[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M77KZ6EXD3FQCV9WVBFS1END":3,"skill-same:01M77KZ6EXD3FQCV9WVBFS1END":48},{"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":26,"commitSha":27,"skillDirectory":28,"forksCount":29,"repositoryHtmlUrl":30,"repositoryHomepageUrl":6,"ownerHtmlUrl":31,"ownerBlogUrl":32,"ownerDescription":33,"updatedAt":34,"install":35,"resources":47},"01M77KZ6EXD3FQCV9WVBFS1END","setup",null,"Verify and configure the improvement plugin for this repository. check (default) inspects the .claude\u002Fimprovement.md config cascade read-only across all three layers, user-global, team, personal override, and reports the effective evidence-source configuration: which layers exist, what churn window and exclusion patterns resolve, and which Tier 2 MCP telemetry sources are declared. apply interviews the user and writes the team file. Zero config is a valid state. Tier 0 evidence needs nothing configured. Use when: 'set up improvement', 'configure improvement', 'is improvement configured', 'improvement setup', 'declare a telemetry source for the improvement finder', 'tune churn exclusions', or \u002Fimprovement:find reports Tier 2 as an evidence gap to close. Re-runnable. Safe to invoke again to reconfigure.","melodic-software\u002Fclaude-code-plugins","melodic-software","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F58273638?v=4","01M77JVQTXNGKCE4MDFY2BC936","Melodic Software","01M77JVQVN9QDWKAH6W03ANP76",30,"MIT","https:\u002F\u002Fgithub.com\u002Fmelodic-software\u002Fclaude-code-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Fimprovement\u002Fskills\u002Fsetup\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\ndescription: \"Verify and configure the improvement plugin for this repository. check (default) inspects the .claude\u002Fimprovement.md config cascade read-only across all three layers, user-global, team, personal override, and reports the effective evidence-source configuration: which layers exist, what churn window and exclusion patterns resolve, and which Tier 2 MCP telemetry sources are declared. apply interviews the user and writes the team file. Zero config is a valid state. Tier 0 evidence needs nothing configured. Use when: 'set up improvement', 'configure improvement', 'is improvement configured', 'improvement setup', 'declare a telemetry source for the improvement finder', 'tune churn exclusions', or \u002Fimprovement:find reports Tier 2 as an evidence gap to close. Re-runnable. Safe to invoke again to reconfigure.\"\nargument-hint: \"[check|apply]\"\nuser-invocable: true\ndisable-model-invocation: true\n---\n\n## Purpose\n\nVerify and manage the consuming repo's tracked evidence-source config at\n`.claude\u002Fimprovement.md`, so `\u002Fimprovement:find` resolves Tier 2 telemetry sources and churn\ntuning from a declared contract instead of having nothing to consult.\n\nThe config is optional in every layer: with none, the finder runs fully on Tier 0 repo-native\nevidence with the bundled churn defaults, and Tier 2 is a recorded evidence gap, so total\nabsence is a reported INFO, never a FAIL. The key contract, layer paths, and merge semantics\nlive in `${CLAUDE_PLUGIN_ROOT}\u002Freference\u002Fconfig.md` (the single home. This skill implements\nthat contract and does not restate it).\n\n## Actions\n\nTwo actions only. No argument or `check` runs the read-only check; `apply` runs the check first,\nthen the interview-and-write flow. Any other argument is an error naming these two actions.\n\n## The config cascade\n\nThree layers, resolved user-global (`~\u002F.claude\u002Fimprovement.md`) → team\n(`.claude\u002Fimprovement.md`, tracked) → personal override (`.claude\u002Fimprovement.local.md`,\ngitignored), merged per the semantics declared in\n`${CLAUDE_PLUGIN_ROOT}\u002Freference\u002Fconfig.md`: per-key override for scalars, union for\n`churn_exclude`, per-source-name merge for `evidence_sources` with an empty-body opt-out. This\nskill writes only the **team** file; changes to the user-global layer or the personal override are the\nuser's to make. Anchor every repo-relative read at the repo root (`${CLAUDE_PROJECT_DIR}` when\nset, otherwise `git rev-parse --show-toplevel`), never at the CWD.\n\n## `check` (read-only)\n\nInspect the effective merged config and report a PASS\u002FFAIL\u002FINFO table with one remediation line\nper FAIL. Modify nothing, and do NOT run a scan. That is `\u002Fimprovement:find`.\n\n1. **Layer presence.** Load every layer you can access and report which exist. Look up the user-global layer with one Read of the literal path `~\u002F.claude\u002Fimprovement.md`, `~` kept; never Glob for it, spell out a home directory, or search the home tree. A Read that finds no file means the layer is absent. All absent →\n   INFO: the finder uses bundled defaults and records Tier 2 as an evidence gap; `apply` writes\n   `.claude\u002Fimprovement.md` to declare sources. When a layer outside the repo (the user-global\n   base) cannot be read, WARN it was not considered rather than presenting the readable layers\n   as the whole effective config.\n2. **Effective configuration.** Report the merged result and which layer supplied each value:\n   the effective `churn_window`, whether bundled exclusion defaults are on, declared\n   `churn_exclude` additions, and every `evidence_sources` entry (name, `mcp_server`, `kind`,\n   scope), including entries a later layer opted out, reported as removed, not broken. A\n   malformed layer degrades soft: name it, resolve as if absent, report the parse error.\n3. **Key validity.** Each `evidence_sources` entry names an `mcp_server`; an entry missing it is\n   FAIL, naming the entry. Where the session can enumerate available MCP servers, note (INFO)\n   any declared server not currently available. At find-time that source becomes an\n   evidence-gap line, which may be expected on this machine.\n4. **Tracked, not ignored.** A present team file must be committed to be team-shared, and\n   \"not ignored\" is not \"tracked\". Probe both: `git check-ignore -v .claude\u002Fimprovement.md`\n   (non-empty result is FAIL with the matching pattern) AND\n   `git ls-files --error-unmatch .claude\u002Fimprovement.md` (non-zero exit is FAIL: the file is\n   present but untracked, the state immediately after `apply` writes it, so teammates will not\n   receive it until it is committed; report \"untracked, commit it\" and, if it is staged but\n   uncommitted, say that instead via `git diff --cached --name-only`). The `.local.md` personal override is\n   expected to be gitignored, a personal override that is tracked or staged is FAIL (a personal deviation\n   can reach team history); an ignored personal override is INFO.\n5. **Other-layer divergence.** INFO: when the personal override or user-global layer changes the team file's\n   effect (a different window, extra exclusions, an added or opted-out source), say so\n   explicitly, since `apply` writes only the team file.\n\n## `apply` (idempotent)\n\nRun `check`, then interview and write the team file. Proceed non-interactively where the\ninvocation and the repo make values unambiguous; ask only where a value genuinely needs the\nuser.\n\n1. **Read the effective config first, across all layers**, and present the merged baseline with\n   per-layer provenance. Exactly as `check` step 2 reports it. When a higher layer's opt-out or\n   override shapes the effective result, warn that editing the team file alone will not change\n   what this machine resolves, and point at the owning layer. Nothing is dropped without the\n   user confirming.\n2. **Draft defaults from the repo.** Before asking: propose `churn_exclude` additions from what\n   actually exists (generated\u002Fbuild output trees, snapshot or designer files not covered by the\n   bundled defaults, the `excludes` ERE in `${CLAUDE_PLUGIN_ROOT}\u002Fskills\u002Ffind\u002Fcontext\u002Fhotspots.md`: one Read of that file, never Glob or Grep the plugin directory); keep `churn_window` at the bundled default unless the repo's history is\n   young or the user says otherwise; and draft `evidence_sources` candidates from the MCP\n   servers the consumer has configured that plausibly carry telemetry, a candidate is a\n   proposal for the user to confirm, never an assumption. No plausible source and none volunteered\n   is a fine outcome: write the churn keys only, and say Tier 2 stays a declared gap.\n3. **Interview, one decision at a time**: window, exclusions, then each telemetry source (name,\n   `mcp_server`, `kind`, scope, hints) with a recommendation the user accepts, edits, or\n   declines.\n4. **Write the team file** in the file format defined by\n   `${CLAUDE_PLUGIN_ROOT}\u002Freference\u002Fconfig.md`, a fenced YAML block holding only the keys the\n   user confirmed. Never write a vendor-specific default the user did not declare.\n5. **Verify after writing.** Re-run the `check` probes on the written file. BOTH\n   `git check-ignore -v .claude\u002Fimprovement.md` (surface a matching pattern and offer a\n   `.gitignore` fix) AND `git ls-files --error-unmatch .claude\u002Fimprovement.md` (non-zero exit\n   means the file is un-ignored but still untracked, the guaranteed state right after this\n   step writes it, so report \"written but untracked: commit it to share with the team\", never\n   success). Then offer the personal override\n   convention: personal overrides go in `.claude\u002Fimprovement.local.md`; recommend the consumer\n   add the recursive `.claude\u002F**\u002F*.local.*` line to `.gitignore` if not already covered. But\n   never edit their `.gitignore` yourself.\n\nRe-running `apply` after everything passes changes nothing and reports \"already configured\".\n\n## Output\n\n`check`: the PASS\u002FFAIL\u002FINFO table plus the effective evidence-source configuration with\nper-layer provenance. `apply`: a tracked `.claude\u002Fimprovement.md`, the re-run check result, and\none paragraph on what was written and how to re-run this setup to reconfigure.\n\n## What this skill does NOT do\n\n- Run an improvement scan. That is `\u002Fimprovement:find`; `check` only inspects config.\n- Install or configure MCP servers. It declares which configured server carries telemetry;\n  provisioning the server is the consumer's own MCP setup.\n- Write the user-global layer, the `.local.md` personal override, or the consumer's `.gitignore`. Team\n  file only; everything else is recommended, not written.\n- Write machine-local state. Configuration lives in the consumer's tracked file, never in the\n  plugin directory or the plugin data directory.\n\n## Gotchas\n\n- **All layers absent is healthy.** Tier 0 evidence needs no config; do not manufacture a FAIL\n  (or a config file) for a repo that has not opted into Tier 2 or churn tuning.\n- **CWD is not the repo root.** Invoked from a subdirectory, a CWD-relative read finds a\n  nonexistent `\u003Csubdir>\u002F.claude\u002Fimprovement.md` and silently reports an absent config. Anchor\n  at the root in every self-contained shell call.\n- **The team file is the only thing `apply` writes.** An effective value contributed by the\n  personal override or user-global layer cannot be changed here; say which layer owns it instead of\n  writing a team value that the higher layer will keep overriding.\n- **No git verdict on the user-global layer.** `git check-ignore` against a path outside the\n  worktree is meaningless (or confidently wrong when the home directory is itself a repo); the tracked\u002Fignored probes apply to the two in-repo layers only.\n","## Purpose\n\nVerify and manage the consuming repo's tracked evidence-source config at\n`.claude\u002Fimprovement.md`, so `\u002Fimprovement:find` resolves Tier 2 telemetry sources and churn\ntuning from a declared contract instead of having nothing to consult.\n\nThe config is optional in every layer: with none, the finder runs fully on Tier 0 repo-native\nevidence with the bundled churn defaults, and Tier 2 is a recorded evidence gap, so total\nabsence is a reported INFO, never a FAIL. The key contract, layer paths, and merge semantics\nlive in `${CLAUDE_PLUGIN_ROOT}\u002Freference\u002Fconfig.md` (the single home. This skill implements\nthat contract and does not restate it).\n\n## Actions\n\nTwo actions only. No argument or `check` runs the read-only check; `apply` runs the check first,\nthen the interview-and-write flow. Any other argument is an error naming these two actions.\n\n## The config cascade\n\nThree layers, resolved user-global (`~\u002F.claude\u002Fimprovement.md`) → team\n(`.claude\u002Fimprovement.md`, tracked) → personal override (`.claude\u002Fimprovement.local.md`,\ngitignored), merged per the semantics declared in\n`${CLAUDE_PLUGIN_ROOT}\u002Freference\u002Fconfig.md`: per-key override for scalars, union for\n`churn_exclude`, per-source-name merge for `evidence_sources` with an empty-body opt-out. This\nskill writes only the **team** file; changes to the user-global layer or the personal override are the\nuser's to make. Anchor every repo-relative read at the repo root (`${CLAUDE_PROJECT_DIR}` when\nset, otherwise `git rev-parse --show-toplevel`), never at the CWD.\n\n## `check` (read-only)\n\nInspect the effective merged config and report a PASS\u002FFAIL\u002FINFO table with one remediation line\nper FAIL. Modify nothing, and do NOT run a scan. That is `\u002Fimprovement:find`.\n\n1. **Layer presence.** Load every layer you can access and report which exist. Look up the user-global layer with one Read of the literal path `~\u002F.claude\u002Fimprovement.md`, `~` kept; never Glob for it, spell out a home directory, or search the home tree. A Read that finds no file means the layer is absent. All absent →\n   INFO: the finder uses bundled defaults and records Tier 2 as an evidence gap; `apply` writes\n   `.claude\u002Fimprovement.md` to declare sources. When a layer outside the repo (the user-global\n   base) cannot be read, WARN it was not considered rather than presenting the readable layers\n   as the whole effective config.\n2. **Effective configuration.** Report the merged result and which layer supplied each value:\n   the effective `churn_window`, whether bundled exclusion defaults are on, declared\n   `churn_exclude` additions, and every `evidence_sources` entry (name, `mcp_server`, `kind`,\n   scope), including entries a later layer opted out, reported as removed, not broken. A\n   malformed layer degrades soft: name it, resolve as if absent, report the parse error.\n3. **Key validity.** Each `evidence_sources` entry names an `mcp_server`; an entry missing it is\n   FAIL, naming the entry. Where the session can enumerate available MCP servers, note (INFO)\n   any declared server not currently available. At find-time that source becomes an\n   evidence-gap line, which may be expected on this machine.\n4. **Tracked, not ignored.** A present team file must be committed to be team-shared, and\n   \"not ignored\" is not \"tracked\". Probe both: `git check-ignore -v .claude\u002Fimprovement.md`\n   (non-empty result is FAIL with the matching pattern) AND\n   `git ls-files --error-unmatch .claude\u002Fimprovement.md` (non-zero exit is FAIL: the file is\n   present but untracked, the state immediately after `apply` writes it, so teammates will not\n   receive it until it is committed; report \"untracked, commit it\" and, if it is staged but\n   uncommitted, say that instead via `git diff --cached --name-only`). The `.local.md` personal override is\n   expected to be gitignored, a personal override that is tracked or staged is FAIL (a personal deviation\n   can reach team history); an ignored personal override is INFO.\n5. **Other-layer divergence.** INFO: when the personal override or user-global layer changes the team file's\n   effect (a different window, extra exclusions, an added or opted-out source), say so\n   explicitly, since `apply` writes only the team file.\n\n## `apply` (idempotent)\n\nRun `check`, then interview and write the team file. Proceed non-interactively where the\ninvocation and the repo make values unambiguous; ask only where a value genuinely needs the\nuser.\n\n1. **Read the effective config first, across all layers**, and present the merged baseline with\n   per-layer provenance. Exactly as `check` step 2 reports it. When a higher layer's opt-out or\n   override shapes the effective result, warn that editing the team file alone will not change\n   what this machine resolves, and point at the owning layer. Nothing is dropped without the\n   user confirming.\n2. **Draft defaults from the repo.** Before asking: propose `churn_exclude` additions from what\n   actually exists (generated\u002Fbuild output trees, snapshot or designer files not covered by the\n   bundled defaults, the `excludes` ERE in `${CLAUDE_PLUGIN_ROOT}\u002Fskills\u002Ffind\u002Fcontext\u002Fhotspots.md`: one Read of that file, never Glob or Grep the plugin directory); keep `churn_window` at the bundled default unless the repo's history is\n   young or the user says otherwise; and draft `evidence_sources` candidates from the MCP\n   servers the consumer has configured that plausibly carry telemetry, a candidate is a\n   proposal for the user to confirm, never an assumption. No plausible source and none volunteered\n   is a fine outcome: write the churn keys only, and say Tier 2 stays a declared gap.\n3. **Interview, one decision at a time**: window, exclusions, then each telemetry source (name,\n   `mcp_server`, `kind`, scope, hints) with a recommendation the user accepts, edits, or\n   declines.\n4. **Write the team file** in the file format defined by\n   `${CLAUDE_PLUGIN_ROOT}\u002Freference\u002Fconfig.md`, a fenced YAML block holding only the keys the\n   user confirmed. Never write a vendor-specific default the user did not declare.\n5. **Verify after writing.** Re-run the `check` probes on the written file. BOTH\n   `git check-ignore -v .claude\u002Fimprovement.md` (surface a matching pattern and offer a\n   `.gitignore` fix) AND `git ls-files --error-unmatch .claude\u002Fimprovement.md` (non-zero exit\n   means the file is un-ignored but still untracked, the guaranteed state right after this\n   step writes it, so report \"written but untracked: commit it to share with the team\", never\n   success). Then offer the personal override\n   convention: personal overrides go in `.claude\u002Fimprovement.local.md`; recommend the consumer\n   add the recursive `.claude\u002F**\u002F*.local.*` line to `.gitignore` if not already covered. But\n   never edit their `.gitignore` yourself.\n\nRe-running `apply` after everything passes changes nothing and reports \"already configured\".\n\n## Output\n\n`check`: the PASS\u002FFAIL\u002FINFO table plus the effective evidence-source configuration with\nper-layer provenance. `apply`: a tracked `.claude\u002Fimprovement.md`, the re-run check result, and\none paragraph on what was written and how to re-run this setup to reconfigure.\n\n## What this skill does NOT do\n\n- Run an improvement scan. That is `\u002Fimprovement:find`; `check` only inspects config.\n- Install or configure MCP servers. It declares which configured server carries telemetry;\n  provisioning the server is the consumer's own MCP setup.\n- Write the user-global layer, the `.local.md` personal override, or the consumer's `.gitignore`. Team\n  file only; everything else is recommended, not written.\n- Write machine-local state. Configuration lives in the consumer's tracked file, never in the\n  plugin directory or the plugin data directory.\n\n## Gotchas\n\n- **All layers absent is healthy.** Tier 0 evidence needs no config; do not manufacture a FAIL\n  (or a config file) for a repo that has not opted into Tier 2 or churn tuning.\n- **CWD is not the repo root.** Invoked from a subdirectory, a CWD-relative read finds a\n  nonexistent `\u003Csubdir>\u002F.claude\u002Fimprovement.md` and silently reports an absent config. Anchor\n  at the root in every self-contained shell call.\n- **The team file is the only thing `apply` writes.** An effective value contributed by the\n  personal override or user-global layer cannot be changed here; say which layer owns it instead of\n  writing a team value that the higher layer will keep overriding.\n- **No git verdict on the user-global layer.** `git check-ignore` against a path outside the\n  worktree is meaningless (or confidently wrong when the home directory is itself a repo); the tracked\u002Fignored probes apply to the two in-repo layers only.\n",{"description":7,"argument-hint":24,"user-invocable":25,"disable-model-invocation":25},"[check|apply]",true,"main","d0e68c3f36b92a1d6381f0ba0060d7a0b8b886c5","plugins\u002Fimprovement\u002Fskills\u002Fsetup",7,"https:\u002F\u002Fgithub.com\u002Fmelodic-software\u002Fclaude-code-plugins","https:\u002F\u002Fgithub.com\u002Fmelodic-software","https:\u002F\u002Fmelodicsoftware.com","Independent software studio building developer tools, automation, and application services, with roots in music.","2026-11-15 05:15:06",{"source":8,"skillName":5,"commands":36,"prompt":46,"zipAvailable":25},[37,40,43],{"tool":38,"command":39},"npx","npx skills add melodic-software\u002Fclaude-code-plugins --skill setup",{"tool":41,"command":42},"bunx","bunx skills add melodic-software\u002Fclaude-code-plugins --skill setup",{"tool":44,"command":45},"pnpm","pnpm dlx skills add melodic-software\u002Fclaude-code-plugins --skill setup","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fmelodic-software\u002Fclaude-code-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Fimprovement\u002Fskills\u002Fsetup\u002FSKILL.md",[],[49,56,63,70,75,82],{"skillNo":50,"name":51,"nameZh":6,"description":52,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":53,"categories":54,"keywords":55,"publishedAt":19},"01M7DYYDG31B9YX7N9MB9QW2GW","track","Backlog reads through the bound tracker: search open and closed items, view one item with its comments and state history. Use when another skill searches or reads tracker items.","https:\u002F\u002Fgithub.com\u002Fmelodic-software\u002Fclaude-code-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Fsession-flow\u002Fevals\u002Frecall-still-reports-reopened-bug\u002Fwork-items-stub\u002Fskills\u002Ftrack\u002FSKILL.md",[],[],{"skillNo":57,"name":58,"nameZh":6,"description":59,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":60,"categories":61,"keywords":62,"publishedAt":19},"01M7DYXVBAVG3NV30PA7GKW9Q6","reviewer-media","Reviewer media for one pull request: decides whether it needs a narrated walkthrough, builds the storyboard from \u002Freview:explain-change's digest, has \u002Fexplainer-video:produce render it and hands the manifest to \u002Fsource-control:pull-request media. Use when asked for 'reviewer media', 'a walkthrough video for this PR', 'a produced video of this change', or when pull-request ready or monitor hands a PR over.","https:\u002F\u002Fgithub.com\u002Fmelodic-software\u002Fclaude-code-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Freview\u002Fskills\u002Freviewer-media\u002FSKILL.md",[],[],{"skillNo":64,"name":65,"nameZh":6,"description":66,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":67,"categories":68,"keywords":69,"publishedAt":19},"01M7DYRZ5YD44AMYH26YR25GZ8","route","Model and effort per agent role (orchestrator, worker, verifier, retrieval) for a multi-agent run, resolved through the bundled defaults and the user, team and personal override config layers into fenced JSON with single-agent and fan-out variants and the layer behind each value. Use when asked 'which model for the subagents', 'route the roles' or 'model routing', or before a skill passes a role's model on a spawn.","https:\u002F\u002Fgithub.com\u002Fmelodic-software\u002Fclaude-code-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Fdiscovery\u002Fevals\u002Fex-model-pin-without-override\u002Fmulti-agent-stub\u002Fskills\u002Froute\u002FSKILL.md",[],[],{"skillNo":71,"name":65,"nameZh":6,"description":66,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":72,"categories":73,"keywords":74,"publishedAt":19},"01M7DYRTVT4FWTGDJY0HCCFJMZ","https:\u002F\u002Fgithub.com\u002Fmelodic-software\u002Fclaude-code-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Fdiscovery\u002Fevals\u002Fex-model-override-from-route\u002Fmulti-agent-stub\u002Fskills\u002Froute\u002FSKILL.md",[],[],{"skillNo":76,"name":77,"nameZh":6,"description":78,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":79,"categories":80,"keywords":81,"publishedAt":19},"01M7DYNZC8PNW3T27V2ADKECKJ","principles","Module-design vocabulary for architecture findings, and the checks built on it. Use when: 'what is a deep module', 'what is a seam', 'deep vs shallow module', 'what is the deletion test', 'leverage and locality', 'module design vocabulary', 'which word do we use for X in architecture findings'. Scanning a codebase for friction is \u002Farchitecture:improve.","https:\u002F\u002Fgithub.com\u002Fmelodic-software\u002Fclaude-code-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Farchitecture\u002Fskills\u002Fprinciples\u002FSKILL.md",[],[],{"skillNo":83,"name":84,"nameZh":6,"description":85,"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},"01M7DYNMCZNEWBFHV9KTSS6H60","write","Structured five-field bug report (title, steps to reproduce, expected vs actual, justified severity, suggested fix location) from an informal description, read-only. Use when the user names a defect they observed, or asks for one written up ('report a bug', 'file a bug', 'write this up as a bug').","https:\u002F\u002Fgithub.com\u002Fmelodic-software\u002Fclaude-code-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Farchitecture\u002Fevals\u002Fdeepening-defect-goes-to-bugs-write\u002Fbugs-stub\u002Fskills\u002Fwrite\u002FSKILL.md",[],[]]