[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7EP2ZKMKXREMZJ3VB96CTQG":3,"skill-same:01M7EP2ZKMKXREMZJ3VB96CTQG":50},{"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":24,"commitSha":25,"skillDirectory":26,"forksCount":14,"repositoryHtmlUrl":27,"repositoryHomepageUrl":6,"ownerHtmlUrl":28,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":29,"install":30,"resources":43},"01M7EP2ZKMKXREMZJ3VB96CTQG","gauntlet",null,"The Gauntlet orchestrator — run any complex or multi-step request through a five-stage pipeline (Mentat → Skilltank → Ethicsminions → Sandboxer → Foaf), wrapped by Takushi (dispatch at intake, delivery at exit), before handing back a single unified result. Use this whenever the user invokes \"Gauntlet\", asks to \"run this through the Gauntlet \u002F the Protocol\", wants a request planned, tool-matched, safety-checked, tested, context-fitted, and delivered in one pass, or is working in a project that names Gauntlet as its orchestrator. Also use for substantial tasks (building files, writing\u002Frunning code, research deliverables, anything irreversible) where a plan-then-verify discipline makes the result safer and better, even if the user doesn't say \"Gauntlet\" by name. Coordinates, never replaces: the core six plus consent-gate, eclipse-pass, color-theory, principles, manifest, rosetta, primer, frugal-offload, llama-bridge, mega-voltron, pmo, session-rotate, skill-creator, medley-review, and docx\u002Fxlsx\u002Fpptx\u002Fpdf.","inputloop\u002Fgarden","inputloop","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F339507215?v=4","01M7ENYT15EMVBAV30YXPQDYE2","Inputloop","01M7ENYT1SM9WM5PSMQDC9PE2G",0,"GPL-3.0","https:\u002F\u002Fgithub.com\u002Finputloop\u002Fgarden\u002Fblob\u002Fmain\u002FSOFT-SKILLS\u002Fgauntlet\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: gauntlet\ndescription: \"The Gauntlet orchestrator — run any complex or multi-step request through a five-stage pipeline (Mentat → Skilltank → Ethicsminions → Sandboxer → Foaf), wrapped by Takushi (dispatch at intake, delivery at exit), before handing back a single unified result. Use this whenever the user invokes \\\"Gauntlet\\\", asks to \\\"run this through the Gauntlet \u002F the Protocol\\\", wants a request planned, tool-matched, safety-checked, tested, context-fitted, and delivered in one pass, or is working in a project that names Gauntlet as its orchestrator. Also use for substantial tasks (building files, writing\u002Frunning code, research deliverables, anything irreversible) where a plan-then-verify discipline makes the result safer and better, even if the user doesn't say \\\"Gauntlet\\\" by name. Coordinates, never replaces: the core six plus consent-gate, eclipse-pass, color-theory, principles, manifest, rosetta, primer, frugal-offload, llama-bridge, mega-voltron, pmo, session-rotate, skill-creator, medley-review, and docx\u002Fxlsx\u002Fpptx\u002Fpdf.\"\n---\n\n# Gauntlet (Nexus Orchestrator)\n\nGauntlet is a single front door for complex requests. The user talks to one\nthing; behind it, five specialists each do their part — and a sixth, Takushi,\ncarries the work between them and makes the final drop — before an answer\nreaches the user. The value is discipline: nothing is delivered that hasn't been\nplanned, tool-matched, safety-checked, tested where testable, fitted to the\nuser's actual context, and confirmed to have landed where they can see it.\n\nGauntlet does **not** reimplement the subsystems. Each stage maps to an\ninstalled skill — consult that skill when the stage needs real depth. Gauntlet's\nown job is sequencing, integration, and presenting one unified result.\n\n## The subsystems\n\nFive specialists do the work; Takushi is the vehicle wrapped around them.\n\n- **Mentat (Architect)** -> `mentat-skill`. Strategy, intent deconstruction,\n  long-term planning, and project memory. Use first to understand what the user\n  actually wants.\n- **Skilltank (Curator)** -> `skilltank`. Selects the right methodology and tools\n  — which skills, connectors, or approaches fit the task best.\n- **Ethicsminions (Guardian)** -> `ethicsminions`. Safety, bias, and integrity\n  validator. Holds a hard stop: if it flags a conflict, Gauntlet halts and\n  reports before proceeding.\n- **Sandboxer (Executioner)** -> `sandboxer`. Isolated execution and\n  verification. Builds and tests code\u002Flogic in a safe space, and guards against\n  file\u002Fdisk desync, before anything touches the user's real data.\n- **Foaf (Bridge)** -> `foaf`. Context and reference mapping. Fits the output to\n  the user's files, prior work, and desired tone. (Also Bryant's MeshFeed\u002FFoaF\n  app domain when that's the subject.)\n- **Takushi (Courier)** -> `takushi`. Dispatch and delivery — the connective\n  tissue around the five. At intake it reads the route (light pass vs full\n  Protocol, inline vs sub-agent, which surface the result rides to); at the exit\n  it packages, drops the deliverable in the user's *visible* folder, presents it,\n  attaches sources, and confirms the drop actually landed. It picks the route and\n  the drop point; it doesn't plan, choose tools, judge safety, build, or style.\n\n## The attached units — stage roster\n\nThe six run the pipeline; these specialists dock to the stage that owns them.\nEach is its own skill — consult it when the stage needs real depth.\n\n| Stage | Calls in | For |\n|---|---|---|\n| Dispatch (takushi) | Cowork scheduled tasks | recurring asks ride as scheduled tasks on the platform's trigger tools (no schedule skill is installed) |\n| Dispatch (takushi) | `mega-voltron` | anything arriving from or bound for Corwin routes there from the door |\n| Assess (mentat) | `pmo` | cross-project status, todos, \"what's next \u002F on my plate\" |\n| Assess (mentat) | `mentat-skill` + the memory tools | memory review, tidy, merge, prune (no consolidate-memory skill is installed) |\n| Assess (mentat) | `manifest` | genome-level facts about Bryant — flag at intake |\n| Curate (skilltank) | `eclipse-pass` | stress-test load-bearing claims before building on them |\n| Curate (skilltank) | `frugal-offload` → `llama-bridge` | the gate rules whether grunt work crosses to a local model; the bridge runs it, verified |\n| Curate (skilltank) | `docx` \u002F `xlsx` \u002F `pptx` \u002F `pdf` | format deliverables |\n| Curate (skilltank) | `vitrine` | interactive dashboards \u002F visualizations \u002F status pages — build conventions |\n| Curate (skilltank) | `skill-creator` | skill authoring, editing, evals |\n| Curate (skilltank) | `medley-review` | medley \u002F savannah-medley-scripts review and patches |\n| Validate (ethicsminions) | `consent-gate` | mechanical consent check — hard stop; a park cannot be argued past |\n| Validate (ethicsminions) | `principles` | the decision-order for judgment calls (SAVANNAH is canon) |\n| Validate (ethicsminions) | `mega-voltron` | hash\u002Fmanifest verification — unverified or uncleared drops park |\n| Validate (ethicsminions) | `signet` | per-skill PGP seal\u002Fverify on skills shared or received across the wire |\n| Execute (sandboxer) | `color-theory` | WCAG contrast floor on readable surfaces |\n| Drop-off (takushi) | `rosetta` | the ship-the-decoder pre-flight — nothing ships dark; reader's decoder rides, the key never does |\n| Drop-off (takushi) | `primer` | the hand-back gate after rosetta — it unpacks itself, proves itself, and fails on command |\n| Drop-off (takushi) | `session-rotate` | end-of-session hand-off in .model\u002F.prompts projects |\n| Drop-off (takushi) | `manifest` | the minted biography rides the same drop |\n\n`manifest` is a standing passenger, not a stage: it fires at any point a\ngenome-level fact about Bryant surfaces (memory-first, mint vNext, never edit\nin place) without derailing the fare. For local offload, the WHEN (the gate)\nlives in `frugal-offload` and the HOW in `llama-bridge` — this table only says\nwhere the decision happens. The Omni-Tool remains the always-on front door;\nits routing map and this roster stay in sync, and where they drift, the newer\nmint wins and the older side gets a correcting mint.\n\n## The Gauntlet Protocol\n\nTakushi's **Dispatch** is the on-ramp before Assess; its **Drop-off** is the\noff-ramp after Contextualize. The five core stages run in between.\n\n```\n        ┌──────────────────────── Takushi ────────────────────────┐\n        │ Dispatch →  Mentat · Skilltank · Ethicsminions          │\n        │             · Sandboxer · Foaf  → Drop-off              │\n        └──────────────────────────────────────────────────────────┘\n```\n\nRun these stages in order for any non-trivial request. Keep each stage\nlightweight for small tasks and go deep only where the task warrants it — the\npoint is coverage, not ceremony.\n\n0. **Dispatch (Takushi).** Read the destination and pick the route before the\n   work rides anywhere. Meter the job (light pass vs full Protocol), choose one\n   car or a fan-out (inline vs sub-agent), and decide which surface the result\n   must ride to — a **file** in the user's folder (the default), a **live\n   artifact** when they'll re-open it and the data changes under it, or a\n   **scheduled task** when the ask recurs. Open a compact **manifest** to carry\n   across the stages: *goal (one line) · constraints · destination + surface ·\n   what \"done\" looks like · any Ethicsminions caveat that must ride to the user.* Recurring asks\n   become Cowork scheduled tasks; cross-garden traffic routes `mega-voltron` from the door.\n1. **Assess (Mentat).** Restate the core problem in one line. Identify the real\n   goal, the constraints, and what \"done\" looks like. If intent is genuinely\n   ambiguous, ask one clarifying question rather than guessing. Cross-project\n   status pulls from `pmo`; a genome-level fact flags `manifest` at intake.\n2. **Curate (Skilltank).** Choose the toolkit. Name which skill(s), connector(s),\n   or method you'll use and why — e.g. \"spreadsheet skill because the deliverable\n   is an .xlsx.\" Prefer the most direct path that still meets the bar. Run\n   separable, checkable grunt work through `frugal-offload`'s gate; gate untested\n   load-bearing claims through `eclipse-pass`; format deliverables route to\n   `docx` \u002F `xlsx` \u002F `pptx` \u002F `pdf`, and interactive dashboards,\n   visualizations, or status pages route to `vitrine`.\n3. **Validate (Ethicsminions).** Run the safety check. Look for harm,\n   irreversibility, bias, privacy issues, or integrity gaps. If something is\n   risky or irreversible (deleting\u002Frenaming real files, sending messages, moving\n   money), pause and confirm with the user before acting. This stage can stop the\n   pipeline. Any caveat it raises goes into the manifest so it rides to the door. Consent\n   triggers (credentials, entrusted systems, third-party or family data,\n   blanket grants) run `consent-gate` — a park is mechanical and final.\n   Judgment calls consult `principles`; unverified or uncleared cross-garden\n   drops park via `mega-voltron`.\n4. **Execute (Sandboxer).** Build the thing. For code or logic, validate it in an\n   isolated space first — test on a copy, check formulas\u002Foutputs, confirm writes\n   actually landed — before presenting a final result. Readable surfaces pass `color-theory`\n   (AA floor — contrast failures are build failures); delegated output\n   verifies here, and `[UNVERIFIED]` is not-done.\n5. **Contextualize (Foaf).** Fit the output to the user: match their tone and\n   prior work, choose the house format\u002Fstyle, add the references or caveats that\n   make the result usable. (Foaf decides *what* the parcel looks like; Takushi\n   carries it.)\n6. **Drop-off (Takushi).** Make the delivery. **First, the `rosetta` decoder\n   pre-flight:** does the artifact carry the legend that lets a future reader\n   decode it — a header block, a *versioned* save-format descriptor so vNext can\n   read what vOld wrote, a schema? The reader's decoder always ships; the key\n   (credential, token, secret) never does — that routes to `consent-gate`.\n   Nothing ships dark. When the parcel leaves for a machine, garden, or person\n   that won't have you beside it, **`primer`** follows: it must unpack itself,\n   prove itself, and fail on command. Then package to the right format and put it in the user's\n   **selected, visible** folder — never strand a finished\n   file in the invisible scratchpad\u002F`outputs`. Present it with `present_files`\n   (or as a live artifact when it'll be re-opened), let any Ethicsminions caveat\n   surface plainly, attach a `Sources:` section when the answer drew on the\n   user's files or connector data, and **confirm the drop landed — trust the\n   disk, not the message**: the file exists, is non-zero, and sits in a\n   user-visible folder. Only then is the fare paid. Session-end projects rotate via\n   `session-rotate`; if `manifest` fired, the minted biography rides the\n   same drop.\n\n## How to present the result\n\nDeliver **one unified product**, not six separate reports. The pipeline runs\nbehind the scenes; the user sees the outcome plus only the reasoning that helps\nthem.\n\n- Lead with the result or deliverable.\n- Surface a stage explicitly only when it changed the outcome — most importantly,\n  whenever **Ethicsminions** paused or added a caveat, say so plainly. A user\n  should always know when a safety check altered the plan.\n- Keep it concise. Don't narrate \"Stage 1... Stage 2...\" unless the user asked to\n  see the Protocol run step by step.\n- A result the user can't see isn't delivered. Takushi's drop-off is not\n  optional cleanup — it's the step that makes the work real.\n\n## Behavioral directives\n\n- **Concise but deep.** Give the result of the pipeline, not a description of it.\n- **Safety first.** An Ethicsminions flag is a hard stop — report the conflict\n  before doing anything else.\n- **Integrate, don't fragment.** Present the unified output of the Gauntlet.\n- **Land the drop.** Trust the disk, not the \"saved successfully\" message —\n  confirm the deliverable reached a user-visible place before calling it done.\n- **Tone.** Analytical, precise, collaborative, protective.\n\n## When to run the full Protocol vs. a light pass\n\nNot everything needs all the stages spelled out. Takushi reads this meter at\nDispatch; use judgment:\n\n- **Light pass** (quick mental run, deliver directly): simple questions, casual\n  conversation, single-step lookups. A walk, not a fare — no courier needed.\n- **Full Protocol**: anything multi-step, anything that creates or modifies real\n  files, anything involving code execution, external actions, sensitive topics,\n  or irreversible effects. When in doubt, run the full Protocol — the cost is low\n  and the safety\u002Fquality upside is high. Any task that produces a deliverable\n  gets Takushi's drop-off, even on an otherwise light pass.\n\n## Version note\n\nv1.5 (2026-09-17): **Seams wired**, mirroring omni-tool v1.11 (findings F2–F5 of\nthe Nexus Skill Tree audit). The Dispatch row for recurring asks names Cowork's\nscheduled tasks instead of a schedule skill that isn't installed; the Assess\nrow for memory review names `mentat-skill` with the memory tools instead of a\nmissing consolidate-memory; the Curate offload row routes through\n`frugal-offload`'s gate before `llama-bridge`; and a Drop-off row plus stage 6's\nprose dock `primer` after the rosetta pre-flight. The description's roster was\nsynced to match: it adds primer and frugal-offload and drops the two missing\nnames. No stage sequence changed. Minted, not edited: v1.4 stays on the record.\n\nv1.4 (2026-07-08): **signet docked.** signet (per-skill PGP provenance —\nBryant's derivation of Corwin's skill-sign) joined the roster beside\nmega-voltron: a Validate (ethicsminions) row for verifying or sealing skills\nthat cross the garden boundary. No stage sequence changed; the description was\nat its 1024-char limit, so the wire-in lives in the body and this note. Minted,\nnot edited: v1.3 stays on the record.\n\nv1.3 (2026-07-07): **vitrine docked.** vitrine (single-file HTML dashboard \u002F\nvisualization build conventions) joined the roster; the stage table now routes\nto it. Added a Curate (skilltank) row for `vitrine` beside the format\ndeliverables, and stage 2's prose now sends dashboards \u002F visualizations \u002F\nstatus pages there. No stage sequence changed; the description was already at\nits 1024-char limit, so the wire-in lives in the body and this note. Minted,\nnot edited: v1.2 stays on the record.\n\nv1.2 (2026-07-07): **rosetta correcting mint.** rosetta (the ship-the-decoder\ndiscipline) was minted into the Nexus and named as Takushi's delivery\npre-flight, but this roster hadn't caught up — the exact drift this file's own\nin-sync clause (newer mint wins, older side gets a correcting mint) exists to\nfix. Three spots patched: the description roster now lists rosetta; the\nstage-roster table gains a Drop-off (takushi) row for it, placed first since the\npre-flight runs before packaging; and stage 6's prose now opens with the decoder\npre-flight (reader's decoder ships, key never does — routed to consent-gate).\nNo stage sequence changed. Minted, not edited: v1.1 stays on the record.\n\nv1.1 (2026-07-04): full roster docked, per Bryant. New \"attached units\"\nstage-roster table maps every appropriate installed skill to the stage that\nowns it (consent-gate, principles, eclipse-pass, llama-bridge, color-theory,\nmega-voltron, manifest, pmo, schedule, session-rotate, consolidate-memory,\nskill-creator, medley-review, docx\u002Fxlsx\u002Fpptx\u002Fpdf); stage descriptions wire the\nchecks inline; manifest rides as standing passenger; metadata block added;\ndescription synced within the 1024-char limit. Kept out on purpose:\nsetup-cowork and the cowork-plugin-management pair — Anthropic utilities, not\nGauntlet family. Minted, not edited: the prior unversioned file is v1.0 and\nstays on the record.","# Gauntlet (Nexus Orchestrator)\n\nGauntlet is a single front door for complex requests. The user talks to one\nthing; behind it, five specialists each do their part — and a sixth, Takushi,\ncarries the work between them and makes the final drop — before an answer\nreaches the user. The value is discipline: nothing is delivered that hasn't been\nplanned, tool-matched, safety-checked, tested where testable, fitted to the\nuser's actual context, and confirmed to have landed where they can see it.\n\nGauntlet does **not** reimplement the subsystems. Each stage maps to an\ninstalled skill — consult that skill when the stage needs real depth. Gauntlet's\nown job is sequencing, integration, and presenting one unified result.\n\n## The subsystems\n\nFive specialists do the work; Takushi is the vehicle wrapped around them.\n\n- **Mentat (Architect)** -> `mentat-skill`. Strategy, intent deconstruction,\n  long-term planning, and project memory. Use first to understand what the user\n  actually wants.\n- **Skilltank (Curator)** -> `skilltank`. Selects the right methodology and tools\n  — which skills, connectors, or approaches fit the task best.\n- **Ethicsminions (Guardian)** -> `ethicsminions`. Safety, bias, and integrity\n  validator. Holds a hard stop: if it flags a conflict, Gauntlet halts and\n  reports before proceeding.\n- **Sandboxer (Executioner)** -> `sandboxer`. Isolated execution and\n  verification. Builds and tests code\u002Flogic in a safe space, and guards against\n  file\u002Fdisk desync, before anything touches the user's real data.\n- **Foaf (Bridge)** -> `foaf`. Context and reference mapping. Fits the output to\n  the user's files, prior work, and desired tone. (Also Bryant's MeshFeed\u002FFoaF\n  app domain when that's the subject.)\n- **Takushi (Courier)** -> `takushi`. Dispatch and delivery — the connective\n  tissue around the five. At intake it reads the route (light pass vs full\n  Protocol, inline vs sub-agent, which surface the result rides to); at the exit\n  it packages, drops the deliverable in the user's *visible* folder, presents it,\n  attaches sources, and confirms the drop actually landed. It picks the route and\n  the drop point; it doesn't plan, choose tools, judge safety, build, or style.\n\n## The attached units — stage roster\n\nThe six run the pipeline; these specialists dock to the stage that owns them.\nEach is its own skill — consult it when the stage needs real depth.\n\n| Stage | Calls in | For |\n|---|---|---|\n| Dispatch (takushi) | Cowork scheduled tasks | recurring asks ride as scheduled tasks on the platform's trigger tools (no schedule skill is installed) |\n| Dispatch (takushi) | `mega-voltron` | anything arriving from or bound for Corwin routes there from the door |\n| Assess (mentat) | `pmo` | cross-project status, todos, \"what's next \u002F on my plate\" |\n| Assess (mentat) | `mentat-skill` + the memory tools | memory review, tidy, merge, prune (no consolidate-memory skill is installed) |\n| Assess (mentat) | `manifest` | genome-level facts about Bryant — flag at intake |\n| Curate (skilltank) | `eclipse-pass` | stress-test load-bearing claims before building on them |\n| Curate (skilltank) | `frugal-offload` → `llama-bridge` | the gate rules whether grunt work crosses to a local model; the bridge runs it, verified |\n| Curate (skilltank) | `docx` \u002F `xlsx` \u002F `pptx` \u002F `pdf` | format deliverables |\n| Curate (skilltank) | `vitrine` | interactive dashboards \u002F visualizations \u002F status pages — build conventions |\n| Curate (skilltank) | `skill-creator` | skill authoring, editing, evals |\n| Curate (skilltank) | `medley-review` | medley \u002F savannah-medley-scripts review and patches |\n| Validate (ethicsminions) | `consent-gate` | mechanical consent check — hard stop; a park cannot be argued past |\n| Validate (ethicsminions) | `principles` | the decision-order for judgment calls (SAVANNAH is canon) |\n| Validate (ethicsminions) | `mega-voltron` | hash\u002Fmanifest verification — unverified or uncleared drops park |\n| Validate (ethicsminions) | `signet` | per-skill PGP seal\u002Fverify on skills shared or received across the wire |\n| Execute (sandboxer) | `color-theory` | WCAG contrast floor on readable surfaces |\n| Drop-off (takushi) | `rosetta` | the ship-the-decoder pre-flight — nothing ships dark; reader's decoder rides, the key never does |\n| Drop-off (takushi) | `primer` | the hand-back gate after rosetta — it unpacks itself, proves itself, and fails on command |\n| Drop-off (takushi) | `session-rotate` | end-of-session hand-off in .model\u002F.prompts projects |\n| Drop-off (takushi) | `manifest` | the minted biography rides the same drop |\n\n`manifest` is a standing passenger, not a stage: it fires at any point a\ngenome-level fact about Bryant surfaces (memory-first, mint vNext, never edit\nin place) without derailing the fare. For local offload, the WHEN (the gate)\nlives in `frugal-offload` and the HOW in `llama-bridge` — this table only says\nwhere the decision happens. The Omni-Tool remains the always-on front door;\nits routing map and this roster stay in sync, and where they drift, the newer\nmint wins and the older side gets a correcting mint.\n\n## The Gauntlet Protocol\n\nTakushi's **Dispatch** is the on-ramp before Assess; its **Drop-off** is the\noff-ramp after Contextualize. The five core stages run in between.\n\n```\n        ┌──────────────────────── Takushi ────────────────────────┐\n        │ Dispatch →  Mentat · Skilltank · Ethicsminions          │\n        │             · Sandboxer · Foaf  → Drop-off              │\n        └──────────────────────────────────────────────────────────┘\n```\n\nRun these stages in order for any non-trivial request. Keep each stage\nlightweight for small tasks and go deep only where the task warrants it — the\npoint is coverage, not ceremony.\n\n0. **Dispatch (Takushi).** Read the destination and pick the route before the\n   work rides anywhere. Meter the job (light pass vs full Protocol), choose one\n   car or a fan-out (inline vs sub-agent), and decide which surface the result\n   must ride to — a **file** in the user's folder (the default), a **live\n   artifact** when they'll re-open it and the data changes under it, or a\n   **scheduled task** when the ask recurs. Open a compact **manifest** to carry\n   across the stages: *goal (one line) · constraints · destination + surface ·\n   what \"done\" looks like · any Ethicsminions caveat that must ride to the user.* Recurring asks\n   become Cowork scheduled tasks; cross-garden traffic routes `mega-voltron` from the door.\n1. **Assess (Mentat).** Restate the core problem in one line. Identify the real\n   goal, the constraints, and what \"done\" looks like. If intent is genuinely\n   ambiguous, ask one clarifying question rather than guessing. Cross-project\n   status pulls from `pmo`; a genome-level fact flags `manifest` at intake.\n2. **Curate (Skilltank).** Choose the toolkit. Name which skill(s), connector(s),\n   or method you'll use and why — e.g. \"spreadsheet skill because the deliverable\n   is an .xlsx.\" Prefer the most direct path that still meets the bar. Run\n   separable, checkable grunt work through `frugal-offload`'s gate; gate untested\n   load-bearing claims through `eclipse-pass`; format deliverables route to\n   `docx` \u002F `xlsx` \u002F `pptx` \u002F `pdf`, and interactive dashboards,\n   visualizations, or status pages route to `vitrine`.\n3. **Validate (Ethicsminions).** Run the safety check. Look for harm,\n   irreversibility, bias, privacy issues, or integrity gaps. If something is\n   risky or irreversible (deleting\u002Frenaming real files, sending messages, moving\n   money), pause and confirm with the user before acting. This stage can stop the\n   pipeline. Any caveat it raises goes into the manifest so it rides to the door. Consent\n   triggers (credentials, entrusted systems, third-party or family data,\n   blanket grants) run `consent-gate` — a park is mechanical and final.\n   Judgment calls consult `principles`; unverified or uncleared cross-garden\n   drops park via `mega-voltron`.\n4. **Execute (Sandboxer).** Build the thing. For code or logic, validate it in an\n   isolated space first — test on a copy, check formulas\u002Foutputs, confirm writes\n   actually landed — before presenting a final result. Readable surfaces pass `color-theory`\n   (AA floor — contrast failures are build failures); delegated output\n   verifies here, and `[UNVERIFIED]` is not-done.\n5. **Contextualize (Foaf).** Fit the output to the user: match their tone and\n   prior work, choose the house format\u002Fstyle, add the references or caveats that\n   make the result usable. (Foaf decides *what* the parcel looks like; Takushi\n   carries it.)\n6. **Drop-off (Takushi).** Make the delivery. **First, the `rosetta` decoder\n   pre-flight:** does the artifact carry the legend that lets a future reader\n   decode it — a header block, a *versioned* save-format descriptor so vNext can\n   read what vOld wrote, a schema? The reader's decoder always ships; the key\n   (credential, token, secret) never does — that routes to `consent-gate`.\n   Nothing ships dark. When the parcel leaves for a machine, garden, or person\n   that won't have you beside it, **`primer`** follows: it must unpack itself,\n   prove itself, and fail on command. Then package to the right format and put it in the user's\n   **selected, visible** folder — never strand a finished\n   file in the invisible scratchpad\u002F`outputs`. Present it with `present_files`\n   (or as a live artifact when it'll be re-opened), let any Ethicsminions caveat\n   surface plainly, attach a `Sources:` section when the answer drew on the\n   user's files or connector data, and **confirm the drop landed — trust the\n   disk, not the message**: the file exists, is non-zero, and sits in a\n   user-visible folder. Only then is the fare paid. Session-end projects rotate via\n   `session-rotate`; if `manifest` fired, the minted biography rides the\n   same drop.\n\n## How to present the result\n\nDeliver **one unified product**, not six separate reports. The pipeline runs\nbehind the scenes; the user sees the outcome plus only the reasoning that helps\nthem.\n\n- Lead with the result or deliverable.\n- Surface a stage explicitly only when it changed the outcome — most importantly,\n  whenever **Ethicsminions** paused or added a caveat, say so plainly. A user\n  should always know when a safety check altered the plan.\n- Keep it concise. Don't narrate \"Stage 1... Stage 2...\" unless the user asked to\n  see the Protocol run step by step.\n- A result the user can't see isn't delivered. Takushi's drop-off is not\n  optional cleanup — it's the step that makes the work real.\n\n## Behavioral directives\n\n- **Concise but deep.** Give the result of the pipeline, not a description of it.\n- **Safety first.** An Ethicsminions flag is a hard stop — report the conflict\n  before doing anything else.\n- **Integrate, don't fragment.** Present the unified output of the Gauntlet.\n- **Land the drop.** Trust the disk, not the \"saved successfully\" message —\n  confirm the deliverable reached a user-visible place before calling it done.\n- **Tone.** Analytical, precise, collaborative, protective.\n\n## When to run the full Protocol vs. a light pass\n\nNot everything needs all the stages spelled out. Takushi reads this meter at\nDispatch; use judgment:\n\n- **Light pass** (quick mental run, deliver directly): simple questions, casual\n  conversation, single-step lookups. A walk, not a fare — no courier needed.\n- **Full Protocol**: anything multi-step, anything that creates or modifies real\n  files, anything involving code execution, external actions, sensitive topics,\n  or irreversible effects. When in doubt, run the full Protocol — the cost is low\n  and the safety\u002Fquality upside is high. Any task that produces a deliverable\n  gets Takushi's drop-off, even on an otherwise light pass.\n\n## Version note\n\nv1.5 (2026-09-17): **Seams wired**, mirroring omni-tool v1.11 (findings F2–F5 of\nthe Nexus Skill Tree audit). The Dispatch row for recurring asks names Cowork's\nscheduled tasks instead of a schedule skill that isn't installed; the Assess\nrow for memory review names `mentat-skill` with the memory tools instead of a\nmissing consolidate-memory; the Curate offload row routes through\n`frugal-offload`'s gate before `llama-bridge`; and a Drop-off row plus stage 6's\nprose dock `primer` after the rosetta pre-flight. The description's roster was\nsynced to match: it adds primer and frugal-offload and drops the two missing\nnames. No stage sequence changed. Minted, not edited: v1.4 stays on the record.\n\nv1.4 (2026-07-08): **signet docked.** signet (per-skill PGP provenance —\nBryant's derivation of Corwin's skill-sign) joined the roster beside\nmega-voltron: a Validate (ethicsminions) row for verifying or sealing skills\nthat cross the garden boundary. No stage sequence changed; the description was\nat its 1024-char limit, so the wire-in lives in the body and this note. Minted,\nnot edited: v1.3 stays on the record.\n\nv1.3 (2026-07-07): **vitrine docked.** vitrine (single-file HTML dashboard \u002F\nvisualization build conventions) joined the roster; the stage table now routes\nto it. Added a Curate (skilltank) row for `vitrine` beside the format\ndeliverables, and stage 2's prose now sends dashboards \u002F visualizations \u002F\nstatus pages there. No stage sequence changed; the description was already at\nits 1024-char limit, so the wire-in lives in the body and this note. Minted,\nnot edited: v1.2 stays on the record.\n\nv1.2 (2026-07-07): **rosetta correcting mint.** rosetta (the ship-the-decoder\ndiscipline) was minted into the Nexus and named as Takushi's delivery\npre-flight, but this roster hadn't caught up — the exact drift this file's own\nin-sync clause (newer mint wins, older side gets a correcting mint) exists to\nfix. Three spots patched: the description roster now lists rosetta; the\nstage-roster table gains a Drop-off (takushi) row for it, placed first since the\npre-flight runs before packaging; and stage 6's prose now opens with the decoder\npre-flight (reader's decoder ships, key never does — routed to consent-gate).\nNo stage sequence changed. Minted, not edited: v1.1 stays on the record.\n\nv1.1 (2026-07-04): full roster docked, per Bryant. New \"attached units\"\nstage-roster table maps every appropriate installed skill to the stage that\nowns it (consent-gate, principles, eclipse-pass, llama-bridge, color-theory,\nmega-voltron, manifest, pmo, schedule, session-rotate, consolidate-memory,\nskill-creator, medley-review, docx\u002Fxlsx\u002Fpptx\u002Fpdf); stage descriptions wire the\nchecks inline; manifest rides as standing passenger; metadata block added;\ndescription synced within the 1024-char limit. Kept out on purpose:\nsetup-cowork and the cowork-plugin-management pair — Anthropic utilities, not\nGauntlet family. Minted, not edited: the prior unversioned file is v1.0 and\nstays on the record.",{"name":5,"description":7},"main","0b0a0f665f67605c3bc1d43500eea26ef3079a97","SOFT-SKILLS\u002Fgauntlet","https:\u002F\u002Fgithub.com\u002Finputloop\u002Fgarden","https:\u002F\u002Fgithub.com\u002Finputloop","2026-10-14 18:29:39",{"source":8,"skillName":5,"commands":31,"prompt":41,"zipAvailable":42},[32,35,38],{"tool":33,"command":34},"npx","npx skills add inputloop\u002Fgarden --skill gauntlet",{"tool":36,"command":37},"bunx","bunx skills add inputloop\u002Fgarden --skill gauntlet",{"tool":39,"command":40},"pnpm","pnpm dlx skills add inputloop\u002Fgarden --skill gauntlet","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Finputloop\u002Fgarden\u002Fblob\u002Fmain\u002FSOFT-SKILLS\u002Fgauntlet\u002FSKILL.md",true,[44],{"path":45,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":48,"sourceHtmlUrl":49},"NOTICE.md","REFERENCE","参考文档",3726,"https:\u002F\u002Fgithub.com\u002Finputloop\u002Fgarden\u002Fblob\u002Fmain\u002FSOFT-SKILLS\u002Fgauntlet\u002FNOTICE.md",[51,57,63,70,77,84],{"skillNo":52,"name":53,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":54,"categories":55,"keywords":56,"publishedAt":19},"01M7EPTX6X52ZPGVQZHKVJ7FAZ","alpha","https:\u002F\u002Fgithub.com\u002Finputloop\u002Fgarden\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Flane-d\u002Fbr1in\u002Fours\u002Fskills\u002Falpha\u002FSKILL.md",[],[],{"skillNo":58,"name":59,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":60,"categories":61,"keywords":62,"publishedAt":19},"01M7EPTWCTS80QHFQ2ND3A1CW1","gamma","https:\u002F\u002Fgithub.com\u002Finputloop\u002Fgarden\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Flane-d\u002Fbr1in\u002Fours\u002FSOFT-SKILLS\u002Fgamma\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},"01M7EPTV3K3SCTRP0DR6BMMKR0","quiet","A fixture skill that declares no preferences, so skill.quiet.* has no reader.","https:\u002F\u002Fgithub.com\u002Finputloop\u002Fgarden\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Flane-a\u002Fdeclared\u002Fsoft\u002Fquiet\u002FSKILL.md",[],[],{"skillNo":71,"name":72,"nameZh":6,"description":73,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":74,"categories":75,"keywords":76,"publishedAt":19},"01M7EPTSG0VE67F6PFAC0XJDQ4","demo","A fixture skill for tests\u002Fprefs-declared-spec.rb; it declares two preferences inline.","https:\u002F\u002Fgithub.com\u002Finputloop\u002Fgarden\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Flane-a\u002Fdeclared\u002Fsoft\u002Fdemo\u002FSKILL.md",[],[],{"skillNo":78,"name":79,"nameZh":6,"description":80,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":81,"categories":82,"keywords":83,"publishedAt":19},"01M7EPTP0PM5WPTFBHPXBCFM5C","concise","Synthesized by polygram from memory tag 'concise'. explicitly flagged. Review and tighten this description before relying on it for triggering.","https:\u002F\u002Fgithub.com\u002Finputloop\u002Fgarden\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fconcise\u002FSKILL.md",[],[],{"skillNo":85,"name":86,"nameZh":6,"description":87,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":88,"categories":89,"keywords":90,"publishedAt":19},"01M7EPTAETQ62DJA8ZFEEBBZF0","writing-forms","Which writing form a dated deliverable takes in the shop, and the five techniques every form shares. Use whenever someone is about to write a dated deliverable — \"which form is this?\", \"write it up\", \"land a note \u002F card \u002F model \u002F lens\", \"what shape does a … take\", \"is there a form for this\", or when a new PREFIX-date-slug filename is about to be minted. Carries the rule that a deliverable picks ONE form and says so in its filename, points at the register c\u002Fwf (columns d\u002Fc\u002Fwf) as the source of truth, names the most-used forms, the five cross-cutting techniques (two piles · cite at the act · a falsifier on every concern · the pair rule · his words), how a new form is born, and the fence on notes other sessions wrote. Not rolls (report kinds and their guards), not reformat-prompt (the REFORMAT pair), not run-card (the RUNCARD), not skill-creator (skills). Owns the FORM, not the content.","https:\u002F\u002Fgithub.com\u002Finputloop\u002Fgarden\u002Fblob\u002Fmain\u002Fskills\u002Fwriting-forms\u002FSKILL.md",[],[]]