[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7BKTY15MNMCTF776YVMZGGW":3,"skill-same:01M7BKTY15MNMCTF776YVMZGGW":99},{"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},"01M7BKTY15MNMCTF776YVMZGGW","experimentation-engineer",null,"Run DX experiments on quickstarts, CLI output variants, error formatting, and onboarding flows, measuring task completion, time, error rate, drop-off, support load, and retention, with OpenFeature-style flagging as the reference. For analyzing passive feedback signals use developer-feedback-analyst.","devarispbrown\u002Fdevex-skills","devarispbrown","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F769307?v=4","01M7BKQ227RE7WBV6G929609KC","DeVaris Brown","01M7BKQ23B73Z8CVNF4R1K072W",1,"MIT","https:\u002F\u002Fgithub.com\u002Fdevarispbrown\u002Fdevex-skills\u002Fblob\u002Fmain\u002Fskills\u002Fexperimentation-engineer\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: experimentation-engineer\ndescription: Run DX experiments on quickstarts, CLI output variants, error formatting, and onboarding flows, measuring task completion, time, error rate, drop-off, support load, and retention, with OpenFeature-style flagging as the reference. For analyzing passive feedback signals use developer-feedback-analyst.\nlicense: MIT\ncompatibility: Claude Code and Agent Skills-compatible coding agents; best with repository access and experimentation\u002Fflag tooling.\nmetadata:\n  version: \"2.9.4\"\n---\n\n# Experimentation Engineer\n\n## Mission\n\nRun controlled experiments on developer-facing surfaces — quickstarts, CLI output variants, error formatting, and onboarding flows — and decide from measured outcomes, not opinions.\n\nEvery experiment is a contract: one falsifiable hypothesis, exactly one control variant, predefined metrics with definitions, guardrails with thresholds, and a flag configuration. Run it, label the evidence, and ship, iterate, or kill based on the result.\n\nNever experiment on developers without guardrails. Never report an experiment that cannot decide anything.\n\nRead `references\u002Fstandards.md` for the canonical thresholds, severity vocabulary, and release gates.\n\n## DX experiment design\n\nDesign every experiment in this order. Do not implement before the plan is complete.\n\n### 1. Frame the hypothesis\n\nState exactly one falsifiable claim:\n\n- the surface changed: quickstart step count, CLI output layout, error message text, onboarding flow\n- the change under test\n- the primary metric it moves\n- the expected direction and the minimum meaningful effect\n\nA hypothesis that cannot fail cannot decide. If the expected effect size is not stated, estimate it before proceeding.\n\n### 2. Define the variants\n\nDefine exactly one control and one or more treatments. The control is the current shipped behavior and is never an afterthought.\n\n- every variant has a stable id, a flag key, and a value\n- exactly one variant is marked `control`\n- treatments differ from the control in exactly one dimension\n- record what the experiment does not cover and why\n\nChanging more than one dimension confounds the result. Split compound changes into separate experiments.\n\n### 3. Define the metrics\n\nPick one primary metric and a small set of secondary metrics. Definitions live in `references\u002Fmetric-definitions.md`; record a name, direction, and definition for every metric.\n\nDefine metrics before launch. A metric added after results arrive is post-hoc and cannot prove the hypothesis.\n\nFor onboarding work the suite magic-path gate remains the floor: a variant that pushes first value past `MAGIC_PATH_MAX_MIN` fails regardless of experiment results.\n\n### 4. Set guardrails\n\nDefine the thresholds that stop the experiment early:\n\n- a ceiling for error rate, drop-off, or support load\n- a floor for the control's completion rate\n- the comparison direction for every threshold\n- the trip action: stop exposure, revert to control, page the owner\n\nA guardrail trip is a P1 finding. Continue the experiment only after the root cause is understood and the plan is amended.\n\n### 5. Configure the flag\n\nImplement the experiment with provider-agnostic flagging per `references\u002Fopenfeature-patterns.md`.\n\n- one flag key, one control value, one treatment value\n- targeting by stable cohort with an even split at launch\n- exposure tracking on every evaluation\n- a kill switch that reverts every variant to control\n- no flag reuse across unrelated experiments\n\n### 6. Validate the manifest\n\nRecord the plan as an experiment manifest JSON and run `scripts\u002Fcheck_experiment_metrics.py` on it.\n\nThe checker exits 0 only when name, hypothesis, variants with exactly one control, metric definitions, guardrail thresholds, and flag config are all present.\n\nA manifest that fails the checker is a sketch, not an experiment. Fix it before launch.\n\n### 7. Run the experiment\n\nRun with the smallest sample that the expected effect and guardrails allow:\n\n- sessions are homogeneous: same task, same environment class, same instrumentation\n- assignment comes from the flag, never from self-selection\n- label every result Observed, CI-observed, or Estimated\n- collect completion, time, error, drop-off, support, and retention data per variant\n- watch guardrails continuously, not after the run\n\nObserved beats CI-observed beats Estimated. An estimate never proves a result.\n\n### 8. Analyze and decide\n\nCompare every treatment to control on the primary metric, then on secondary metrics and guardrails:\n\n- report statistical and practical significance\n- state per-variant evidence; never average across variants\n- report the effect on task completion, time, error rate, drop-off, support load, and retention\n- decide exactly one of: ship the treatment, iterate on a follow-up hypothesis, or kill it and keep control\n- record the decision, the evidence, and the flag state in the experiment report\n\nA result that does not move the primary metric within the expected effect is a kill or an iterate, not a ship.\n\n## Experiment contract\n\nAn experiment is launchable only when its manifest records:\n\n1. `name` — a unique identifier for the experiment\n2. `hypothesis` — the falsifiable claim with direction and expected effect\n3. `variants` — exactly one control and at least one treatment\n4. `metrics` — primary and secondary, each with a definition\n5. `guardrails` — thresholds, comparison direction, and trip action\n6. `flag` — key, values, targeting, rollout, kill switch\n\n## Evidence contract\n\nEvery number in the report carries one of the suite evidence labels:\n\n- **Observed:** executed from a clean or representative environment\n- **CI-observed:** executed in automation; may understate human time\n- **Estimated:** reasoned without execution; can never prove a result\n\nUnlabeled numbers are UNVERIFIED and cannot drive a ship decision.\n\n## References\n\n- `references\u002Fexperiment-design.md` — hypothesis, variants, guardrails, ethics of user experiments\n- `references\u002Fonboarding-experiments.md` — quickstart and onboarding A\u002FB patterns\n- `references\u002Fcli-variants.md` — CLI output and error formatting experiments\n- `references\u002Fmetric-definitions.md` — task completion, time, error rate, drop-off, support load, retention\n- `references\u002Fopenfeature-patterns.md` — OpenFeature-style flagging patterns\n\n## Required output\n\nProduce the experiment report from `assets\u002Fexperiment-plan-template.md` containing:\n\n1. Experiment identity, hypothesis, and variant table\n2. Metrics with definitions and evidence labels\n3. Guardrails and trip status\n4. Results per variant: task completion, time, error rate, drop-off, support load, retention\n5. Decision — ship, iterate, or kill — with rationale\n6. Flag state and rollout plan for the shipped variant\n\n## Definition of done\n\nAn experiment is done when:\n\n- the hypothesis is falsifiable and states direction and expected effect\n- exactly one control variant exists and treatments differ in one dimension\n- every metric has a definition recorded before launch\n- guardrails carry thresholds, comparison direction, and a trip action\n- the manifest passes `scripts\u002Fcheck_experiment_metrics.py` with exit 0\n- the run is complete and every result carries an evidence label\n- the report compares every treatment to control, never averaging variants\n- the decision is exactly one of ship, iterate, or kill, recorded with the flag state\n- guardrail trips are reported as P1 findings, never hidden in averages\n\nFor passive feedback signals — surveys, sentiment, support tickets — use the `developer-feedback-analyst` skill if available. Experimentation Engineer measures; it does not replace passive-signal analysis.\n","# Experimentation Engineer\n\n## Mission\n\nRun controlled experiments on developer-facing surfaces — quickstarts, CLI output variants, error formatting, and onboarding flows — and decide from measured outcomes, not opinions.\n\nEvery experiment is a contract: one falsifiable hypothesis, exactly one control variant, predefined metrics with definitions, guardrails with thresholds, and a flag configuration. Run it, label the evidence, and ship, iterate, or kill based on the result.\n\nNever experiment on developers without guardrails. Never report an experiment that cannot decide anything.\n\nRead `references\u002Fstandards.md` for the canonical thresholds, severity vocabulary, and release gates.\n\n## DX experiment design\n\nDesign every experiment in this order. Do not implement before the plan is complete.\n\n### 1. Frame the hypothesis\n\nState exactly one falsifiable claim:\n\n- the surface changed: quickstart step count, CLI output layout, error message text, onboarding flow\n- the change under test\n- the primary metric it moves\n- the expected direction and the minimum meaningful effect\n\nA hypothesis that cannot fail cannot decide. If the expected effect size is not stated, estimate it before proceeding.\n\n### 2. Define the variants\n\nDefine exactly one control and one or more treatments. The control is the current shipped behavior and is never an afterthought.\n\n- every variant has a stable id, a flag key, and a value\n- exactly one variant is marked `control`\n- treatments differ from the control in exactly one dimension\n- record what the experiment does not cover and why\n\nChanging more than one dimension confounds the result. Split compound changes into separate experiments.\n\n### 3. Define the metrics\n\nPick one primary metric and a small set of secondary metrics. Definitions live in `references\u002Fmetric-definitions.md`; record a name, direction, and definition for every metric.\n\nDefine metrics before launch. A metric added after results arrive is post-hoc and cannot prove the hypothesis.\n\nFor onboarding work the suite magic-path gate remains the floor: a variant that pushes first value past `MAGIC_PATH_MAX_MIN` fails regardless of experiment results.\n\n### 4. Set guardrails\n\nDefine the thresholds that stop the experiment early:\n\n- a ceiling for error rate, drop-off, or support load\n- a floor for the control's completion rate\n- the comparison direction for every threshold\n- the trip action: stop exposure, revert to control, page the owner\n\nA guardrail trip is a P1 finding. Continue the experiment only after the root cause is understood and the plan is amended.\n\n### 5. Configure the flag\n\nImplement the experiment with provider-agnostic flagging per `references\u002Fopenfeature-patterns.md`.\n\n- one flag key, one control value, one treatment value\n- targeting by stable cohort with an even split at launch\n- exposure tracking on every evaluation\n- a kill switch that reverts every variant to control\n- no flag reuse across unrelated experiments\n\n### 6. Validate the manifest\n\nRecord the plan as an experiment manifest JSON and run `scripts\u002Fcheck_experiment_metrics.py` on it.\n\nThe checker exits 0 only when name, hypothesis, variants with exactly one control, metric definitions, guardrail thresholds, and flag config are all present.\n\nA manifest that fails the checker is a sketch, not an experiment. Fix it before launch.\n\n### 7. Run the experiment\n\nRun with the smallest sample that the expected effect and guardrails allow:\n\n- sessions are homogeneous: same task, same environment class, same instrumentation\n- assignment comes from the flag, never from self-selection\n- label every result Observed, CI-observed, or Estimated\n- collect completion, time, error, drop-off, support, and retention data per variant\n- watch guardrails continuously, not after the run\n\nObserved beats CI-observed beats Estimated. An estimate never proves a result.\n\n### 8. Analyze and decide\n\nCompare every treatment to control on the primary metric, then on secondary metrics and guardrails:\n\n- report statistical and practical significance\n- state per-variant evidence; never average across variants\n- report the effect on task completion, time, error rate, drop-off, support load, and retention\n- decide exactly one of: ship the treatment, iterate on a follow-up hypothesis, or kill it and keep control\n- record the decision, the evidence, and the flag state in the experiment report\n\nA result that does not move the primary metric within the expected effect is a kill or an iterate, not a ship.\n\n## Experiment contract\n\nAn experiment is launchable only when its manifest records:\n\n1. `name` — a unique identifier for the experiment\n2. `hypothesis` — the falsifiable claim with direction and expected effect\n3. `variants` — exactly one control and at least one treatment\n4. `metrics` — primary and secondary, each with a definition\n5. `guardrails` — thresholds, comparison direction, and trip action\n6. `flag` — key, values, targeting, rollout, kill switch\n\n## Evidence contract\n\nEvery number in the report carries one of the suite evidence labels:\n\n- **Observed:** executed from a clean or representative environment\n- **CI-observed:** executed in automation; may understate human time\n- **Estimated:** reasoned without execution; can never prove a result\n\nUnlabeled numbers are UNVERIFIED and cannot drive a ship decision.\n\n## References\n\n- `references\u002Fexperiment-design.md` — hypothesis, variants, guardrails, ethics of user experiments\n- `references\u002Fonboarding-experiments.md` — quickstart and onboarding A\u002FB patterns\n- `references\u002Fcli-variants.md` — CLI output and error formatting experiments\n- `references\u002Fmetric-definitions.md` — task completion, time, error rate, drop-off, support load, retention\n- `references\u002Fopenfeature-patterns.md` — OpenFeature-style flagging patterns\n\n## Required output\n\nProduce the experiment report from `assets\u002Fexperiment-plan-template.md` containing:\n\n1. Experiment identity, hypothesis, and variant table\n2. Metrics with definitions and evidence labels\n3. Guardrails and trip status\n4. Results per variant: task completion, time, error rate, drop-off, support load, retention\n5. Decision — ship, iterate, or kill — with rationale\n6. Flag state and rollout plan for the shipped variant\n\n## Definition of done\n\nAn experiment is done when:\n\n- the hypothesis is falsifiable and states direction and expected effect\n- exactly one control variant exists and treatments differ in one dimension\n- every metric has a definition recorded before launch\n- guardrails carry thresholds, comparison direction, and a trip action\n- the manifest passes `scripts\u002Fcheck_experiment_metrics.py` with exit 0\n- the run is complete and every result carries an evidence label\n- the report compares every treatment to control, never averaging variants\n- the decision is exactly one of ship, iterate, or kill, recorded with the flag state\n- guardrail trips are reported as P1 findings, never hidden in averages\n\nFor passive feedback signals — surveys, sentiment, support tickets — use the `developer-feedback-analyst` skill if available. Experimentation Engineer measures; it does not replace passive-signal analysis.\n",{"name":5,"license":15,"metadata":24,"description":7,"compatibility":26},{"version":25},"2.9.4","Claude Code and Agent Skills-compatible coding agents; best with repository access and experimentation\u002Fflag tooling.","main","8ecb082bfab6f317fa3cada530206ffa7c2917bb","skills\u002Fexperimentation-engineer",0,"https:\u002F\u002Fgithub.com\u002Fdevarispbrown\u002Fdevex-skills","https:\u002F\u002Fgithub.com\u002Fdevarispbrown","floodgateacademy.com","2026-09-09 03:10:32",{"source":8,"skillName":5,"commands":36,"prompt":46,"zipAvailable":47},[37,40,43],{"tool":38,"command":39},"npx","npx skills add devarispbrown\u002Fdevex-skills --skill experimentation-engineer",{"tool":41,"command":42},"bunx","bunx skills add devarispbrown\u002Fdevex-skills --skill experimentation-engineer",{"tool":44,"command":45},"pnpm","pnpm dlx skills add devarispbrown\u002Fdevex-skills --skill experimentation-engineer","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fdevarispbrown\u002Fdevex-skills\u002Fblob\u002Fmain\u002Fskills\u002Fexperimentation-engineer\u002FSKILL.md",true,[49,55,59,63,67,73,77,81,85,89,93],{"path":50,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":53,"sourceHtmlUrl":54},"assets\u002Fexperiment-manifest.clean.json","ASSET","资产文件",1536,"https:\u002F\u002Fgithub.com\u002Fdevarispbrown\u002Fdevex-skills\u002Fblob\u002Fmain\u002Fskills\u002Fexperimentation-engineer\u002Fassets\u002Fexperiment-manifest.clean.json",{"path":56,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":57,"sourceHtmlUrl":58},"assets\u002Fexperiment-manifest.example.json",523,"https:\u002F\u002Fgithub.com\u002Fdevarispbrown\u002Fdevex-skills\u002Fblob\u002Fmain\u002Fskills\u002Fexperimentation-engineer\u002Fassets\u002Fexperiment-manifest.example.json",{"path":60,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":61,"sourceHtmlUrl":62},"assets\u002Fexperiment-plan-template.md",1338,"https:\u002F\u002Fgithub.com\u002Fdevarispbrown\u002Fdevex-skills\u002Fblob\u002Fmain\u002Fskills\u002Fexperimentation-engineer\u002Fassets\u002Fexperiment-plan-template.md",{"path":64,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":65,"sourceHtmlUrl":66},"assets\u002Fsmoke.json",167,"https:\u002F\u002Fgithub.com\u002Fdevarispbrown\u002Fdevex-skills\u002Fblob\u002Fmain\u002Fskills\u002Fexperimentation-engineer\u002Fassets\u002Fsmoke.json",{"path":68,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":71,"sourceHtmlUrl":72},"references\u002Fcli-variants.md","REFERENCE","参考文档",1935,"https:\u002F\u002Fgithub.com\u002Fdevarispbrown\u002Fdevex-skills\u002Fblob\u002Fmain\u002Fskills\u002Fexperimentation-engineer\u002Freferences\u002Fcli-variants.md",{"path":74,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":75,"sourceHtmlUrl":76},"references\u002Fexperiment-design.md",2991,"https:\u002F\u002Fgithub.com\u002Fdevarispbrown\u002Fdevex-skills\u002Fblob\u002Fmain\u002Fskills\u002Fexperimentation-engineer\u002Freferences\u002Fexperiment-design.md",{"path":78,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":79,"sourceHtmlUrl":80},"references\u002Fmetric-definitions.md",2580,"https:\u002F\u002Fgithub.com\u002Fdevarispbrown\u002Fdevex-skills\u002Fblob\u002Fmain\u002Fskills\u002Fexperimentation-engineer\u002Freferences\u002Fmetric-definitions.md",{"path":82,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":83,"sourceHtmlUrl":84},"references\u002Fonboarding-experiments.md",2315,"https:\u002F\u002Fgithub.com\u002Fdevarispbrown\u002Fdevex-skills\u002Fblob\u002Fmain\u002Fskills\u002Fexperimentation-engineer\u002Freferences\u002Fonboarding-experiments.md",{"path":86,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":87,"sourceHtmlUrl":88},"references\u002Fopenfeature-patterns.md",2326,"https:\u002F\u002Fgithub.com\u002Fdevarispbrown\u002Fdevex-skills\u002Fblob\u002Fmain\u002Fskills\u002Fexperimentation-engineer\u002Freferences\u002Fopenfeature-patterns.md",{"path":90,"resourceType":69,"resourceTypeDesc":70,"sizeBytes":91,"sourceHtmlUrl":92},"references\u002Fstandards.md",3067,"https:\u002F\u002Fgithub.com\u002Fdevarispbrown\u002Fdevex-skills\u002Fblob\u002Fmain\u002Fskills\u002Fexperimentation-engineer\u002Freferences\u002Fstandards.md",{"path":94,"resourceType":95,"resourceTypeDesc":96,"sizeBytes":97,"sourceHtmlUrl":98},"scripts\u002Fcheck_experiment_metrics.py","SCRIPT","脚本",3686,"https:\u002F\u002Fgithub.com\u002Fdevarispbrown\u002Fdevex-skills\u002Fblob\u002Fmain\u002Fskills\u002Fexperimentation-engineer\u002Fscripts\u002Fcheck_experiment_metrics.py",[100,106,113,119,125,131],{"skillNo":101,"name":102,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":103,"categories":104,"keywords":105,"publishedAt":19},"01M7BKX6J9JRE4333WETDZ02YA","web-console-dx","https:\u002F\u002Fgithub.com\u002Fdevarispbrown\u002Fdevex-skills\u002Fblob\u002Fmain\u002Fskills\u002Fweb-console-dx\u002FSKILL.md",[],[],{"skillNo":107,"name":108,"nameZh":6,"description":109,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":110,"categories":111,"keywords":112,"publishedAt":19},"01M7BKX16CMSFAF11E19Y3838A","test-data-and-fixtures","Author fixtures, seed data, factories, fake services, mock servers, record-replay cassettes, golden files, synthetic datasets, and sanitization, with a project test-data create pattern. For sandbox route design use sandbox-experience; for per-system-type test strategy use quality-engineer.","https:\u002F\u002Fgithub.com\u002Fdevarispbrown\u002Fdevex-skills\u002Fblob\u002Fmain\u002Fskills\u002Ftest-data-and-fixtures\u002FSKILL.md",[],[],{"skillNo":114,"name":115,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":116,"categories":117,"keywords":118,"publishedAt":19},"01M7BKWX1EA5BTM9CF8Y4K7PBJ","security-supply-chain","https:\u002F\u002Fgithub.com\u002Fdevarispbrown\u002Fdevex-skills\u002Fblob\u002Fmain\u002Fskills\u002Fsecurity-supply-chain\u002FSKILL.md",[],[],{"skillNo":120,"name":121,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":122,"categories":123,"keywords":124,"publishedAt":19},"01M7BKWS9R0SAZ1WK3JV0EFD9E","sdk-engineer","https:\u002F\u002Fgithub.com\u002Fdevarispbrown\u002Fdevex-skills\u002Fblob\u002Fmain\u002Fskills\u002Fsdk-engineer\u002FSKILL.md",[],[],{"skillNo":126,"name":127,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":128,"categories":129,"keywords":130,"publishedAt":19},"01M7BKWN2MAK1WN87H7VZD3KP6","sandbox-experience","https:\u002F\u002Fgithub.com\u002Fdevarispbrown\u002Fdevex-skills\u002Fblob\u002Fmain\u002Fskills\u002Fsandbox-experience\u002FSKILL.md",[],[],{"skillNo":132,"name":133,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":134,"categories":135,"keywords":136,"publishedAt":19},"01M7BKWH9WMK5M9H41802TGNJC","release-guardian","https:\u002F\u002Fgithub.com\u002Fdevarispbrown\u002Fdevex-skills\u002Fblob\u002Fmain\u002Fskills\u002Frelease-guardian\u002FSKILL.md",[],[]]