not running benchmarks with it. Use when setting up the dev environment, running Rally's tests or linters, navigating its source, debugging Rally's own code, or making changes to Rally itself.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[287,290,291,294],{"name":288,"slug":289,"type":15},"Debugging","debugging",{"name":9,"slug":8,"type":15},{"name":292,"slug":293,"type":15},"Engineering","engineering",{"name":295,"slug":296,"type":15},"Local Development","local-development","2026-07-12T07:46:35.976807",{"slug":299,"name":299,"fn":300,"description":301,"org":302,"tags":303,"stars":278,"repoUrl":279,"updatedAt":312},"running-benchmarks","run Rally benchmarks against Elasticsearch","Run Rally benchmarks (races) against Elasticsearch — an existing\u002Fexternal cluster or a Rally-provisioned distribution — and read the summary report. Use when running a race (any pipeline, track, challenge, target-hosts, or auth) or when interpreting throughput, latency, and service_time results.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[304,305,308,309],{"name":9,"slug":8,"type":15},{"name":306,"slug":307,"type":15},"Elasticsearch","elasticsearch",{"name":276,"slug":277,"type":15},{"name":310,"slug":311,"type":15},"Testing","testing","2026-07-12T07:46:37.277964",{"slug":314,"name":314,"fn":315,"description":316,"org":317,"tags":318,"stars":329,"repoUrl":330,"updatedAt":331},"cloud-access-management","manage Elastic Cloud organization access","Manage Elastic Cloud organization access: invite users, assign roles to Serverless projects, and create or revoke Cloud API keys. Use when granting, modifying, or auditing user access.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[319,322,323,326],{"name":320,"slug":321,"type":15},"Cloud","cloud",{"name":9,"slug":8,"type":15},{"name":324,"slug":325,"type":15},"Operations","operations",{"name":327,"slug":328,"type":15},"Permissions","permissions",547,"https:\u002F\u002Fgithub.com\u002Felastic\u002Fagent-skills","2026-07-12T07:46:44.946285",{"slug":333,"name":333,"fn":334,"description":335,"org":336,"tags":337,"stars":329,"repoUrl":330,"updatedAt":343},"cloud-create-project","create Elastic Cloud Serverless projects","Creates Elastic Cloud Serverless projects (Elasticsearch, Observability, or Security) via the REST API, saves credentials to file, and bootstraps a scoped Elasticsearch API key. Use when creating a new serverless project, provisioning a search or observability environment, or spinning up a new Elastic Cloud project.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[338,339,342],{"name":320,"slug":321,"type":15},{"name":340,"slug":341,"type":15},"Deployment","deployment",{"name":306,"slug":307,"type":15},"2026-07-12T07:46:42.353362",{"slug":345,"name":345,"fn":346,"description":347,"org":348,"tags":349,"stars":329,"repoUrl":330,"updatedAt":353},"cloud-manage-project","manage Elastic Cloud Serverless projects","Manages existing Elastic Cloud Serverless projects: list, get, update, delete, reset credentials, resume, and load saved credentials. Connects to existing projects by resolving endpoints and acquiring scoped Elasticsearch API keys. Use when performing day-2 operations on serverless projects, connecting to an existing project, loading or resetting project credentials, or looking up project details.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[350,351,352],{"name":320,"slug":321,"type":15},{"name":306,"slug":307,"type":15},{"name":324,"slug":325,"type":15},"2026-07-12T07:46:41.097412",{"slug":355,"name":355,"fn":356,"description":357,"org":358,"tags":359,"stars":329,"repoUrl":330,"updatedAt":368},"cloud-network-security","manage Elastic Cloud network security","Manage Serverless network security (traffic filters): create, update, and delete IP filters and AWS PrivateLink VPC filters. Use when restricting network access or configuring private connectivity.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[360,361,362,365],{"name":320,"slug":321,"type":15},{"name":306,"slug":307,"type":15},{"name":363,"slug":364,"type":15},"Networking","networking",{"name":366,"slug":367,"type":15},"Security","security","2026-07-12T07:46:43.675992",{"slug":370,"name":370,"fn":371,"description":372,"org":373,"tags":374,"stars":329,"repoUrl":330,"updatedAt":380},"cloud-setup","configure Elastic Cloud authentication","Configures Elastic Cloud authentication and environment defaults. Use when setting up EC_API_KEY, configuring Cloud API access, or when another cloud skill requires credentials.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[375,378,379],{"name":376,"slug":377,"type":15},"Authentication","authentication",{"name":320,"slug":321,"type":15},{"name":306,"slug":307,"type":15},"2026-07-12T07:46:39.783105",{"slug":382,"name":382,"fn":383,"description":384,"org":385,"tags":386,"stars":329,"repoUrl":330,"updatedAt":395},"elasticsearch-audit","configure Elasticsearch security audit logs","Enable, configure, and query Elasticsearch security audit logs. Use when the task involves audit logging setup, event filtering, or investigating security incidents like failed logins.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[387,390,391,394],{"name":388,"slug":389,"type":15},"Audit","audit",{"name":306,"slug":307,"type":15},{"name":392,"slug":393,"type":15},"Logs","logs",{"name":366,"slug":367,"type":15},"2026-07-12T07:47:35.092599",{"slug":397,"name":397,"fn":398,"description":399,"org":400,"tags":401,"stars":329,"repoUrl":330,"updatedAt":406},"elasticsearch-authn","configure Elasticsearch authentication realms","Authenticate to Elasticsearch using native, file-based, LDAP\u002FAD, SAML, OIDC, Kerberos, JWT, or certificate realms. Use when connecting with credentials, choosing a realm, or managing API keys. Assumes the target realms are already configured.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[402,403,404,405],{"name":376,"slug":377,"type":15},{"name":9,"slug":8,"type":15},{"name":306,"slug":307,"type":15},{"name":366,"slug":367,"type":15},"2026-07-12T07:47:41.474547",{"slug":408,"name":408,"fn":409,"description":410,"org":411,"tags":412,"stars":329,"repoUrl":330,"updatedAt":419},"elasticsearch-authz","manage Elasticsearch RBAC and security roles","Manage Elasticsearch RBAC: native users, roles, role mappings, document- and field-level security. Use when creating users or roles, assigning privileges, or mapping external realms like LDAP\u002FSAML.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[413,414,415,418],{"name":9,"slug":8,"type":15},{"name":306,"slug":307,"type":15},{"name":416,"slug":417,"type":15},"RBAC","rbac",{"name":366,"slug":367,"type":15},"2026-07-12T07:47:36.394177",{"slug":421,"name":421,"fn":422,"description":423,"org":424,"tags":425,"stars":329,"repoUrl":330,"updatedAt":432},"elasticsearch-esql","query Elasticsearch data with ES|QL","Execute ES|QL (Elasticsearch Query Language) queries, use when the user wants to query Elasticsearch data, analyze logs, aggregate metrics, explore data, or create charts and dashboards from ES|QL results.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[426,427,428,429],{"name":269,"slug":270,"type":15},{"name":272,"slug":273,"type":15},{"name":306,"slug":307,"type":15},{"name":430,"slug":431,"type":15},"SQL","sql","2026-07-12T07:47:40.249533",92]