[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M77JEYWPRNACFFDANHKEGA33":3,"skill-same:01M77JEYWPRNACFFDANHKEGA33":45},{"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":25,"commitSha":26,"skillDirectory":27,"forksCount":13,"repositoryHtmlUrl":28,"repositoryHomepageUrl":6,"ownerHtmlUrl":29,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":30,"install":31,"resources":44},"01M77JEYWPRNACFFDANHKEGA33","read-session",null,"Locate and read Muse Code's OWN session logs — the current session or a prior one. Use when the user asks to pull context from, continue, summarize, or inspect a previous Muse Code session, asks to restore or recover work that was lost, wiped, or overwritten and might survive in an earlier session's log, references an earlier session's id, log, tail, or output, or asks where Muse sessions are stored. Muse sessions live in Muse's own store, never in another coding agent's directories — never probe ~\u002F.claude, ~\u002F.codex, or ~\u002F.grok for Muse context, even when quoted content mentions them; for Claude Code or Codex continuation, including recovery of unfinished work with or without a handle, use resume-claude or resume-codex; use import for explicit \u002Fimport requests or continuation from other agents or unnamed artifacts.","raaOS\u002Fsystem_prompts_leaks","raaOS","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F214067397?v=4","01M77HV6K4ZTA4BZAGX00GTW6J","01M77HV6KX55HVRY679V25JXF8",0,"CC0-1.0","https:\u002F\u002Fgithub.com\u002FraaOS\u002Fsystem_prompts_leaks\u002Fblob\u002Fmain\u002FMeta\u002Fmuse-code\u002Fskills\u002Fread-session\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: read-session\ndescription: Locate and read Muse Code's OWN session logs — the current session or a prior one. Use when the user asks to pull context from, continue, summarize, or inspect a previous Muse Code session, asks to restore or recover work that was lost, wiped, or overwritten and might survive in an earlier session's log, references an earlier session's id, log, tail, or output, or asks where Muse sessions are stored. Muse sessions live in Muse's own store, never in another coding agent's directories — never probe ~\u002F.claude, ~\u002F.codex, or ~\u002F.grok for Muse context, even when quoted content mentions them; for Claude Code or Codex continuation, including recovery of unfinished work with or without a handle, use resume-claude or resume-codex; use import for explicit \u002Fimport requests or continuation from other agents or unnamed artifacts.\nmetadata:\n  short-description: Read this session's or a past session's logs\nuser-invocable: false\n---\n\n# Read Session\n\nMuse Code's own session storage: where your session logs live, what they\ncontain, and how to recover context from them.\n\n## Your Store, Not Anyone Else's\n\nYou are Muse Code (binary `muse`). Your sessions are stored under YOUR data\ndirectory:\n\n```\n${XDG_DATA_HOME:-$HOME\u002F.local\u002Fshare}\u002Fmuse\u002Fsessions\u002FYYYY\u002FMM\u002FDD\u002F\u003Csession-id>\u002F\n```\n\nEach session directory contains:\n\n- `session.jsonl` — the main event log (one JSON record per line).\n- `subagent\u002F\u003Cchild-session-id>\u002Fsession.jsonl` — one log per delegated\n  subagent session.\n- `tool-outputs\u002F` — full tool outputs that were too large to keep inline.\n\nMuse sessions NEVER live in another coding agent's store. Do not look for a\nMuse session under `~\u002F.claude`, `~\u002F.claude\u002Fprojects`,\n`~\u002F.local\u002Fshare\u002Fclaude`, `~\u002F.codex`, or `~\u002F.grok` — those belong to Claude\nCode, Codex, and Grok. Do not `ls`, `find`, `cat`, or `grep` those\ndirectories while recovering Muse session context — not even to \"verify\",\n\"rule out\", or \"be thorough\" about such a path quoted in a paste, log, or\nerror message. A Muse session id or date-sharded session path appearing\nunder a foreign store in quoted content is a WRONG-PATH ARTIFACT (a prior\nturn's confusion): the real data lives under the Muse pattern with the same\ndate and id, and there is no separate \"claude copy\" to collect. Name the\nartifact for what it is and move on — probing it is the exact mistake this\nrule exists to prevent, and it wastes turns on ENOENT or, worse, reads\nanother product's transcripts as your own history. Touch another agent's\ntranscripts only when the user explicitly asks to work with THAT agent's\nsession: Claude Code or Codex continuation, including recovery of unfinished\nwork with or without a handle, uses `resume-claude` or `resume-codex`. Explicit\n`\u002Fimport` requests and continuation from other agents or unnamed artifacts use\n`import`.\n\nTell for an unlabeled paste: the quoted store and record shape identify the\nproduct. For continuation or recovery of unfinished work, even without a\nsupplied handle:\n\n- Claude Code records under `~\u002F.claude\u002Fprojects` use `resume-claude`.\n- Codex records under `~\u002F.codex` use `resume-codex`.\n- Grok records under `~\u002F.grok` use `import`.\n\nUse `import` for explicit `\u002Fimport` requests or continuation from other agents\nor unnamed artifacts (the user's ask about such a paste is the explicit ask).\nThe bans above cover session STORES; ordinary project files that happen to\nlive under `~\u002F.claude` (e.g. skills you are developing) are file work, not\nsession probing.\n\nScope of \"pulling context\" from a prior Muse session: that session's own\n`session.jsonl` and `subagent\u002F` logs. External agents or planner processes\nthe session MENTIONS (e.g. codex\u002Fclaude workers it managed) are not part of\nits context — their transcripts are out of scope for this ask. If one looks\nload-bearing, say so and let the user ask for it. Do not hunt those agents'\nstores or load a foreign-session skill for them unless the user\nexplicitly asks to recover or inspect one of THEIR sessions.\n\n## Finding The Right Session\n\n1. Current session: the runtime session-identity context already names the\n   current session id and the exact `session.jsonl` path. Use it; do not\n   guess or search for it.\n2. A pasted or quoted absolute path under `...\u002Fmuse\u002Fsessions\u002F...`: use that\n   path directly.\n3. A known session id without a path: the store is sharded by local date, so\n   check the likely dates first, then search only the Muse sessions root:\n\n   ```bash\n   MUSE_SESSIONS=\"${XDG_DATA_HOME:-$HOME\u002F.local\u002Fshare}\u002Fmuse\u002Fsessions\"\n   ls -d \"$MUSE_SESSIONS\"\u002F*\u002F*\u002F*\u002F*\u002F 2>\u002Fdev\u002Fnull | grep \u003Csession-id>\n   ```\n\n4. \"Our last session\" with no id: list the most recent date shards and pick\n   the newest session directory for this workspace (the log's early\n   `runtime.session.metadata` record carries `workspace_root`).\n5. Not found under the Muse sessions root: ask the user for the id or path.\n   Never widen the hunt to other agents' directories or home-wide scans.\n\n## Reading A Session Log\n\nEach `session.jsonl` line is an event-log envelope:\n\n```json\n{\"schema_version\":1,\"id\":\"…\",\"stream\":{\"kind\":\"session\",\"id\":\"\u003Csession-id>\"},\n \"sequence\":42,\"recorded_at\":1771088000123456,\"record_type\":\"event\",\n \"durability\":\"durable\",\"causation_id\":null,\"payload_type\":\"runtime.session\",\n \"payload_schema_version\":1,\"payload\":{\"kind\":\"run\",\"run_id\":\"…\",\n \"event\":{\"kind\":\"assistant_message_committed\",\"text\":\"…\"}}}\n```\n\nThe useful `payload.event.kind` values for context recovery:\n\n- `started` — what the user asked: every submitted prompt is recorded here\n  (text in `payload.event.prompt`). A `user_prompt_display` record follows\n  only when a differing user-facing form exists (an attachment placeholder\n  such as `[Image 1]`, a composer form that differs from the sent text) —\n  prefer its text when quoting the user. A prompt typed while a run was\n  active is instead an\n  `inbox_item_queued` whose `source.source` is `\"user_steer\"` (text in\n  `payload.event.payload.prompt`, or in `payload.event.body` when the record\n  carries no payload); background and scheduled runtime deliveries share\n  that kind, so never quote those as the user.\n- `assistant_message_committed` — what the agent concluded (decisions,\n  summaries, handoffs usually live here, late in the log).\n- `assistant_tool_calls_committed` \u002F `tool_result_batch_committed` — what was\n  actually done and what it returned.\n- `terminal` — turn boundaries.\n\nRead discipline for long logs: read the TAIL first (later records matter\nmost), then only enough earlier evidence to understand context. Use bounded\n`tail`\u002F`grep` slices; never load a whole multi-megabyte log into context.\nSubagent findings live in `subagent\u002F\u003Cid>\u002Fsession.jsonl`, not the main log.\nFor product debugging of the current session, prefer the doctor skill's\nsession-evidence helper.\n\n## Restoring Lost Work\n\nWhen the user asks to restore or recover lost, wiped, or overwritten work\nand a prior session's log holds the only copy of that content (as recorded\ntool-call mutations), reconstruct each requested file exactly; do not guess\nbetween versions:\n\n1. Enumerate every mutated workspace path first — `write_file`,\n   `edit_file`, `apply_patch`, and shell writes — and build the complete\n   file list before restoring anything. Restore the full scope of the\n   ask: for a general \"restore what was lost\" ask that means every lost\n   file, while an explicitly narrower ask wins as stated.\n2. A file's final state is its mutation history replayed in record order\n   (`sequence`\u002F`recorded_at`): the LAST `write_file` content for the path\n   in record order — or the newest full-file `apply_patch`\u002Fshell write\n   when that came later — with every LATER `edit_file` delta applied in\n   the same order (patch hunks and shell edits likewise). Skip an\n   `edit_file`\u002F`apply_patch` delta whose tool result reported an error;\n   an errored shell command may still have mutated the file first, so\n   check whether later records reflect its write. Recency comes from record\n   order, never by content length or size: the longest version is often a\n   superseded draft, and refactors make the final version SHORTER.\n   Reconstructions built from `write_file` records alone silently drop\n   every later edit.\n3. Restore the exact recorded content, not a paraphrase. Re-typing from\n   memory, rewording, \"improving\", or summarizing content that is\n   recoverable verbatim is data loss, not a restore — extract the bytes\n   from the record and write those. (A user asking only to summarize a\n   prior session is not a restore; this rule governs restoring files.)\n4. Report the result per file: which records it was restored from (the\n   last full write plus the deltas applied), and what was restored and\n   what was not, with a reason for anything skipped.\n5. If two candidate final versions are genuinely ambiguous (e.g.\n   divergent edit branches after a fork or resume), present both\n   candidates with their record timestamps and let the user pick; never\n   silently prefer the longer one.\n\n## First-Party Commands\n\nPrefer these over hand-rolled parsing when they fit:\n\n- `muse resume \u003Csession-id>` or `muse resume --last` — interactive\n  continuation.\n- `muse exec --session-id \u003Csession-id> \"\u003Cfollow-up>\"` — headless\n  continuation, only on explicit request.\n- `muse export --session \u003Cid-or-session.jsonl> --redacted --out \u003Cfile>` —\n  a shareable, redacted export.\n- `muse trace inspect --session-log \u003Csession.jsonl> --render-mode compact` —\n  model-call level inspection.\n\nTreat session logs as read-only evidence: never modify, move, or delete them.\n","# Read Session\n\nMuse Code's own session storage: where your session logs live, what they\ncontain, and how to recover context from them.\n\n## Your Store, Not Anyone Else's\n\nYou are Muse Code (binary `muse`). Your sessions are stored under YOUR data\ndirectory:\n\n```\n${XDG_DATA_HOME:-$HOME\u002F.local\u002Fshare}\u002Fmuse\u002Fsessions\u002FYYYY\u002FMM\u002FDD\u002F\u003Csession-id>\u002F\n```\n\nEach session directory contains:\n\n- `session.jsonl` — the main event log (one JSON record per line).\n- `subagent\u002F\u003Cchild-session-id>\u002Fsession.jsonl` — one log per delegated\n  subagent session.\n- `tool-outputs\u002F` — full tool outputs that were too large to keep inline.\n\nMuse sessions NEVER live in another coding agent's store. Do not look for a\nMuse session under `~\u002F.claude`, `~\u002F.claude\u002Fprojects`,\n`~\u002F.local\u002Fshare\u002Fclaude`, `~\u002F.codex`, or `~\u002F.grok` — those belong to Claude\nCode, Codex, and Grok. Do not `ls`, `find`, `cat`, or `grep` those\ndirectories while recovering Muse session context — not even to \"verify\",\n\"rule out\", or \"be thorough\" about such a path quoted in a paste, log, or\nerror message. A Muse session id or date-sharded session path appearing\nunder a foreign store in quoted content is a WRONG-PATH ARTIFACT (a prior\nturn's confusion): the real data lives under the Muse pattern with the same\ndate and id, and there is no separate \"claude copy\" to collect. Name the\nartifact for what it is and move on — probing it is the exact mistake this\nrule exists to prevent, and it wastes turns on ENOENT or, worse, reads\nanother product's transcripts as your own history. Touch another agent's\ntranscripts only when the user explicitly asks to work with THAT agent's\nsession: Claude Code or Codex continuation, including recovery of unfinished\nwork with or without a handle, uses `resume-claude` or `resume-codex`. Explicit\n`\u002Fimport` requests and continuation from other agents or unnamed artifacts use\n`import`.\n\nTell for an unlabeled paste: the quoted store and record shape identify the\nproduct. For continuation or recovery of unfinished work, even without a\nsupplied handle:\n\n- Claude Code records under `~\u002F.claude\u002Fprojects` use `resume-claude`.\n- Codex records under `~\u002F.codex` use `resume-codex`.\n- Grok records under `~\u002F.grok` use `import`.\n\nUse `import` for explicit `\u002Fimport` requests or continuation from other agents\nor unnamed artifacts (the user's ask about such a paste is the explicit ask).\nThe bans above cover session STORES; ordinary project files that happen to\nlive under `~\u002F.claude` (e.g. skills you are developing) are file work, not\nsession probing.\n\nScope of \"pulling context\" from a prior Muse session: that session's own\n`session.jsonl` and `subagent\u002F` logs. External agents or planner processes\nthe session MENTIONS (e.g. codex\u002Fclaude workers it managed) are not part of\nits context — their transcripts are out of scope for this ask. If one looks\nload-bearing, say so and let the user ask for it. Do not hunt those agents'\nstores or load a foreign-session skill for them unless the user\nexplicitly asks to recover or inspect one of THEIR sessions.\n\n## Finding The Right Session\n\n1. Current session: the runtime session-identity context already names the\n   current session id and the exact `session.jsonl` path. Use it; do not\n   guess or search for it.\n2. A pasted or quoted absolute path under `...\u002Fmuse\u002Fsessions\u002F...`: use that\n   path directly.\n3. A known session id without a path: the store is sharded by local date, so\n   check the likely dates first, then search only the Muse sessions root:\n\n   ```bash\n   MUSE_SESSIONS=\"${XDG_DATA_HOME:-$HOME\u002F.local\u002Fshare}\u002Fmuse\u002Fsessions\"\n   ls -d \"$MUSE_SESSIONS\"\u002F*\u002F*\u002F*\u002F*\u002F 2>\u002Fdev\u002Fnull | grep \u003Csession-id>\n   ```\n\n4. \"Our last session\" with no id: list the most recent date shards and pick\n   the newest session directory for this workspace (the log's early\n   `runtime.session.metadata` record carries `workspace_root`).\n5. Not found under the Muse sessions root: ask the user for the id or path.\n   Never widen the hunt to other agents' directories or home-wide scans.\n\n## Reading A Session Log\n\nEach `session.jsonl` line is an event-log envelope:\n\n```json\n{\"schema_version\":1,\"id\":\"…\",\"stream\":{\"kind\":\"session\",\"id\":\"\u003Csession-id>\"},\n \"sequence\":42,\"recorded_at\":1771088000123456,\"record_type\":\"event\",\n \"durability\":\"durable\",\"causation_id\":null,\"payload_type\":\"runtime.session\",\n \"payload_schema_version\":1,\"payload\":{\"kind\":\"run\",\"run_id\":\"…\",\n \"event\":{\"kind\":\"assistant_message_committed\",\"text\":\"…\"}}}\n```\n\nThe useful `payload.event.kind` values for context recovery:\n\n- `started` — what the user asked: every submitted prompt is recorded here\n  (text in `payload.event.prompt`). A `user_prompt_display` record follows\n  only when a differing user-facing form exists (an attachment placeholder\n  such as `[Image 1]`, a composer form that differs from the sent text) —\n  prefer its text when quoting the user. A prompt typed while a run was\n  active is instead an\n  `inbox_item_queued` whose `source.source` is `\"user_steer\"` (text in\n  `payload.event.payload.prompt`, or in `payload.event.body` when the record\n  carries no payload); background and scheduled runtime deliveries share\n  that kind, so never quote those as the user.\n- `assistant_message_committed` — what the agent concluded (decisions,\n  summaries, handoffs usually live here, late in the log).\n- `assistant_tool_calls_committed` \u002F `tool_result_batch_committed` — what was\n  actually done and what it returned.\n- `terminal` — turn boundaries.\n\nRead discipline for long logs: read the TAIL first (later records matter\nmost), then only enough earlier evidence to understand context. Use bounded\n`tail`\u002F`grep` slices; never load a whole multi-megabyte log into context.\nSubagent findings live in `subagent\u002F\u003Cid>\u002Fsession.jsonl`, not the main log.\nFor product debugging of the current session, prefer the doctor skill's\nsession-evidence helper.\n\n## Restoring Lost Work\n\nWhen the user asks to restore or recover lost, wiped, or overwritten work\nand a prior session's log holds the only copy of that content (as recorded\ntool-call mutations), reconstruct each requested file exactly; do not guess\nbetween versions:\n\n1. Enumerate every mutated workspace path first — `write_file`,\n   `edit_file`, `apply_patch`, and shell writes — and build the complete\n   file list before restoring anything. Restore the full scope of the\n   ask: for a general \"restore what was lost\" ask that means every lost\n   file, while an explicitly narrower ask wins as stated.\n2. A file's final state is its mutation history replayed in record order\n   (`sequence`\u002F`recorded_at`): the LAST `write_file` content for the path\n   in record order — or the newest full-file `apply_patch`\u002Fshell write\n   when that came later — with every LATER `edit_file` delta applied in\n   the same order (patch hunks and shell edits likewise). Skip an\n   `edit_file`\u002F`apply_patch` delta whose tool result reported an error;\n   an errored shell command may still have mutated the file first, so\n   check whether later records reflect its write. Recency comes from record\n   order, never by content length or size: the longest version is often a\n   superseded draft, and refactors make the final version SHORTER.\n   Reconstructions built from `write_file` records alone silently drop\n   every later edit.\n3. Restore the exact recorded content, not a paraphrase. Re-typing from\n   memory, rewording, \"improving\", or summarizing content that is\n   recoverable verbatim is data loss, not a restore — extract the bytes\n   from the record and write those. (A user asking only to summarize a\n   prior session is not a restore; this rule governs restoring files.)\n4. Report the result per file: which records it was restored from (the\n   last full write plus the deltas applied), and what was restored and\n   what was not, with a reason for anything skipped.\n5. If two candidate final versions are genuinely ambiguous (e.g.\n   divergent edit branches after a fork or resume), present both\n   candidates with their record timestamps and let the user pick; never\n   silently prefer the longer one.\n\n## First-Party Commands\n\nPrefer these over hand-rolled parsing when they fit:\n\n- `muse resume \u003Csession-id>` or `muse resume --last` — interactive\n  continuation.\n- `muse exec --session-id \u003Csession-id> \"\u003Cfollow-up>\"` — headless\n  continuation, only on explicit request.\n- `muse export --session \u003Cid-or-session.jsonl> --redacted --out \u003Cfile>` —\n  a shareable, redacted export.\n- `muse trace inspect --session-log \u003Csession.jsonl> --render-mode compact` —\n  model-call level inspection.\n\nTreat session logs as read-only evidence: never modify, move, or delete them.\n",{"name":5,"metadata":23,"description":7,"user-invocable":19},{"short-description":24},"Read this session's or a past session's logs","main","2975a5ac469614541ae22929d43b6da2635e1ee9","Meta\u002Fmuse-code\u002Fskills\u002Fread-session","https:\u002F\u002Fgithub.com\u002FraaOS\u002Fsystem_prompts_leaks","https:\u002F\u002Fgithub.com\u002FraaOS","2026-11-14 21:19:17",{"source":8,"skillName":5,"commands":32,"prompt":42,"zipAvailable":43},[33,36,39],{"tool":34,"command":35},"npx","npx skills add raaOS\u002Fsystem_prompts_leaks --skill read-session",{"tool":37,"command":38},"bunx","bunx skills add raaOS\u002Fsystem_prompts_leaks --skill read-session",{"tool":40,"command":41},"pnpm","pnpm dlx skills add raaOS\u002Fsystem_prompts_leaks --skill read-session","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FraaOS\u002Fsystem_prompts_leaks\u002Fblob\u002Fmain\u002FMeta\u002Fmuse-code\u002Fskills\u002Fread-session\u002FSKILL.md",true,[],[46,52,59,66,73,80],{"skillNo":47,"name":48,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":49,"categories":50,"keywords":51,"publishedAt":18},"01M77JV1M0ZZTZZREM9Q96TKZZ","model-council","https:\u002F\u002Fgithub.com\u002FraaOS\u002Fsystem_prompts_leaks\u002Fblob\u002Fmain\u002FPerplexity\u002Fskills\u002Fmodel-council\u002FSKILL.md",[],[],{"skillNo":53,"name":54,"nameZh":6,"description":55,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":56,"categories":57,"keywords":58,"publishedAt":18},"01M77JTYND3ETS9EGCX3B9CW6E","writing-style","Draft or edit writing in the user's voice for the person who will read it and where it will appear. Use for messages, emails, documents or slides written on their behalf, not for your own replies.","https:\u002F\u002Fgithub.com\u002FraaOS\u002Fsystem_prompts_leaks\u002Fblob\u002Fmain\u002FOpenAI\u002Fdots\u002Fskills\u002Fwriting-style\u002FSKILL.md",[],[],{"skillNo":60,"name":61,"nameZh":6,"description":62,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":63,"categories":64,"keywords":65,"publishedAt":18},"01M77JTVWEHY80JC2H1JG1JMQB","write-page","Create or edit requested Page\u002FSpace content, or use for prose you have already decided to save as a standalone Markdown file, only if the user did not explicitly request a Markdown file. Honor established formats, destinations, and repository documentation. A selected Page alone does not authorize a write.","https:\u002F\u002Fgithub.com\u002FraaOS\u002Fsystem_prompts_leaks\u002Fblob\u002Fmain\u002FOpenAI\u002Fdots\u002Fskills\u002Fwrite-page\u002FSKILL.md",[],[],{"skillNo":67,"name":68,"nameZh":6,"description":69,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":70,"categories":71,"keywords":72,"publishedAt":18},"01M77JTQ5ND23QPF6RJYJPXHSZ","software-engineering","For use only in threads for this proactive personal assistant. Investigate software issues; inspect, write, review, or test code; work on a repository or local engineering file; or create, fix, or monitor a pull request or CI run. Use for actual engineering or product QA work, not a general programming explanation.","https:\u002F\u002Fgithub.com\u002FraaOS\u002Fsystem_prompts_leaks\u002Fblob\u002Fmain\u002FOpenAI\u002Fdots\u002Fskills\u002Fsoftware-engineering\u002FSKILL.md",[],[],{"skillNo":74,"name":75,"nameZh":6,"description":76,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":77,"categories":78,"keywords":79,"publishedAt":18},"01M77JTM3ZDPCYE8C1FW7G7S2B","slides-artifact","Use when the user asks to create, edit, or review slides or a presentation (PowerPoint or Google Slides), including when the task clearly calls for an audience-ready slide deck. Also use to answer questions that require inspecting an existing deck; not for general presentation advice or a chat-only outline.","https:\u002F\u002Fgithub.com\u002FraaOS\u002Fsystem_prompts_leaks\u002Fblob\u002Fmain\u002FOpenAI\u002Fdots\u002Fskills\u002Fslides-artifact\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},"01M77JTHTWZN2J27KD2H66XDBZ","slack","Use when reading Slack conversations, deciding when to respond, or sending messages, files, and reactions in Slack.","https:\u002F\u002Fgithub.com\u002FraaOS\u002Fsystem_prompts_leaks\u002Fblob\u002Fmain\u002FOpenAI\u002Fdots\u002Fskills\u002Fslack\u002FSKILL.md",[],[]]