[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M77MNPVAWMMKV9QR1YYFNFNA":3,"skill-same:01M77MNPVAWMMKV9QR1YYFNFNA":55},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":17,"publishedAt":18,"linkOnly":19,"markdownContent":20,"markdownBody":21,"frontMatter":22,"refName":23,"commitSha":24,"skillDirectory":25,"forksCount":26,"repositoryHtmlUrl":27,"repositoryHomepageUrl":28,"ownerHtmlUrl":29,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":30,"install":31,"resources":44},"01M77MNPVAWMMKV9QR1YYFNFNA","hive-self-improve",null,"Load when improving the QUALITY of an already-shipped framework component — simplification, duplication and dead-code removal, test-coverage gaps, doc\u002Fconsistency drift, developer ergonomics, and performance — by running the real code to ground every finding, then converting each into a right-tiered ticket. NOT for security or trust-boundary hardening (that is hive-adversarial-improve) and NOT for routine single-diff review (that is the reviewer agent). Triggers on \"hive-self-improve this\", \"find quality gaps in \u003Ccomponent>\", \"what can we simplify\u002Fclean up in \u003Ccomponent>\", \"improve \u003Ccomponent>\".","HexingBot\u002Fhivemind","HexingBot","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F296674764?v=4","01M77MMXMW5SWSXKHB0VA4N1FF","01M77MMXNV3PZA3795D6PYX5C9",1,"MIT","https:\u002F\u002Fgithub.com\u002FHexingBot\u002Fhivemind\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fhive-self-improve\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: hive-self-improve\ndescription: Load when improving the QUALITY of an already-shipped framework component — simplification, duplication and dead-code removal, test-coverage gaps, doc\u002Fconsistency drift, developer ergonomics, and performance — by running the real code to ground every finding, then converting each into a right-tiered ticket. NOT for security or trust-boundary hardening (that is hive-adversarial-improve) and NOT for routine single-diff review (that is the reviewer agent). Triggers on \"hive-self-improve this\", \"find quality gaps in \u003Ccomponent>\", \"what can we simplify\u002Fclean up in \u003Ccomponent>\", \"improve \u003Ccomponent>\".\n---\n\n# Hive Self-Improve\n\n**Context guard — framework repo only.** This skill targets the hivemind **framework's own**\ncomponents (its `src\u002F`, `tests\u002F`, `tasks\u002F`) — not a downstream project built with hivemind. Before\nproceeding, confirm you are in the framework repo: call `isFrameworkRepo({ repoRoot })` from\n`src\u002Fframework-context.js` against the current working repo root. If it returns **false** (a\nconsumer project), **STOP** — do not proceed with this skill — and direct the user to the\n`hive-self-improve-current-project` variant instead, which runs the same protocol against the\nconsumer's own project code. If `src\u002Fframework-context.js` cannot be imported (e.g. a stripped\ncheckout), fall back to the equivalent check by hand: the repo root has a `.claude-plugin\u002Fplugin.json`\nwhose `name` field is `hivemind`, AND a `src\u002F` directory, AND a `bin\u002Finit.js` file. All three must\nhold for this skill to apply; if any is missing, STOP and route to the `-current-project` variant.\n\nThis skill turns an ad-hoc \"this could be cleaner \u002F better covered \u002F clearer\" conversation into a\nrepeatable, evidence-producing exercise. It is a **durable quality-improvement capability**, not a\none-off brainstorm: every run is grounded in the real shipped code actually executed, seeded from a\ncited improvement-dimension catalog, and converts every finding into a right-tiered ticket so the\nimprovement survives after the session ends.\n\nIt is the constructive counterpart to `hive-adversarial-improve`. That skill hardens a **trust\nboundary** against hostile inputs; this one raises the everyday quality of a component that is\nalready correct — the ~90% of \"make the framework better\" work that is not security-sensitive. The\ntwo are designed to run side by side: most components deserve a self-improve pass, and only the ones\nsitting on a trust boundary additionally deserve an adversarial one.\n\n## When to load this skill\n\nLoad whenever the task is to raise the quality of one specific, already-shipped component or\npipeline along a **constructive** dimension: it works, but it could be simpler, better covered,\nbetter documented, faster, or more ergonomic. Examples: collapsing duplicated logic across two\nmodules, closing a use-case coverage gap, reconciling a doc that has drifted from the code it\ndescribes, tightening a CLI's error messages.\n\nDo **not** load it for:\n\n- **Security or trust-boundary hardening** — where untrusted content crosses into trusted execution\n  (third-party skill adoption, tool-use dispatch, installers, parsers of fetched content). That is\n  `hive-adversarial-improve`. See the hard handoff in protocol step 5.\n- **Routine review of a single diff** — that is the `reviewer` agent.\n- **Open-ended research into an unfamiliar library or pattern** — that is the `researcher` agent.\n\n## The two load-bearing rules\n\nEverything below is scaffolding around these two rules. Drop either one and the exercise stops\nproducing durable value:\n\n1. **Every finding is grounded in the REAL code actually run — never speculation.** A finding must\n   point at concrete evidence you produced this session: a duplication with both `file:line`\n   locations, a coverage gap named against the specific acceptance criterion it leaves untested, a\n   doc-vs-code drift with both sides quoted, a timing number from an actual run. A quality review\n   that lists \"things that feel improvable\" without running the code or citing the artifact is a\n   brainstorm, not a self-improve pass — and it systematically ships cosmetic churn while missing\n   the real complexity.\n2. **Every surviving finding becomes a ticket at the LIGHTEST defensible verification tier, with\n   the evidence attached.** A finding that lives only in a chat transcript evaporates the moment the\n   session ends. Most self-improve findings are `tests-after` (behavior provable by running the\n   code, including schema\u002Fstate-mutation\u002Fparsing findings) or `uat-only` (docs, glue, ergonomics) —\n   see this repo's verification-tier rubric; the `tdd` tier it once described is retired (TASK-212,\n   2026-08-13 human decision). The concrete evidence from rule 1 is what makes the ticket actionable\n   rather than a vague \"clean this up.\"\n\n## The 7-step protocol\n\n1. **Name the component and pick the dimensions in scope.** Choose from\n   `references\u002Fimprovement-dimensions.md` — simplification\u002Freuse, test-coverage, doc\u002Fconsistency\n   drift, developer experience, performance, structural cohesion. Do not scan every dimension by\n   reflex; pick the ones that fit the component and say which you are deliberately skipping.\n2. **Establish a baseline by running the real thing.** Before proposing any change, capture the\n   current state as evidence: run the affected specs and record pass\u002Fskip counts, time the hot path\n   if performance is in scope, run the CLI and capture its actual output. This baseline is what\n   every later finding is measured against and what step 7 re-runs to prove the improvement landed.\n3. **Scan each in-scope dimension for concrete findings.** Work from the baseline, not from\n   imagination. Each finding carries its evidence inline (the duplicated block, the uncovered AC,\n   the drifted doc line). Grep and read the real code; do not infer structure you have not opened.\n4. **Triage against the useless-vs-valuable checklist.** Run every candidate through\n   `references\u002Fuseless-vs-valuable.md` and discard the cosmetic ones. Style-only reformatting,\n   speculative abstraction for an imagined future, and any \"improvement\" that would add a spec the\n   repo's new-test budget forbids are noise — dropping them is the point of this step.\n5. **Security handoff (hard gate).** If any finding touches a trust boundary — untrusted content\n   reaching trusted execution, an auth\u002Fvalidation seam, a scanner or gate that rejects hostile\n   input — STOP on that finding and route it to `hive-adversarial-improve` instead of handling it\n   here. This skill deliberately does not do adversarial security work; forcing a security finding\n   through the constructive path under-tests it. Record the handoff so the finding is not lost.\n6. **Convert each surviving finding into a right-tiered ticket.** Assign the lightest defensible\n   `verification_tier` (rule 2), attach the step-3 evidence, and state the acceptance criterion as a\n   concrete, verifiable delta (\"these two blocks become one call site\", \"this AC is now covered by a\n   spec\", \"this doc line now matches the code and a doc-lock sensor guards it\").\n7. **Verify the improvement by re-running the baseline.** Once a ticket lands, re-run the exact\n   step-2 baseline. The delta — fewer lines at the same behavior, a newly-covered AC, a green\n   doc-lock, a faster hot path — is the evidence the improvement is real and regression-free.\n\n## References\n\n- `references\u002Fimprovement-dimensions.md` — the catalog of constructive quality dimensions with\n  \"what good looks like\" and how to find each one by running the real code. Seed step 1 from here.\n- `references\u002Fuseless-vs-valuable.md` — the checklist for telling a self-improve pass that produced\n  durable value from one that shipped cosmetic churn. Run it in step 4 and again before reporting.\n\n## Priority scoring\n\nFindings and the tickets they produce reuse this repo's existing HIGH \u002F MEDIUM \u002F LOW convention (see\n`.claude\u002Fagents\u002Freviewer.md`) — the same scale the reviewer already runs on — rather than a separate\nnumeric rubric. A quality finding is rarely HIGH; most are MEDIUM (real complexity or a real\ncoverage gap) or LOW (a small clarity or consistency win). Be honest about severity: inflating a\ncosmetic tidy to MEDIUM is exactly the noise step 4 exists to remove.\n\n## Runs on Fable end to end\n\nUnlike its adversarial sibling, this skill is constructive throughout — no hostile-input authoring,\nno trust-boundary probing — so it runs comfortably with Fable 5 as the orchestrator and a `sonnet`\n`developer` subagent for the ticketed changes. The orchestrator names the component, picks\ndimensions, triages, and mints tickets; the `developer` subagent lands each change at its tier and\nre-runs the baseline. The only step that hands work elsewhere is step 5's security gate, which\nroutes to `hive-adversarial-improve` (where the dual-use-sensitive work belongs).\n\n## How it complements the adversarial skill\n\nA mature component gets both: a `hive-self-improve` pass keeps it simple, covered, and clear, and —\nif it sits on a trust boundary — a `hive-adversarial-improve` pass keeps it robust against hostile\ninput. Run self-improve first on a component that is merely messy; run adversarial-improve first on\none whose main risk is untrusted input. Findings from either can legitimately hand off to the other:\na self-improve pass that uncovers a validation seam routes it to adversarial-improve (step 5), and\nan adversarial pass that trips over duplicated gate logic can note it back for a self-improve ticket.\n","# Hive Self-Improve\n\n**Context guard — framework repo only.** This skill targets the hivemind **framework's own**\ncomponents (its `src\u002F`, `tests\u002F`, `tasks\u002F`) — not a downstream project built with hivemind. Before\nproceeding, confirm you are in the framework repo: call `isFrameworkRepo({ repoRoot })` from\n`src\u002Fframework-context.js` against the current working repo root. If it returns **false** (a\nconsumer project), **STOP** — do not proceed with this skill — and direct the user to the\n`hive-self-improve-current-project` variant instead, which runs the same protocol against the\nconsumer's own project code. If `src\u002Fframework-context.js` cannot be imported (e.g. a stripped\ncheckout), fall back to the equivalent check by hand: the repo root has a `.claude-plugin\u002Fplugin.json`\nwhose `name` field is `hivemind`, AND a `src\u002F` directory, AND a `bin\u002Finit.js` file. All three must\nhold for this skill to apply; if any is missing, STOP and route to the `-current-project` variant.\n\nThis skill turns an ad-hoc \"this could be cleaner \u002F better covered \u002F clearer\" conversation into a\nrepeatable, evidence-producing exercise. It is a **durable quality-improvement capability**, not a\none-off brainstorm: every run is grounded in the real shipped code actually executed, seeded from a\ncited improvement-dimension catalog, and converts every finding into a right-tiered ticket so the\nimprovement survives after the session ends.\n\nIt is the constructive counterpart to `hive-adversarial-improve`. That skill hardens a **trust\nboundary** against hostile inputs; this one raises the everyday quality of a component that is\nalready correct — the ~90% of \"make the framework better\" work that is not security-sensitive. The\ntwo are designed to run side by side: most components deserve a self-improve pass, and only the ones\nsitting on a trust boundary additionally deserve an adversarial one.\n\n## When to load this skill\n\nLoad whenever the task is to raise the quality of one specific, already-shipped component or\npipeline along a **constructive** dimension: it works, but it could be simpler, better covered,\nbetter documented, faster, or more ergonomic. Examples: collapsing duplicated logic across two\nmodules, closing a use-case coverage gap, reconciling a doc that has drifted from the code it\ndescribes, tightening a CLI's error messages.\n\nDo **not** load it for:\n\n- **Security or trust-boundary hardening** — where untrusted content crosses into trusted execution\n  (third-party skill adoption, tool-use dispatch, installers, parsers of fetched content). That is\n  `hive-adversarial-improve`. See the hard handoff in protocol step 5.\n- **Routine review of a single diff** — that is the `reviewer` agent.\n- **Open-ended research into an unfamiliar library or pattern** — that is the `researcher` agent.\n\n## The two load-bearing rules\n\nEverything below is scaffolding around these two rules. Drop either one and the exercise stops\nproducing durable value:\n\n1. **Every finding is grounded in the REAL code actually run — never speculation.** A finding must\n   point at concrete evidence you produced this session: a duplication with both `file:line`\n   locations, a coverage gap named against the specific acceptance criterion it leaves untested, a\n   doc-vs-code drift with both sides quoted, a timing number from an actual run. A quality review\n   that lists \"things that feel improvable\" without running the code or citing the artifact is a\n   brainstorm, not a self-improve pass — and it systematically ships cosmetic churn while missing\n   the real complexity.\n2. **Every surviving finding becomes a ticket at the LIGHTEST defensible verification tier, with\n   the evidence attached.** A finding that lives only in a chat transcript evaporates the moment the\n   session ends. Most self-improve findings are `tests-after` (behavior provable by running the\n   code, including schema\u002Fstate-mutation\u002Fparsing findings) or `uat-only` (docs, glue, ergonomics) —\n   see this repo's verification-tier rubric; the `tdd` tier it once described is retired (TASK-212,\n   2026-08-13 human decision). The concrete evidence from rule 1 is what makes the ticket actionable\n   rather than a vague \"clean this up.\"\n\n## The 7-step protocol\n\n1. **Name the component and pick the dimensions in scope.** Choose from\n   `references\u002Fimprovement-dimensions.md` — simplification\u002Freuse, test-coverage, doc\u002Fconsistency\n   drift, developer experience, performance, structural cohesion. Do not scan every dimension by\n   reflex; pick the ones that fit the component and say which you are deliberately skipping.\n2. **Establish a baseline by running the real thing.** Before proposing any change, capture the\n   current state as evidence: run the affected specs and record pass\u002Fskip counts, time the hot path\n   if performance is in scope, run the CLI and capture its actual output. This baseline is what\n   every later finding is measured against and what step 7 re-runs to prove the improvement landed.\n3. **Scan each in-scope dimension for concrete findings.** Work from the baseline, not from\n   imagination. Each finding carries its evidence inline (the duplicated block, the uncovered AC,\n   the drifted doc line). Grep and read the real code; do not infer structure you have not opened.\n4. **Triage against the useless-vs-valuable checklist.** Run every candidate through\n   `references\u002Fuseless-vs-valuable.md` and discard the cosmetic ones. Style-only reformatting,\n   speculative abstraction for an imagined future, and any \"improvement\" that would add a spec the\n   repo's new-test budget forbids are noise — dropping them is the point of this step.\n5. **Security handoff (hard gate).** If any finding touches a trust boundary — untrusted content\n   reaching trusted execution, an auth\u002Fvalidation seam, a scanner or gate that rejects hostile\n   input — STOP on that finding and route it to `hive-adversarial-improve` instead of handling it\n   here. This skill deliberately does not do adversarial security work; forcing a security finding\n   through the constructive path under-tests it. Record the handoff so the finding is not lost.\n6. **Convert each surviving finding into a right-tiered ticket.** Assign the lightest defensible\n   `verification_tier` (rule 2), attach the step-3 evidence, and state the acceptance criterion as a\n   concrete, verifiable delta (\"these two blocks become one call site\", \"this AC is now covered by a\n   spec\", \"this doc line now matches the code and a doc-lock sensor guards it\").\n7. **Verify the improvement by re-running the baseline.** Once a ticket lands, re-run the exact\n   step-2 baseline. The delta — fewer lines at the same behavior, a newly-covered AC, a green\n   doc-lock, a faster hot path — is the evidence the improvement is real and regression-free.\n\n## References\n\n- `references\u002Fimprovement-dimensions.md` — the catalog of constructive quality dimensions with\n  \"what good looks like\" and how to find each one by running the real code. Seed step 1 from here.\n- `references\u002Fuseless-vs-valuable.md` — the checklist for telling a self-improve pass that produced\n  durable value from one that shipped cosmetic churn. Run it in step 4 and again before reporting.\n\n## Priority scoring\n\nFindings and the tickets they produce reuse this repo's existing HIGH \u002F MEDIUM \u002F LOW convention (see\n`.claude\u002Fagents\u002Freviewer.md`) — the same scale the reviewer already runs on — rather than a separate\nnumeric rubric. A quality finding is rarely HIGH; most are MEDIUM (real complexity or a real\ncoverage gap) or LOW (a small clarity or consistency win). Be honest about severity: inflating a\ncosmetic tidy to MEDIUM is exactly the noise step 4 exists to remove.\n\n## Runs on Fable end to end\n\nUnlike its adversarial sibling, this skill is constructive throughout — no hostile-input authoring,\nno trust-boundary probing — so it runs comfortably with Fable 5 as the orchestrator and a `sonnet`\n`developer` subagent for the ticketed changes. The orchestrator names the component, picks\ndimensions, triages, and mints tickets; the `developer` subagent lands each change at its tier and\nre-runs the baseline. The only step that hands work elsewhere is step 5's security gate, which\nroutes to `hive-adversarial-improve` (where the dual-use-sensitive work belongs).\n\n## How it complements the adversarial skill\n\nA mature component gets both: a `hive-self-improve` pass keeps it simple, covered, and clear, and —\nif it sits on a trust boundary — a `hive-adversarial-improve` pass keeps it robust against hostile\ninput. Run self-improve first on a component that is merely messy; run adversarial-improve first on\none whose main risk is untrusted input. Findings from either can legitimately hand off to the other:\na self-improve pass that uncovers a validation seam routes it to adversarial-improve (step 5), and\nan adversarial pass that trips over duplicated gate logic can note it back for a self-improve ticket.\n",{"name":5,"description":7},"main","c8856147b7a3246a1e54b66126d093183805d1f3",".claude\u002Fskills\u002Fhive-self-improve",0,"https:\u002F\u002Fgithub.com\u002FHexingBot\u002Fhivemind","https:\u002F\u002Fhivemind-ef84a.web.app","https:\u002F\u002Fgithub.com\u002FHexingBot","2026-10-10 00:50:01",{"source":8,"skillName":5,"commands":32,"prompt":42,"zipAvailable":43},[33,36,39],{"tool":34,"command":35},"npx","npx skills add HexingBot\u002Fhivemind --skill hive-self-improve",{"tool":37,"command":38},"bunx","bunx skills add HexingBot\u002Fhivemind --skill hive-self-improve",{"tool":40,"command":41},"pnpm","pnpm dlx skills add HexingBot\u002Fhivemind --skill hive-self-improve","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FHexingBot\u002Fhivemind\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fhive-self-improve\u002FSKILL.md",true,[45,51],{"path":46,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":49,"sourceHtmlUrl":50},"references\u002Fimprovement-dimensions.md","REFERENCE","参考文档",4008,"https:\u002F\u002Fgithub.com\u002FHexingBot\u002Fhivemind\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fhive-self-improve\u002Freferences\u002Fimprovement-dimensions.md",{"path":52,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":53,"sourceHtmlUrl":54},"references\u002Fuseless-vs-valuable.md",3376,"https:\u002F\u002Fgithub.com\u002FHexingBot\u002Fhivemind\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fhive-self-improve\u002Freferences\u002Fuseless-vs-valuable.md",[56,63,72,80,87,95],{"skillNo":57,"name":58,"nameZh":6,"description":59,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":60,"categories":61,"keywords":62,"publishedAt":18},"01M77MWBWY2AK0GJ91MMKAPZ0M","unknown-skill","A demo skill with no license signal at all, used to exercise the \"unknown\" classification branch of the human license gate (TASK-135).","https:\u002F\u002Fgithub.com\u002FHexingBot\u002Fhivemind\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fskills\u002Funknown-skill\u002FSKILL.md",[],[],{"skillNo":64,"name":65,"nameZh":6,"description":66,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":67,"licenseFileUrl":68,"sourceHtmlUrl":69,"categories":70,"keywords":71,"publishedAt":18},"01M77MW9H43Z0ATZ15421K8J7D","spoofed-license-skill","A demo skill used to test TASK-150's SPDX-header-vs-LICENSE-file conflict surfacing.","GPL-3.0","https:\u002F\u002Fgithub.com\u002FHexingBot\u002Fhivemind\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fskills\u002Fspoofed-license-skill\u002FLICENSE","https:\u002F\u002Fgithub.com\u002FHexingBot\u002Fhivemind\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fskills\u002Fspoofed-license-skill\u002FSKILL.md",[],[],{"skillNo":73,"name":74,"nameZh":6,"description":75,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":76,"sourceHtmlUrl":77,"categories":78,"keywords":79,"publishedAt":18},"01M77MW7715ENRZN6F01FDM81J","permissive-skill","A demo skill used to test assimilation of a permissively-licensed third party skill.","https:\u002F\u002Fgithub.com\u002FHexingBot\u002Fhivemind\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fskills\u002Fpermissive-skill\u002FLICENSE","https:\u002F\u002Fgithub.com\u002FHexingBot\u002Fhivemind\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fskills\u002Fpermissive-skill\u002FSKILL.md",[],[],{"skillNo":81,"name":82,"nameZh":6,"description":83,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":84,"categories":85,"keywords":86,"publishedAt":18},"01M77MW3CRAPWETDAZT7E08BVS","malicious-skill","A fixture skill deliberately carrying risky content patterns, used only to exercise src\u002Fskill-scan.js (TASK-122). Never assimilated for real.","https:\u002F\u002Fgithub.com\u002FHexingBot\u002Fhivemind\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fskills\u002Fmalicious-skill\u002FSKILL.md",[],[],{"skillNo":88,"name":89,"nameZh":6,"description":90,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":67,"licenseFileUrl":91,"sourceHtmlUrl":92,"categories":93,"keywords":94,"publishedAt":18},"01M77MW1RSWW1MTQGM9TY3X37G","copyleft-skill","A demo skill used to test the human license gate for a copyleft third party skill.","https:\u002F\u002Fgithub.com\u002FHexingBot\u002Fhivemind\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fskills\u002Fcopyleft-skill\u002FLICENSE","https:\u002F\u002Fgithub.com\u002FHexingBot\u002Fhivemind\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fskills\u002Fcopyleft-skill\u002FSKILL.md",[],[],{"skillNo":96,"name":97,"nameZh":6,"description":98,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":99,"categories":100,"keywords":101,"publishedAt":18},"01M77MVWT1EVEJ8ZYZTCG4CENE","watch","Watch a video (URL or local path). Downloads with yt-dlp, extracts auto-scaled frames with ffmpeg, pulls the transcript from captions (or Whisper API fallback), and hands the result to Claude so it can answer questions about what's in the video. (assimilated for Hivemind)","https:\u002F\u002Fgithub.com\u002FHexingBot\u002Fhivemind\u002Fblob\u002Fmain\u002Fskills\u002Fwatch\u002FSKILL.md",[],[]]