[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6M98DJ3CYR928KTM44S5DM4":3,"skill-same:01M6M98DJ3CYR928KTM44S5DM4":166},{"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":16,"sourceHtmlUrl":17,"categories":18,"keywords":24,"publishedAt":46,"linkOnly":47,"markdownContent":48,"markdownBody":49,"frontMatter":50,"refName":53,"commitSha":54,"skillDirectory":55,"forksCount":56,"repositoryHtmlUrl":57,"repositoryHomepageUrl":58,"ownerHtmlUrl":59,"ownerBlogUrl":60,"ownerDescription":61,"updatedAt":62,"install":63,"resources":76},"01M6M98DJ3CYR928KTM44S5DM4","suede-graph-flo-xr",null,"Use when asked \"compare implementation plans before building\" for one multi-file repository change. Suede AI runs Suede Thought Graph shipping search: it searches, scores, prunes and refutes plans, selects one winner, then builds in an isolated worktree with a sandboxed gate and handoff. Halts on hazards, collisions or exhausted budget. NOT FOR: bulk worker tasks, existing-diff review (use suede-code), CI wiring (use suede-ci-gate), or copy-only work (use suede-ship-copy).","hashgraph-online\u002Fawesome-codex-plugins","hashgraph-online","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F160371962?v=4","01M6CNB6NRJVNNGN14TR01Q6D3","HOL (Hashgraph Online)","01M6JT1HSKKG9V68KQAQGBA7XP",1571,"BSD-3-Clause","https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002FJasonColapietro\u002Fsuede-creator-skills\u002Fskills\u002Fsuede-graph-flo-xr\u002FLICENSE.graph-of-thoughts-BSD.txt","https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002FJasonColapietro\u002Fsuede-creator-skills\u002Fskills\u002Fsuede-graph-flo-xr\u002FSKILL.md",[19],{"categoryNo":20,"categoryName":21,"taxonomyType":22,"taxonomyTypeDesc":23},"01JBSEEDCAT000000000000040","财务","PROFESSION","适用职业",[25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45],"suede graph flo xr","graph","ci","when","wiring","bulk","worker","only","diff","work","ship","Budget","review","repository","worktree","search","shipping","code","tasks","change","handoff","2026-11-10 10:12:30",false,"---\nname: suede-graph-flo-xr\ndescription: \"Use when asked \\\"compare implementation plans before building\\\" for one multi-file repository change. Suede AI runs Suede Thought Graph shipping search: it searches, scores, prunes and refutes plans, selects one winner, then builds in an isolated worktree with a sandboxed gate and handoff. Halts on hazards, collisions or exhausted budget. NOT FOR: bulk worker tasks, existing-diff review (use suede-code), CI wiring (use suede-ci-gate), or copy-only work (use suede-ship-copy).\"\nmetadata:\n  display-name: \"Suede AI Graph Flo XR Skills\"\n---\n\n# Suede AI Graph Flo XR Skills\n\nUse the bundled `workflows\u002Fsuede-graph-flo-xr.js` workflow to search competing plans for\none multi-file repository change. It makes an evidence-backed selection before\nany implementation lane mutates the worktree.\n\n## Compatibility\n\nBundled execution requires Claude Code Workflow on macOS, sandbox-exec, Node.js, Git and the six registered agent profiles. Other hosts can read the orchestration contract but must use direct implementation rather than claim Workflow execution.\n\n## Intake and budget gate\n\nBefore launch, require all three inputs:\n\n- **Repo**: an absolute repository path. Relative paths and `~` fail closed.\n- **Scope**: the requested multi-file change, including any protected paths or\n  constraints.\n- **Budget**: `light`, `standard`, or `deep`.\n\nAlso detect and pass optional context when available: `deploys` (whether the\nrepo has a deploy surface), `liveUrl` (the read-only production surface), and\n`vault` (the external decision\u002Fhandoff context path). Their absence does not\nblock a non-deploying repository, but do not silently discard known values.\n\nWhen the user names a model for the workers (`workerModel`: `sonnet`, `opus`,\n`haiku`, or `fable`), pass it: every worker call then runs on that model while\norchestration stays on the session model. Omitted, workers inherit the session\nmodel silently; if the session sits on an expensive model and the user did not\nchoose it for workers, say so before launch instead of letting the default\ndecide. A run can spend up to 200 worker calls, so an unchosen inherited model\nis a cost decision nobody made.\n\nIf repo or scope is missing, halt. Report the missing input in one line, offer\nto provide the repo path, describe the desired change, or route a one-file edit\nto direct implementation, then wait for the user's choice.\n\nState the selected range and projected worst-case calls before launching:\n`light` projects and permits **55**, `standard` projects and permits **110**, and\n`deep` projects and permits **200** total agent calls. Do not infer a budget from\nscope or silently raise a ceiling. If the user has not chosen one, ask and wait.\n\n## Runtime prerequisites\n\nThe bundled JavaScript workflow is a Claude Code workflow for macOS. It requires\n`sandbox-exec` and the six registered `suede-graph-flo-xr-*` agent profiles. Install the\nfull `suede-skills` plugin, the `suede-agent-workflows` plugin, or use this\nrepository's `install.sh`, which copies the profiles into `~\u002F.claude\u002Fagents`.\n\nClaude Workflow exposes no Node `process` global, so the workflow cannot infer\nits package namespace. The calling skill must derive it from how this skill was\ninvoked and pass it on every launch. When the invoked name carries a plugin\nprefix, `agentNamespace` is that prefix verbatim: `suede-skills` from the full\nplugin, `suede-agent-workflows` from the focused orchestration plugin. A bare\ninvoked name with no prefix, installed by `install.sh` or copied by hand, takes\nthe empty string. This is runtime context, not a user choice. A missing or\nunknown value fails before the first agent call.\n\nThe workflow also cannot locate its own bundled helper scripts. Pass `helperDir`:\nthe absolute path of the invoked skill's `workflows\u002Fhelpers` directory (for this\ninstall, `\u003Cskill base directory>\u002Fworkflows\u002Fhelpers`). The clamped Bash commands\nrun these `.cjs` helpers: the per-spawn clamp cannot verify a rule that is\nmulti-line or longer than roughly 400 characters, so inline `node -e` payloads\nare not usable. A missing or whitespace-containing path fails before the first\nagent call; a missing helper file surfaces as the Scout setup failure.\nPayload-carrying helper invocations are admitted by pinned prefixes (helper\npath plus worktree, temp root, or base SHA) rather than exact strings; each\nhelper validates its remaining argv, and the diff attestations, not the clamp,\nremain the check that what was applied matches the selected bundle.\n\nThe selected patch reaches the applier as bounded base64 chunks staged into the\nrun's private temp root, because the clamp verifier cannot parse a command\ncarrying a multi-kilobyte inline payload. Each append carries its offset and an\nFNV-1a checksum, and `--apply` verifies total length and payload checksum before\ndecoding, so a mistyped chunk fails fast with a retry instruction instead of\nproducing a corrupt patch.\n\nA skill-folder-only install, a generic skills-CLI install, and the Codex plugin do\nnot by themselves register or execute Claude Workflow agent profiles. In those\nenvironments, treat this file as the orchestration contract and route the change\nto direct implementation; do not claim the bundled workflow ran. To enable it in\nClaude Code after a manual single-skill copy, also copy the skill's bundled\n`agents\u002Fsuede-graph-flo-xr-*.md` files into `~\u002F.claude\u002Fagents` and restart Claude Code.\n\nThe requested Scout setup command probes `\u002Fusr\u002Fbin\u002Fsandbox-exec` as its first\nsubprocess, before fetch or worktree creation. If that command is invoked and\nthe probe fails, Scout reports failure before its setup mutation. The returned\nScout evidence is still a model attestation, not a host execution receipt. Gate\nalso holds on any later reported sandbox rejection. Never retry an acceptance\ncommand outside the sandbox to turn that hold into a pass.\n\n## Run the graph search\n\nInvoke:\n\n```js\nWorkflow({\n  scriptPath: \"skills\u002Fsuede-graph-flo-xr\u002Fworkflows\u002Fsuede-graph-flo-xr.js\",\n  args: { repo, scope, agentBudget, agentNamespace, helperDir, workerModel, deploys, liveUrl, vault }\n})\n```\n\nThe workflow executes these operations in dependency order:\n\n1. **Generate** independent implementation plans from the scout and research\n   evidence.\n2. **Score** each plan for coverage, evidence, feasibility, safety, and\n   efficiency.\n3. **KeepBestN** deterministically prunes the scored beam.\n4. **Refute** attacks the surviving plans with evidence-backed objections.\n5. **Improve** repairs plans whose refutations are not fatal.\n6. **Aggregate** combines compatible surviving lanes without merging conflicting\n   file ownership.\n7. **Select** chooses one deterministic winner.\n\nOnly the plan selected by **Select** may mutate files. Rejected, pruned, and\nunselected thoughts remain evidence only; never build them speculatively.\n\n## Boundaries\n\nThe workflow halts before the next agent call or entire mutating batch when its\nbudget is exhausted; it does not undo mutations that completed earlier. It\nhalts before any mutation unless an independent read-only verifier confirms a\nclean, registered origin\u002Fmain worktree at one direct `${REPO}.worktrees\u002Fship-*`\nchild with the same Git common directory and non-symlink candidate files whose\nrealpaths remain inside it. Case-folded or Unicode-normalized path aliases fail\nclosed before graph search. It also halts for\na tracked secret, a live target worktree, a protected-WIP collision, a duplicate\nfile owner, an overflowed safety manifest, or no selectable plan. Scout parses\nNUL-delimited Git porcelain so both sides of renames remain protected, parses\n`lsof -Fn` CWD fields with path-component boundaries, and never discards fresh\ndirty or live claims merely because committed history was cherry-landed. A\nselected Build or Fix result that is blocked, missing context, fails, or reports\nno changed path also halts before the next verification stage; a result that is\ndone with stated concerns proceeds, and its concerns travel into the review\nstage and the handoff. Patch authors receive the whole scope checklist as\ncontract context, so a name another lane pins is imported, not guessed. On a halt, name the\nblocker in one line and offer 2–4 applicable\nresolutions (for example: narrow scope, exempt protected WIP, resolve the\ncollision, choose a higher budget, or provide missing context), and wait. Do\nnot relaunch or mutate while halted.\n\n### Reading a search halt\n\nAn empty search used to report `no safe graph winner` however it ended, so an\ninfrastructure flake and a genuine evidence conflict printed the same line. The\nhalt output now names which happened, and `haltDetail` carries the counts behind it:\n\n| Reason | What it means |\n|---|---|\n| `every candidate lost its score to an agent failure` | No thought in the run was ever scored. Infrastructure, not evidence: rerun. |\n| `no candidate reached Select` | The search emptied upstream for some other reason; read `graph.dropped`. |\n| `every finalist lost its score before Select` | Finalists existed and were pruned as unscored. |\n| `every finalist was pruned before Select` | Finalists were pruned for a non-score reason. |\n| `every finalist carries a degraded or missing score` | Finalists reached Select without a valid score. |\n| `every finalist failed deterministic plan eligibility` | Real rejection. `haltDetail.eligibilityRejections` lists every reason. |\n| `no safe graph winner` | None of the above fits: read the graph. |\n\n`haltDetail.infrastructureDegraded` is independent of the reason: both can be\ntrue at once. Read the reason for what stopped Select and that flag for what\ndegraded the pool feeding it.\n\nScore calls are read-only and idempotent, so a transport-level death is retried:\ntwice per call, capped run-wide at 5% of the agent ceiling, and refused entirely\nonce the remaining budget falls to the reserved floor (20% of the ceiling). A\nmalformed score is never retried: the schema is enforced at the tool layer, so\nan invalid score is a judgment to keep, not a connection to redial. Every\nattempt and every refused retry lands in `graph.scoreRetries`, and\n`scoreReliability` rides out in the result on every run, halted or not: a flake\nthat costs two finalists still degrades a run that goes on to ship.\n\nClaude's registered agent profiles enforce tool separation: local readers have\nno shell, write, or web tools; public-web readers have no local-file or shell\ntools; patch authors have no mutation tools; and appliers\u002Fverifiers have only\nBash plus structured output. Patch authors return unified diffs, one clamped\napplier applies them, and a separately budget-reserved clamped verifier compares\nthe exact path set and diff digest immediately after every Build or Fix Apply,\nbefore any reader or Gate call. Patch validation rejects symlinks, gitlinks,\nbinary patches, renames, copies, and file-type transitions before Apply. Gate\nruns only allowlisted local validation\ncommands under macOS `sandbox-exec`, with no network, host reads limited to\nruntime\u002Fsystem roots, the worktree, its `.git` common directory derived again\ninside the exact Gate clamp, and the run's private temp root. The model-reported\ncommon directory is never interpolated into sandbox permissions. Writes are limited to known generated artifacts\nand that private temp root. The allowlist includes bounded project-local checks\nfor Node, Python, Go, Rust, Make, Swift Package Manager, Xcode simulator builds\nwith derived data under the private temp root, and offline Gradle validation.\nNested module `build` roots are derived only from selected files under that\nmodule's `src` tree and are rejected if a symlink or realpath can escape the\nworktree. A second diff attestation runs after Gate and hashes\nthe binary Git diff plus every reported file's mode, size, and bytes, including\nuntracked additions.\n\nGate removes credential-like and interpreter-injection environment variables,\nthen redirects home, temporary, and cache paths before an acceptance command\nstarts. If a check depends on removed credentials, report it as unverified;\nnever rerun it outside the sandbox merely to obtain a pass.\n\nA successfully applied blocker patch is not treated as semantically cleared.\nThe original blocker remains in `fixedBlockersPendingVerification`. The Gate\nattempt records its exact command set and reported output, but it cannot prove\nthose commands ran because the Workflow API exposes no trusted required-tool\nexecution receipt. The workflow therefore sets `claimedPassed` from the agent\nreport, forces `passed:false`, sets `gateVerified:false`, and keeps the verdict\nand handoff status at `hold`. Only a trusted outer runner with immutable\nexecution receipts can promote that evidence.\n\nThese controls have a precise trust boundary. `bashCommandClamp` constrains a\nBash command when an agent invokes it; Claude Workflow does not provide a\nrequired-tool-call receipt, so a structured verifier response remains a model\nattestation rather than cryptographic proof that Bash ran. Likewise,\n`authority`, `allowedRepo`, `allowedFiles`, and `allowedCommands` are audit\nmetadata, not filesystem permissions. Local reader tools are separated from web\ntools but are not path-sandboxed by the Workflow API. Report these facts in any\nsecurity-sensitive handoff and do not describe the result as host-certified.\n\nProduction inspection is read-only. This skill never deploys, publishes,\nreleases, pushes, merges, changes credentials, deletes or reverts protected\nwork, or claims live verification. It does not choose the user's budget or\ndecide that missing scope can be skipped. Its ship verdict is evidence for the\nuser, not authority to perform an external action.\n\n## Handoff and completion\n\nRead the workflow's returned `runKey`, the validated unique `ship-\u003CUUID>` leaf\nfrom its isolated worktree. On a completed run, use the returned handoff\nmarkdown. On a post-Scout halt, write a factual halt handoff from the structured\nresult and graph trace without spending another agent call; include any Build\nor Fix lanes that completed before the halt. If Scout returns an invalid path\nbefore `runKey` validation, report the halt without writing a run-keyed\nhandoff. Otherwise, save it to\n`.suede-graph-flo-xr\u002F${runKey}\u002Fhandoff.md` at the target repo root, then verify it exists:\n\n```bash\ntest -f \".suede-graph-flo-xr\u002F${runKey}\u002Fhandoff.md\"\n```\n\nReport that path, the selected plan if any, gate result, changed files, commands\nrun, and explicit caveats. A completed local graph does not prove a deployment.\n\n## License file\n\nPart of `workflows\u002Fsuede-graph-flo-xr.js` is distributed under BSD-3-Clause\nterms carried in `LICENSE.graph-of-thoughts-BSD.txt`. Keep that file with every\nsource or binary redistribution of the workflow.\n\n## Routing\n\n- High-volume, well-specified, independent worker tasks → a separate private\n  worker-fleet pass.\n- Review or grade an existing diff → `suede-code`.\n- CI, required checks, or branch-protection wiring → `suede-ci-gate`.\n- Copy-only search and publication readiness → `suede-ship-copy`.\n- From `suede-code`, `suede-ci-gate`, or `suede-ship-copy`: route a\n  multi-file implementation-plan search with one\n  selected mutating winner back to `suede-graph-flo-xr`.\n\n## Worked contract\n\n[Worked contract](references\u002Fusage-contract.md) gives a bounded input, expected output and failure check for this workflow.\n","# Suede AI Graph Flo XR Skills\n\nUse the bundled `workflows\u002Fsuede-graph-flo-xr.js` workflow to search competing plans for\none multi-file repository change. It makes an evidence-backed selection before\nany implementation lane mutates the worktree.\n\n## Compatibility\n\nBundled execution requires Claude Code Workflow on macOS, sandbox-exec, Node.js, Git and the six registered agent profiles. Other hosts can read the orchestration contract but must use direct implementation rather than claim Workflow execution.\n\n## Intake and budget gate\n\nBefore launch, require all three inputs:\n\n- **Repo**: an absolute repository path. Relative paths and `~` fail closed.\n- **Scope**: the requested multi-file change, including any protected paths or\n  constraints.\n- **Budget**: `light`, `standard`, or `deep`.\n\nAlso detect and pass optional context when available: `deploys` (whether the\nrepo has a deploy surface), `liveUrl` (the read-only production surface), and\n`vault` (the external decision\u002Fhandoff context path). Their absence does not\nblock a non-deploying repository, but do not silently discard known values.\n\nWhen the user names a model for the workers (`workerModel`: `sonnet`, `opus`,\n`haiku`, or `fable`), pass it: every worker call then runs on that model while\norchestration stays on the session model. Omitted, workers inherit the session\nmodel silently; if the session sits on an expensive model and the user did not\nchoose it for workers, say so before launch instead of letting the default\ndecide. A run can spend up to 200 worker calls, so an unchosen inherited model\nis a cost decision nobody made.\n\nIf repo or scope is missing, halt. Report the missing input in one line, offer\nto provide the repo path, describe the desired change, or route a one-file edit\nto direct implementation, then wait for the user's choice.\n\nState the selected range and projected worst-case calls before launching:\n`light` projects and permits **55**, `standard` projects and permits **110**, and\n`deep` projects and permits **200** total agent calls. Do not infer a budget from\nscope or silently raise a ceiling. If the user has not chosen one, ask and wait.\n\n## Runtime prerequisites\n\nThe bundled JavaScript workflow is a Claude Code workflow for macOS. It requires\n`sandbox-exec` and the six registered `suede-graph-flo-xr-*` agent profiles. Install the\nfull `suede-skills` plugin, the `suede-agent-workflows` plugin, or use this\nrepository's `install.sh`, which copies the profiles into `~\u002F.claude\u002Fagents`.\n\nClaude Workflow exposes no Node `process` global, so the workflow cannot infer\nits package namespace. The calling skill must derive it from how this skill was\ninvoked and pass it on every launch. When the invoked name carries a plugin\nprefix, `agentNamespace` is that prefix verbatim: `suede-skills` from the full\nplugin, `suede-agent-workflows` from the focused orchestration plugin. A bare\ninvoked name with no prefix, installed by `install.sh` or copied by hand, takes\nthe empty string. This is runtime context, not a user choice. A missing or\nunknown value fails before the first agent call.\n\nThe workflow also cannot locate its own bundled helper scripts. Pass `helperDir`:\nthe absolute path of the invoked skill's `workflows\u002Fhelpers` directory (for this\ninstall, `\u003Cskill base directory>\u002Fworkflows\u002Fhelpers`). The clamped Bash commands\nrun these `.cjs` helpers: the per-spawn clamp cannot verify a rule that is\nmulti-line or longer than roughly 400 characters, so inline `node -e` payloads\nare not usable. A missing or whitespace-containing path fails before the first\nagent call; a missing helper file surfaces as the Scout setup failure.\nPayload-carrying helper invocations are admitted by pinned prefixes (helper\npath plus worktree, temp root, or base SHA) rather than exact strings; each\nhelper validates its remaining argv, and the diff attestations, not the clamp,\nremain the check that what was applied matches the selected bundle.\n\nThe selected patch reaches the applier as bounded base64 chunks staged into the\nrun's private temp root, because the clamp verifier cannot parse a command\ncarrying a multi-kilobyte inline payload. Each append carries its offset and an\nFNV-1a checksum, and `--apply` verifies total length and payload checksum before\ndecoding, so a mistyped chunk fails fast with a retry instruction instead of\nproducing a corrupt patch.\n\nA skill-folder-only install, a generic skills-CLI install, and the Codex plugin do\nnot by themselves register or execute Claude Workflow agent profiles. In those\nenvironments, treat this file as the orchestration contract and route the change\nto direct implementation; do not claim the bundled workflow ran. To enable it in\nClaude Code after a manual single-skill copy, also copy the skill's bundled\n`agents\u002Fsuede-graph-flo-xr-*.md` files into `~\u002F.claude\u002Fagents` and restart Claude Code.\n\nThe requested Scout setup command probes `\u002Fusr\u002Fbin\u002Fsandbox-exec` as its first\nsubprocess, before fetch or worktree creation. If that command is invoked and\nthe probe fails, Scout reports failure before its setup mutation. The returned\nScout evidence is still a model attestation, not a host execution receipt. Gate\nalso holds on any later reported sandbox rejection. Never retry an acceptance\ncommand outside the sandbox to turn that hold into a pass.\n\n## Run the graph search\n\nInvoke:\n\n```js\nWorkflow({\n  scriptPath: \"skills\u002Fsuede-graph-flo-xr\u002Fworkflows\u002Fsuede-graph-flo-xr.js\",\n  args: { repo, scope, agentBudget, agentNamespace, helperDir, workerModel, deploys, liveUrl, vault }\n})\n```\n\nThe workflow executes these operations in dependency order:\n\n1. **Generate** independent implementation plans from the scout and research\n   evidence.\n2. **Score** each plan for coverage, evidence, feasibility, safety, and\n   efficiency.\n3. **KeepBestN** deterministically prunes the scored beam.\n4. **Refute** attacks the surviving plans with evidence-backed objections.\n5. **Improve** repairs plans whose refutations are not fatal.\n6. **Aggregate** combines compatible surviving lanes without merging conflicting\n   file ownership.\n7. **Select** chooses one deterministic winner.\n\nOnly the plan selected by **Select** may mutate files. Rejected, pruned, and\nunselected thoughts remain evidence only; never build them speculatively.\n\n## Boundaries\n\nThe workflow halts before the next agent call or entire mutating batch when its\nbudget is exhausted; it does not undo mutations that completed earlier. It\nhalts before any mutation unless an independent read-only verifier confirms a\nclean, registered origin\u002Fmain worktree at one direct `${REPO}.worktrees\u002Fship-*`\nchild with the same Git common directory and non-symlink candidate files whose\nrealpaths remain inside it. Case-folded or Unicode-normalized path aliases fail\nclosed before graph search. It also halts for\na tracked secret, a live target worktree, a protected-WIP collision, a duplicate\nfile owner, an overflowed safety manifest, or no selectable plan. Scout parses\nNUL-delimited Git porcelain so both sides of renames remain protected, parses\n`lsof -Fn` CWD fields with path-component boundaries, and never discards fresh\ndirty or live claims merely because committed history was cherry-landed. A\nselected Build or Fix result that is blocked, missing context, fails, or reports\nno changed path also halts before the next verification stage; a result that is\ndone with stated concerns proceeds, and its concerns travel into the review\nstage and the handoff. Patch authors receive the whole scope checklist as\ncontract context, so a name another lane pins is imported, not guessed. On a halt, name the\nblocker in one line and offer 2–4 applicable\nresolutions (for example: narrow scope, exempt protected WIP, resolve the\ncollision, choose a higher budget, or provide missing context), and wait. Do\nnot relaunch or mutate while halted.\n\n### Reading a search halt\n\nAn empty search used to report `no safe graph winner` however it ended, so an\ninfrastructure flake and a genuine evidence conflict printed the same line. The\nhalt output now names which happened, and `haltDetail` carries the counts behind it:\n\n| Reason | What it means |\n|---|---|\n| `every candidate lost its score to an agent failure` | No thought in the run was ever scored. Infrastructure, not evidence: rerun. |\n| `no candidate reached Select` | The search emptied upstream for some other reason; read `graph.dropped`. |\n| `every finalist lost its score before Select` | Finalists existed and were pruned as unscored. |\n| `every finalist was pruned before Select` | Finalists were pruned for a non-score reason. |\n| `every finalist carries a degraded or missing score` | Finalists reached Select without a valid score. |\n| `every finalist failed deterministic plan eligibility` | Real rejection. `haltDetail.eligibilityRejections` lists every reason. |\n| `no safe graph winner` | None of the above fits: read the graph. |\n\n`haltDetail.infrastructureDegraded` is independent of the reason: both can be\ntrue at once. Read the reason for what stopped Select and that flag for what\ndegraded the pool feeding it.\n\nScore calls are read-only and idempotent, so a transport-level death is retried:\ntwice per call, capped run-wide at 5% of the agent ceiling, and refused entirely\nonce the remaining budget falls to the reserved floor (20% of the ceiling). A\nmalformed score is never retried: the schema is enforced at the tool layer, so\nan invalid score is a judgment to keep, not a connection to redial. Every\nattempt and every refused retry lands in `graph.scoreRetries`, and\n`scoreReliability` rides out in the result on every run, halted or not: a flake\nthat costs two finalists still degrades a run that goes on to ship.\n\nClaude's registered agent profiles enforce tool separation: local readers have\nno shell, write, or web tools; public-web readers have no local-file or shell\ntools; patch authors have no mutation tools; and appliers\u002Fverifiers have only\nBash plus structured output. Patch authors return unified diffs, one clamped\napplier applies them, and a separately budget-reserved clamped verifier compares\nthe exact path set and diff digest immediately after every Build or Fix Apply,\nbefore any reader or Gate call. Patch validation rejects symlinks, gitlinks,\nbinary patches, renames, copies, and file-type transitions before Apply. Gate\nruns only allowlisted local validation\ncommands under macOS `sandbox-exec`, with no network, host reads limited to\nruntime\u002Fsystem roots, the worktree, its `.git` common directory derived again\ninside the exact Gate clamp, and the run's private temp root. The model-reported\ncommon directory is never interpolated into sandbox permissions. Writes are limited to known generated artifacts\nand that private temp root. The allowlist includes bounded project-local checks\nfor Node, Python, Go, Rust, Make, Swift Package Manager, Xcode simulator builds\nwith derived data under the private temp root, and offline Gradle validation.\nNested module `build` roots are derived only from selected files under that\nmodule's `src` tree and are rejected if a symlink or realpath can escape the\nworktree. A second diff attestation runs after Gate and hashes\nthe binary Git diff plus every reported file's mode, size, and bytes, including\nuntracked additions.\n\nGate removes credential-like and interpreter-injection environment variables,\nthen redirects home, temporary, and cache paths before an acceptance command\nstarts. If a check depends on removed credentials, report it as unverified;\nnever rerun it outside the sandbox merely to obtain a pass.\n\nA successfully applied blocker patch is not treated as semantically cleared.\nThe original blocker remains in `fixedBlockersPendingVerification`. The Gate\nattempt records its exact command set and reported output, but it cannot prove\nthose commands ran because the Workflow API exposes no trusted required-tool\nexecution receipt. The workflow therefore sets `claimedPassed` from the agent\nreport, forces `passed:false`, sets `gateVerified:false`, and keeps the verdict\nand handoff status at `hold`. Only a trusted outer runner with immutable\nexecution receipts can promote that evidence.\n\nThese controls have a precise trust boundary. `bashCommandClamp` constrains a\nBash command when an agent invokes it; Claude Workflow does not provide a\nrequired-tool-call receipt, so a structured verifier response remains a model\nattestation rather than cryptographic proof that Bash ran. Likewise,\n`authority`, `allowedRepo`, `allowedFiles`, and `allowedCommands` are audit\nmetadata, not filesystem permissions. Local reader tools are separated from web\ntools but are not path-sandboxed by the Workflow API. Report these facts in any\nsecurity-sensitive handoff and do not describe the result as host-certified.\n\nProduction inspection is read-only. This skill never deploys, publishes,\nreleases, pushes, merges, changes credentials, deletes or reverts protected\nwork, or claims live verification. It does not choose the user's budget or\ndecide that missing scope can be skipped. Its ship verdict is evidence for the\nuser, not authority to perform an external action.\n\n## Handoff and completion\n\nRead the workflow's returned `runKey`, the validated unique `ship-\u003CUUID>` leaf\nfrom its isolated worktree. On a completed run, use the returned handoff\nmarkdown. On a post-Scout halt, write a factual halt handoff from the structured\nresult and graph trace without spending another agent call; include any Build\nor Fix lanes that completed before the halt. If Scout returns an invalid path\nbefore `runKey` validation, report the halt without writing a run-keyed\nhandoff. Otherwise, save it to\n`.suede-graph-flo-xr\u002F${runKey}\u002Fhandoff.md` at the target repo root, then verify it exists:\n\n```bash\ntest -f \".suede-graph-flo-xr\u002F${runKey}\u002Fhandoff.md\"\n```\n\nReport that path, the selected plan if any, gate result, changed files, commands\nrun, and explicit caveats. A completed local graph does not prove a deployment.\n\n## License file\n\nPart of `workflows\u002Fsuede-graph-flo-xr.js` is distributed under BSD-3-Clause\nterms carried in `LICENSE.graph-of-thoughts-BSD.txt`. Keep that file with every\nsource or binary redistribution of the workflow.\n\n## Routing\n\n- High-volume, well-specified, independent worker tasks → a separate private\n  worker-fleet pass.\n- Review or grade an existing diff → `suede-code`.\n- CI, required checks, or branch-protection wiring → `suede-ci-gate`.\n- Copy-only search and publication readiness → `suede-ship-copy`.\n- From `suede-code`, `suede-ci-gate`, or `suede-ship-copy`: route a\n  multi-file implementation-plan search with one\n  selected mutating winner back to `suede-graph-flo-xr`.\n\n## Worked contract\n\n[Worked contract](references\u002Fusage-contract.md) gives a bounded input, expected output and failure check for this workflow.\n",{"name":5,"metadata":51,"description":7},{"display-name":52},"Suede AI Graph Flo XR Skills","main","2b63f856ce19c2a797929cacabd9e0fc936812fa","plugins\u002FJasonColapietro\u002Fsuede-creator-skills\u002Fskills\u002Fsuede-graph-flo-xr",490,"https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins","https:\u002F\u002Fhol.org\u002Fplugins\u002Fbest-codex-plugins","https:\u002F\u002Fgithub.com\u002Fhashgraph-online","https:\u002F\u002Fhol.org","Open-source standards and developer tools for interoperable AI agents: discovery, identity, communication, payments, privacy and security.","2026-11-14 02:12:02",{"source":8,"skillName":5,"commands":64,"prompt":74,"zipAvailable":75},[65,68,71],{"tool":66,"command":67},"npx","npx skills add hashgraph-online\u002Fawesome-codex-plugins --skill suede-graph-flo-xr",{"tool":69,"command":70},"bunx","bunx skills add hashgraph-online\u002Fawesome-codex-plugins --skill suede-graph-flo-xr",{"tool":72,"command":73},"pnpm","pnpm dlx skills add hashgraph-online\u002Fawesome-codex-plugins --skill suede-graph-flo-xr","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002FJasonColapietro\u002Fsuede-creator-skills\u002Fskills\u002Fsuede-graph-flo-xr\u002FSKILL.md",true,[77,83,87,91,95,99,103,107,111,116,120,126,130,134,138,142,146,150,154,158,162],{"path":78,"resourceType":79,"resourceTypeDesc":80,"sizeBytes":81,"sourceHtmlUrl":82},"agents\u002Fopenai.yaml","REFERENCE","参考文档",461,"https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002FJasonColapietro\u002Fsuede-creator-skills\u002Fskills\u002Fsuede-graph-flo-xr\u002Fagents\u002Fopenai.yaml",{"path":84,"resourceType":79,"resourceTypeDesc":80,"sizeBytes":85,"sourceHtmlUrl":86},"agents\u002Fsuede-graph-flo-xr-applier.md",544,"https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002FJasonColapietro\u002Fsuede-creator-skills\u002Fskills\u002Fsuede-graph-flo-xr\u002Fagents\u002Fsuede-graph-flo-xr-applier.md",{"path":88,"resourceType":79,"resourceTypeDesc":80,"sizeBytes":89,"sourceHtmlUrl":90},"agents\u002Fsuede-graph-flo-xr-code-reader.md",650,"https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002FJasonColapietro\u002Fsuede-creator-skills\u002Fskills\u002Fsuede-graph-flo-xr\u002Fagents\u002Fsuede-graph-flo-xr-code-reader.md",{"path":92,"resourceType":79,"resourceTypeDesc":80,"sizeBytes":93,"sourceHtmlUrl":94},"agents\u002Fsuede-graph-flo-xr-patch-author.md",676,"https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002FJasonColapietro\u002Fsuede-creator-skills\u002Fskills\u002Fsuede-graph-flo-xr\u002Fagents\u002Fsuede-graph-flo-xr-patch-author.md",{"path":96,"resourceType":79,"resourceTypeDesc":80,"sizeBytes":97,"sourceHtmlUrl":98},"agents\u002Fsuede-graph-flo-xr-scout.md",682,"https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002FJasonColapietro\u002Fsuede-creator-skills\u002Fskills\u002Fsuede-graph-flo-xr\u002Fagents\u002Fsuede-graph-flo-xr-scout.md",{"path":100,"resourceType":79,"resourceTypeDesc":80,"sizeBytes":101,"sourceHtmlUrl":102},"agents\u002Fsuede-graph-flo-xr-verifier.md",575,"https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002FJasonColapietro\u002Fsuede-creator-skills\u002Fskills\u002Fsuede-graph-flo-xr\u002Fagents\u002Fsuede-graph-flo-xr-verifier.md",{"path":104,"resourceType":79,"resourceTypeDesc":80,"sizeBytes":105,"sourceHtmlUrl":106},"agents\u002Fsuede-graph-flo-xr-web-reader.md",654,"https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002FJasonColapietro\u002Fsuede-creator-skills\u002Fskills\u002Fsuede-graph-flo-xr\u002Fagents\u002Fsuede-graph-flo-xr-web-reader.md",{"path":108,"resourceType":79,"resourceTypeDesc":80,"sizeBytes":109,"sourceHtmlUrl":110},"CARD.md",5765,"https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002FJasonColapietro\u002Fsuede-creator-skills\u002Fskills\u002Fsuede-graph-flo-xr\u002FCARD.md",{"path":112,"resourceType":113,"resourceTypeDesc":114,"sizeBytes":115,"sourceHtmlUrl":16},"LICENSE.graph-of-thoughts-BSD.txt","LICENSE","许可证",2532,{"path":117,"resourceType":79,"resourceTypeDesc":80,"sizeBytes":118,"sourceHtmlUrl":119},"references\u002Fusage-contract.md",914,"https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002FJasonColapietro\u002Fsuede-creator-skills\u002Fskills\u002Fsuede-graph-flo-xr\u002Freferences\u002Fusage-contract.md",{"path":121,"resourceType":122,"resourceTypeDesc":123,"sizeBytes":124,"sourceHtmlUrl":125},"workflows\u002Fhelpers\u002Fapply-patch.cjs","SCRIPT","脚本",5441,"https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002FJasonColapietro\u002Fsuede-creator-skills\u002Fskills\u002Fsuede-graph-flo-xr\u002Fworkflows\u002Fhelpers\u002Fapply-patch.cjs",{"path":127,"resourceType":122,"resourceTypeDesc":123,"sizeBytes":128,"sourceHtmlUrl":129},"workflows\u002Fhelpers\u002Fcandidate-audit.cjs",2317,"https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002FJasonColapietro\u002Fsuede-creator-skills\u002Fskills\u002Fsuede-graph-flo-xr\u002Fworkflows\u002Fhelpers\u002Fcandidate-audit.cjs",{"path":131,"resourceType":122,"resourceTypeDesc":123,"sizeBytes":132,"sourceHtmlUrl":133},"workflows\u002Fhelpers\u002Fdiff-digest.cjs",1725,"https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002FJasonColapietro\u002Fsuede-creator-skills\u002Fskills\u002Fsuede-graph-flo-xr\u002Fworkflows\u002Fhelpers\u002Fdiff-digest.cjs",{"path":135,"resourceType":122,"resourceTypeDesc":123,"sizeBytes":136,"sourceHtmlUrl":137},"workflows\u002Fhelpers\u002Fgate-environment.cjs",1995,"https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002FJasonColapietro\u002Fsuede-creator-skills\u002Fskills\u002Fsuede-graph-flo-xr\u002Fworkflows\u002Fhelpers\u002Fgate-environment.cjs",{"path":139,"resourceType":122,"resourceTypeDesc":123,"sizeBytes":140,"sourceHtmlUrl":141},"workflows\u002Fhelpers\u002Fgate-sandbox.cjs",4948,"https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002FJasonColapietro\u002Fsuede-creator-skills\u002Fskills\u002Fsuede-graph-flo-xr\u002Fworkflows\u002Fhelpers\u002Fgate-sandbox.cjs",{"path":143,"resourceType":122,"resourceTypeDesc":123,"sizeBytes":144,"sourceHtmlUrl":145},"workflows\u002Fhelpers\u002Fscout-setup.cjs",4758,"https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002FJasonColapietro\u002Fsuede-creator-skills\u002Fskills\u002Fsuede-graph-flo-xr\u002Fworkflows\u002Fhelpers\u002Fscout-setup.cjs",{"path":147,"resourceType":122,"resourceTypeDesc":123,"sizeBytes":148,"sourceHtmlUrl":149},"workflows\u002Fsuede-graph-flo-xr.js",174421,"https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002FJasonColapietro\u002Fsuede-creator-skills\u002Fskills\u002Fsuede-graph-flo-xr\u002Fworkflows\u002Fsuede-graph-flo-xr.js",{"path":151,"resourceType":122,"resourceTypeDesc":123,"sizeBytes":152,"sourceHtmlUrl":153},"workflows\u002Ftests\u002Ftest_apply_patch_staging_safety.mjs",4135,"https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002FJasonColapietro\u002Fsuede-creator-skills\u002Fskills\u002Fsuede-graph-flo-xr\u002Fworkflows\u002Ftests\u002Ftest_apply_patch_staging_safety.mjs",{"path":155,"resourceType":122,"resourceTypeDesc":123,"sizeBytes":156,"sourceHtmlUrl":157},"workflows\u002Ftests\u002Ftest_gate_sandbox_staging_safety.mjs",2497,"https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002FJasonColapietro\u002Fsuede-creator-skills\u002Fskills\u002Fsuede-graph-flo-xr\u002Fworkflows\u002Ftests\u002Ftest_gate_sandbox_staging_safety.mjs",{"path":159,"resourceType":122,"resourceTypeDesc":123,"sizeBytes":160,"sourceHtmlUrl":161},"workflows\u002Ftests\u002Ftest_plan_path_safety.mjs",17846,"https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002FJasonColapietro\u002Fsuede-creator-skills\u002Fskills\u002Fsuede-graph-flo-xr\u002Fworkflows\u002Ftests\u002Ftest_plan_path_safety.mjs",{"path":163,"resourceType":122,"resourceTypeDesc":123,"sizeBytes":164,"sourceHtmlUrl":165},"workflows\u002Ftests\u002Ftest_score_resilience.mjs",14372,"https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002FJasonColapietro\u002Fsuede-creator-skills\u002Fskills\u002Fsuede-graph-flo-xr\u002Fworkflows\u002Ftests\u002Ftest_score_resilience.mjs",[167,176,183,190,214,238],{"skillNo":168,"name":169,"nameZh":6,"description":170,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":171,"licenseFileUrl":6,"sourceHtmlUrl":172,"categories":173,"keywords":174,"publishedAt":175},"01M746ZWACVKMZ1BJXJZHYJSSX","migrate-to-e2e","Moves an existing Playwright, Cypress, Selenium, Appium, Detox, Maestro, or TesterArmy suite to e2e one slice at a time, with the old suite running beside the new one until each ported test is proven. Keeps a MIGRATION.md progress file so any later session resumes where the last stopped. Optionally plants a defect to show both suites catch it. Use for \u002Fmigrate-to-e2e, \"migrate our tests to e2e\", \"port these Maestro flows\", \"move our TesterArmy tests\", or when a project has an e2e config and an older suite still running.","Apache-2.0","https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Ftester-army\u002Fe2e\u002Fskills\u002Fmigrate-to-e2e\u002FSKILL.md",[],[],"2026-11-11 15:11:14",{"skillNo":177,"name":178,"nameZh":6,"description":179,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":171,"licenseFileUrl":6,"sourceHtmlUrl":180,"categories":181,"keywords":182,"publishedAt":175},"01M746ZPTAX1Q0FBKAB1CYND7E","e2e","Writes, runs, and debugs end-to-end tests for web and mobile apps with e2e, the agentic test runner, which mixes agent.act goals with exact locator assertions on a Playwright browser or an iOS or Android device. Also drives a project's running app live over the e2e MCP server to look at a screen, check a UI change, take a screenshot or record a video, and try a locator; and runs bug bashes, parallel e2e explore runs with each bug proven by a repro test. Use when a project depends on e2e or has an e2e.config.ts, when asked for end-to-end, browser, mobile, or agentic UI tests, to check a change in the running app, to QA or bug bash an app or branch, or when an e2e run fails.","https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Ftester-army\u002Fe2e\u002Fskills\u002Fe2e\u002FSKILL.md",[],[],{"skillNo":184,"name":185,"nameZh":6,"description":186,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":171,"licenseFileUrl":6,"sourceHtmlUrl":187,"categories":188,"keywords":189,"publishedAt":175},"01M746ZJN784HVHMDP9MQGCSGG","create-verification-skill","Generate a project-local verify-\u003Capp> skill on top of e2e, so any coding agent can launch the app, check it is worth driving, exercise a feature the way a user does, keep evidence, and bug bash it. Writes the skill, a feature map with one explore charter per feature, and proves it once. Use for \u002Fcreate-verification-skill, \"make a verification skill for this repo\", or when a project has no scripted way to prove UI behavior.","https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Ftester-army\u002Fe2e\u002Fskills\u002Fcreate-verification-skill\u002FSKILL.md",[],[],{"skillNo":191,"name":192,"nameZh":6,"description":193,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":171,"licenseFileUrl":6,"sourceHtmlUrl":194,"categories":195,"keywords":196,"publishedAt":46},"01M716A6ETNB9B1SQ3GXZC75VR","deja-search","deja-vu memory — search the user's past AI coding sessions with the deja CLI. Holds what they already settled and never wrote into the repo (naming rules, banned or already-tried tools, required checks before merge, preferred values, who owns what). Use before your first edit in a task, before you call a change done or ready to merge, before you recommend a tool, library, design or reviewer, before re-debugging an error that may already be solved, when they say 'didn't we fix this before', 'what did we decide about X' or 'I already have that', and before telling them something on this machine does not exist.","https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Fvshulcz\u002Fdeja-vu\u002Fskills\u002Fdeja-search\u002FSKILL.md",[],[40,197,198,199,200,201,202,203,204,205,206,28,207,208,209,210,44,211,212,213],"deja search","never","error","ready","memory","re","decide","fix","design","debugging","you","done","reviewer","io","cli","merge","past",{"skillNo":215,"name":216,"nameZh":6,"description":217,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":171,"licenseFileUrl":6,"sourceHtmlUrl":218,"categories":219,"keywords":220,"publishedAt":46},"01M716A3J2W5SENSE9VGMXB375","deja-reflect","End-of-session reflection and memory curation with deja-vu. Review what was decided, what commands broke and what repaired them, and record durable project rules or decisions with deja remember so the next session starts with them.","https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Fvshulcz\u002Fdeja-vu\u002Fskills\u002Fdeja-reflect\u002FSKILL.md",[],[221,222,37,210,223,224,225,226,227,228,229,230,231,232,201,233,234,235,236,237],"deja reflect","reflect","with","record","github","next","https","node","Rules","decisions","brew","remember","project rules","project","session","end","reflection",{"skillNo":239,"name":240,"nameZh":6,"description":241,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":171,"licenseFileUrl":6,"sourceHtmlUrl":242,"categories":243,"keywords":255,"publishedAt":46},"01M715T3RR4BAYSR57HPPTJ0PF","revid","Make, edit, export and post AI videos with Revid (revid.ai) through the Revid MCP server. Use whenever the user wants a short-form or long-form video made (from an idea, a script, a URL, an article, a PDF, a tweet, a product page, their own clips, or a song), wants to find what is going viral on TikTok, YouTube, Instagram or X before making something, wants to edit or re-export an existing Revid project, generate a standalone AI image or clip, make a series episode, or publish and schedule a finished video to connected social accounts. Covers setup (hosted MCP with OAuth, or an API key), the 36 tools, credits, polling and the rules for publishing. Routes to the more specific revid-* skills when one fits.","https:\u002F\u002Fgithub.com\u002Fhashgraph-online\u002Fawesome-codex-plugins\u002Fblob\u002Fmain\u002Fplugins\u002Ftmakersas\u002Frevid-skills\u002Fskills\u002Frevid\u002FSKILL.md",[244,249,252],{"categoryNo":245,"categoryName":246,"taxonomyType":247,"taxonomyTypeDesc":248},"01JBSEEDCAT000000000000016","图像与多媒体","SCENARIO","使用场景",{"categoryNo":250,"categoryName":251,"taxonomyType":247,"taxonomyTypeDesc":248},"01JBSEEDCAT000000000000023","Skill与智能体开发",{"categoryNo":253,"categoryName":254,"taxonomyType":247,"taxonomyTypeDesc":248},"01JBSEEDCAT000000000000010","后端与接口",[240,256,257,258,259,260,261,234,262,202,263,264,265,266,28,267,268,269,270,271],"URL","Video","short-form","standalone","PDF","viral","tiktok","short","social","MCP","song","image","mcp server","schedule","api","publish"]