[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M75P025M3Y7QNTZFEZTH54V2":3,"skill-same:01M75P025M3Y7QNTZFEZTH54V2":49},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":27,"commitSha":28,"skillDirectory":29,"forksCount":30,"repositoryHtmlUrl":31,"repositoryHomepageUrl":6,"ownerHtmlUrl":32,"ownerBlogUrl":33,"ownerDescription":6,"updatedAt":34,"install":35,"resources":48},"01M75P025M3Y7QNTZFEZTH54V2","activation-contracts",null,"Use when a role must run without an interactive request, when wiring CI or a scheduler to an agent, or when choosing triggered versus interactive dispatch. Defines the event → dispatch → output contract.","pixeloven\u002Fcrew","pixeloven","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F5751260?v=4","01M75P0048G371B0AWRKQ7FJN8","PixelOven","01M75P0051RK1ZEY0SA9HCMX2Z",1,"MIT","https:\u002F\u002Fgithub.com\u002Fpixeloven\u002Fcrew\u002Fblob\u002Fmain\u002Fskills\u002Factivation-contracts\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\nname: activation-contracts\ndescription: Use when a role must run without an interactive request, when wiring CI or a scheduler to an agent, or when choosing triggered versus interactive dispatch. Defines the event → dispatch → output contract.\ntier: concept\nrequires: [external:github]\n---\n\nRoles describe autonomy readily — \"runs on a schedule\", \"fires on alerts\", \"routes incoming work\" — and then wait to be asked, because nothing wires the event to the dispatch. A role with no trigger is a role that only runs when a human remembers it exists, which in practice means whichever session notices absorbs its work.\n\nThis skill is the contract that makes a trigger portable, and reference wiring for the two shapes that cover most cases.\n\n## Two activation models\n\nA role can be either, or both: intake and diagnosis are triggered work that someone can also dispatch by hand. Say which in your routing table, because they need different things from you.\n\n| Model | Woken by | Needs | Typical roles |\n|---|---|---|---|\n| **Interactive dispatch** | the agent you're talking to | nothing — the routing table is enough | planning, implementation, review |\n| **Autonomous trigger** | an event or a schedule, with no human in the loop | infrastructure you deploy | intake, diagnosis, curation |\n\nA routing table that lists both without distinguishing them implies every role is one dispatch away, or hides a dispatchable role behind a trigger label. A role marked \"dispatch or trigger\" is one dispatch away now, and its trigger fires only once wired; an unwired trigger is why triggered work stays dormant.\n\nA planning role such as Lead is neither: it is the session agent, started as the session (on Claude Code, `claude --agent crew:lead`) rather than dispatched or triggered.\n\n## The trigger interface\n\nAny activation — a CI workflow, a cron schedule, a webhook — supplies the same four things. Keep them separable so the wiring can change without touching the role.\n\n1. **Event** — what happened, with enough payload to act on (issue number, alert name, run id). Normalise it: the role shouldn't parse three different shapes for \"something needs triage\".\n2. **Dispatch** — invoke the role with a packet meeting the contract in `orchestration-patterns`. A triggered worker starts cold *and* has no human to ask, so under-specifying here fails silently rather than loudly.\n3. **Output location** — where the result lands: an issue comment, a label, a review note, a file. Decide before wiring. A triggered role whose output goes nowhere observable is indistinguishable from one that never ran.\n4. **Failure signal** — what happens when the run fails. Silence is the default and the wrong one; a trigger that fails quietly is worse than no trigger, because it looks like nothing needed doing.\n\nSet a **budget** on every trigger — turn or token caps where the harness offers them. Triggered work has no human watching cost accumulate.\n\n## Choosing what to trigger\n\nTrigger a role when its work is **recurring, bounded, and cheap to be wrong about**. Intake classification, health sweeps, and lint triage all qualify: they run often, each run is small, and a mistake costs a comment rather than a commit.\n\nDon't trigger work that needs judgment about scope, touches protected seams, or writes to anything durable without review. Those want a human deciding to start them.\n\n**Why wiring them pays.** Left untriggered, a role's work doesn't vanish — it gets absorbed into whichever session happens to notice, which is the one already carrying a plan's worth of context. What you lose there is isolation: intake classification and health sweeps land in a working context that has to keep holding everything else, and the session blocks on work nobody needed it to do. That is the argument for wiring them, and it is also why *not* delegating to them mid-session is a real cost rather than a stylistic preference.\n\nThe cost argument on top of it is **the consumer's to make true, not this foundation's**. Roles ship with no model pinned — they inherit whatever the dispatching session or trigger runs — so a triggered role is only cheaper if you route it that way. If cheap-tier triggers are the point, set the model where your harness selects it (a trigger's own config, a gateway key, a subagent-model default); don't assume the roles arrived that way.\n\n## Reference wiring\n\nTwo templates, both starting points rather than drop-ins:\n\n- **`templates\u002Factivation\u002Fgithub-action-intake.yml`** — issue and PR events → classify, label, route. The most common first trigger, and the cheapest to prove out.\n- **`templates\u002Factivation\u002Fscheduled-sweep.yaml`** — a Kubernetes CronJob for periodic diagnosis or curation. Adapt the schedule and the runner image; the shape is what matters.\n\nBoth write results where a human will see them, and both fail loudly.\n\n## Prerequisites that fail silently\n\nWire these before wiring the trigger, or the first run will look like a no-op:\n\n- **Labels must exist.** A classifier applying `domain:*` labels to a repo without them silently applies nothing. Create the taxonomy first.\n- **The runner needs credentials scoped to its job** — read the event, write the output, nothing more. A triggered role is unattended; least privilege is the whole safety margin.\n- **Deduplicate.** A schedule that files an issue each run produces a queue of duplicates within a week. Check for an open one first.\n- **Verify by observing, not by config.** After wiring, trigger it once for real and confirm the output landed. The `doctor` skill's shadowing check exists because config that looks right and does nothing is the recurring failure here.\n","Roles describe autonomy readily — \"runs on a schedule\", \"fires on alerts\", \"routes incoming work\" — and then wait to be asked, because nothing wires the event to the dispatch. A role with no trigger is a role that only runs when a human remembers it exists, which in practice means whichever session notices absorbs its work.\n\nThis skill is the contract that makes a trigger portable, and reference wiring for the two shapes that cover most cases.\n\n## Two activation models\n\nA role can be either, or both: intake and diagnosis are triggered work that someone can also dispatch by hand. Say which in your routing table, because they need different things from you.\n\n| Model | Woken by | Needs | Typical roles |\n|---|---|---|---|\n| **Interactive dispatch** | the agent you're talking to | nothing — the routing table is enough | planning, implementation, review |\n| **Autonomous trigger** | an event or a schedule, with no human in the loop | infrastructure you deploy | intake, diagnosis, curation |\n\nA routing table that lists both without distinguishing them implies every role is one dispatch away, or hides a dispatchable role behind a trigger label. A role marked \"dispatch or trigger\" is one dispatch away now, and its trigger fires only once wired; an unwired trigger is why triggered work stays dormant.\n\nA planning role such as Lead is neither: it is the session agent, started as the session (on Claude Code, `claude --agent crew:lead`) rather than dispatched or triggered.\n\n## The trigger interface\n\nAny activation — a CI workflow, a cron schedule, a webhook — supplies the same four things. Keep them separable so the wiring can change without touching the role.\n\n1. **Event** — what happened, with enough payload to act on (issue number, alert name, run id). Normalise it: the role shouldn't parse three different shapes for \"something needs triage\".\n2. **Dispatch** — invoke the role with a packet meeting the contract in `orchestration-patterns`. A triggered worker starts cold *and* has no human to ask, so under-specifying here fails silently rather than loudly.\n3. **Output location** — where the result lands: an issue comment, a label, a review note, a file. Decide before wiring. A triggered role whose output goes nowhere observable is indistinguishable from one that never ran.\n4. **Failure signal** — what happens when the run fails. Silence is the default and the wrong one; a trigger that fails quietly is worse than no trigger, because it looks like nothing needed doing.\n\nSet a **budget** on every trigger — turn or token caps where the harness offers them. Triggered work has no human watching cost accumulate.\n\n## Choosing what to trigger\n\nTrigger a role when its work is **recurring, bounded, and cheap to be wrong about**. Intake classification, health sweeps, and lint triage all qualify: they run often, each run is small, and a mistake costs a comment rather than a commit.\n\nDon't trigger work that needs judgment about scope, touches protected seams, or writes to anything durable without review. Those want a human deciding to start them.\n\n**Why wiring them pays.** Left untriggered, a role's work doesn't vanish — it gets absorbed into whichever session happens to notice, which is the one already carrying a plan's worth of context. What you lose there is isolation: intake classification and health sweeps land in a working context that has to keep holding everything else, and the session blocks on work nobody needed it to do. That is the argument for wiring them, and it is also why *not* delegating to them mid-session is a real cost rather than a stylistic preference.\n\nThe cost argument on top of it is **the consumer's to make true, not this foundation's**. Roles ship with no model pinned — they inherit whatever the dispatching session or trigger runs — so a triggered role is only cheaper if you route it that way. If cheap-tier triggers are the point, set the model where your harness selects it (a trigger's own config, a gateway key, a subagent-model default); don't assume the roles arrived that way.\n\n## Reference wiring\n\nTwo templates, both starting points rather than drop-ins:\n\n- **`templates\u002Factivation\u002Fgithub-action-intake.yml`** — issue and PR events → classify, label, route. The most common first trigger, and the cheapest to prove out.\n- **`templates\u002Factivation\u002Fscheduled-sweep.yaml`** — a Kubernetes CronJob for periodic diagnosis or curation. Adapt the schedule and the runner image; the shape is what matters.\n\nBoth write results where a human will see them, and both fail loudly.\n\n## Prerequisites that fail silently\n\nWire these before wiring the trigger, or the first run will look like a no-op:\n\n- **Labels must exist.** A classifier applying `domain:*` labels to a repo without them silently applies nothing. Create the taxonomy first.\n- **The runner needs credentials scoped to its job** — read the event, write the output, nothing more. A triggered role is unattended; least privilege is the whole safety margin.\n- **Deduplicate.** A schedule that files an issue each run produces a queue of duplicates within a week. Check for an open one first.\n- **Verify by observing, not by config.** After wiring, trigger it once for real and confirm the output landed. The `doctor` skill's shadowing check exists because config that looks right and does nothing is the recurring failure here.\n",{"name":5,"tier":24,"requires":25,"description":7},"concept",[26],"external:github","main","686d42537f9efddb1aeece969b1a4fee2e4d1a88","skills\u002Factivation-contracts",0,"https:\u002F\u002Fgithub.com\u002Fpixeloven\u002Fcrew","https:\u002F\u002Fgithub.com\u002Fpixeloven","www.pixeloven.com","2026-11-07 09:18:41",{"source":8,"skillName":5,"commands":36,"prompt":46,"zipAvailable":47},[37,40,43],{"tool":38,"command":39},"npx","npx skills add pixeloven\u002Fcrew --skill activation-contracts",{"tool":41,"command":42},"bunx","bunx skills add pixeloven\u002Fcrew --skill activation-contracts",{"tool":44,"command":45},"pnpm","pnpm dlx skills add pixeloven\u002Fcrew --skill activation-contracts","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fpixeloven\u002Fcrew\u002Fblob\u002Fmain\u002Fskills\u002Factivation-contracts\u002FSKILL.md",true,[],[50,57,64,71,77,83],{"skillNo":51,"name":52,"nameZh":6,"description":53,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":54,"categories":55,"keywords":56,"publishedAt":19},"01M75P2X72QCY4H2H6C7D3KQ1Z","orchard-gh-runs","Use when reading a failed GitHub Actions run with the gh CLI in this project, before re-running it or reporting a CI failure, so the failing step's log is read instead of the whole run.","https:\u002F\u002Fgithub.com\u002Fpixeloven\u002Fcrew\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fproposals\u002Forchard\u002F.agents\u002Fskills\u002Forchard-gh-runs\u002FSKILL.md",[],[],{"skillNo":58,"name":59,"nameZh":6,"description":60,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":61,"categories":62,"keywords":63,"publishedAt":19},"01M75P2T76B22B314Q3W25GDCK","orchard-gh","Hand-edited after promotion; the learned marker was dropped, so Doctor and check_skills --local can no longer see it.","https:\u002F\u002Fgithub.com\u002Fpixeloven\u002Fcrew\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fproposals\u002Finvalid\u002F.agents\u002Fskills\u002Forchard-gh\u002FSKILL.md",[],[],{"skillNo":65,"name":66,"nameZh":6,"description":67,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":68,"categories":69,"keywords":70,"publishedAt":19},"01M75P2Q410WX0H3ZZ61R210F4","demo-topology","Use when exercising the inventory contract fixture so that this fixture description comfortably clears the schema-v2 minimum length.","https:\u002F\u002Fgithub.com\u002Fpixeloven\u002Fcrew\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Finventory\u002Fproject\u002F.agents\u002Fskills\u002Fdemo-topology\u002FSKILL.md",[],[],{"skillNo":72,"name":73,"nameZh":6,"description":67,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":74,"categories":75,"keywords":76,"publishedAt":19},"01M75P2KTWZ94KNSKHPGFB6Y6J","helper-jq","https:\u002F\u002Fgithub.com\u002Fpixeloven\u002Fcrew\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Finventory\u002Fhome\u002F.codex\u002Fplugins\u002Fcache\u002Facme\u002Fhelper\u002F2.0.0\u002Fskills\u002Fhelper-jq\u002FSKILL.md",[],[],{"skillNo":78,"name":79,"nameZh":6,"description":67,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":80,"categories":81,"keywords":82,"publishedAt":19},"01M75P2GT2CXZR4EZE2DS9GTS7","user-notes","https:\u002F\u002Fgithub.com\u002Fpixeloven\u002Fcrew\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Finventory\u002Fhome\u002F.claude\u002Fskills\u002Fuser-notes\u002FSKILL.md",[],[],{"skillNo":84,"name":85,"nameZh":6,"description":67,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":86,"categories":87,"keywords":88,"publishedAt":19},"01M75P2D61GDSH9NFNEBV4HDBS","toolkit-gh","https:\u002F\u002Fgithub.com\u002Fpixeloven\u002Fcrew\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Finventory\u002Fhome\u002F.claude\u002Fplugins\u002Fcache\u002Facme\u002Ftoolkit\u002F1.2.0\u002Fskills\u002Ftoolkit-gh\u002FSKILL.md",[],[]]