[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7APXATF7KTBAK8SRC88CD26":3,"skill-same:01M7APXATF7KTBAK8SRC88CD26":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":27,"commitSha":28,"skillDirectory":29,"forksCount":30,"repositoryHtmlUrl":31,"repositoryHomepageUrl":6,"ownerHtmlUrl":32,"ownerBlogUrl":33,"ownerDescription":34,"updatedAt":35,"install":36,"resources":49},"01M7APXATF7KTBAK8SRC88CD26","gilgamesh-grind",null,"Phase Grind externally-verified work loop — governs sandbox-first mutation within authority, the named-oracle gate after every meaningful change, the loop budget (time\u002Fturns\u002Ftokens), and the mechanical stopping policy over exactly {continue, recover, escalate, terminate}. Use after Lock freezes the plan; load when Gilgamesh enters the work loop. Do NOT use during Gauge\u002FInventory\u002FLock\u002FAttest.","Rynaro\u002Fmagicite","Rynaro","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F5271543?v=4","01M70CFYFMB316W23THVQAQ42M","Henrique A. Lavezzo","01M7APN2KCVWBNY727T57KR2ZJ",2,"Apache-2.0","https:\u002F\u002Fgithub.com\u002FRynaro\u002Fmagicite\u002Fblob\u002Fmain\u002F.eidolons\u002Fgilgamesh\u002Fskills\u002Fgrind\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: gilgamesh-grind\ndescription: Phase Grind externally-verified work loop — governs sandbox-first mutation within authority, the named-oracle gate after every meaningful change, the loop budget (time\u002Fturns\u002Ftokens), and the mechanical stopping policy over exactly {continue, recover, escalate, terminate}. Use after Lock freezes the plan; load when Gilgamesh enters the work loop. Do NOT use during Gauge\u002FInventory\u002FLock\u002FAttest.\nmetadata:\n  methodology: Gilgamesh\n  phase: Grind\n---\n\n# Grind Skill — Gilgamesh (Phase Grind loop)\n\n## When to use\n\nLoad once the plan is locked (Phase L complete). Governs the\nmutate-in-sandbox → run-oracle → decide-stop loop. Never invoked in Gauge, Inventory,\nLock, or Attest.\n\nGrind is where the work happens, and where the Excalipoor rule bites hardest: a\nmutation is not believed until an **external oracle** says green.\n\n---\n\n## The action substrate (small and composable)\n\nGrind uses a handful of semantically-shaped actions (digest §2 — a small substrate\nbeats a large tool catalog):\n\n1. **read** — just-in-time gather via atlas-aci, within `authority.read`.\n2. **edit-in-sandbox** — apply a mutation via the nexus harness applier\n   `eidolons sandbox apply --proposal \u003Cp> --root \u003Cscratch>`. NEVER the real tree.\n3. **run-oracle** — run the deliverable's NAMED external verifier in the sandbox\n   (`eidolons sandbox run` \u002F `eidolons sandbox loop`).\n\nNothing else mutates state. The sandbox is the only write zone (`authority.write =\nsandbox`); the parent commits.\n\n---\n\n## Externalized verification (the gate)\n\nAfter **every meaningful mutation**, run the deliverable's NAMED external oracle. The\noracle is one of:\n\n| Class | Green signal |\n|---|---|\n| **test** | test runner exits 0 |\n| **typecheck** | typechecker exits 0 |\n| **lint** | linter exits 0 |\n| **compile** | compiler exits 0 |\n| **diff** | produced diff matches the expected shape |\n| **schema-validate** | validator accepts the artefact |\n| **parser** | parser accepts \u002F round-trips the artefact |\n| **env-feedback** | the environment reports the intended effect |\n\nSuccess is silent (record the green outcome, move on); failure is verbose (capture the\n**full** output into `TaskState.verification_outcomes`, keep it in context, adjust).\nNever self-critique, never LLM-as-judge. One green external signal per deliverable is\nthe pre-completion gate; no result reaches PROPOSE without it.\n\n**An oracle's output is a `VERIFY-\u003Cname>` result, never an evidence anchor.** A\ncommand's stdout\u002Fstderr — including anything written to an ephemeral path like\n`\u002Ftmp\u002F*` — proves a claim at Attest time via the `VERIFY-\u003Cname>` line, not via a\n`path:line` anchor. `EVIDENCE-\u003Ckey>`\u002F`PROPOSAL-TARGET` anchors resolve only to a\ncommitted repo file (the source or script the fact derives from); a command-output\npath is never a legal anchor target (`skills\u002Fattest\u002FSKILL.md` — repo-only anchor rule).\n\n### Route around tool-surface gaps — never route around the oracle itself (verify-routing ladder)\n\nYour tool allowlist is scoped narrower than the repo's full verification surface by\ndesign (no generic `Bash(*)`). If the named oracle's most direct invocation form\nfalls outside it, that is **not** grounds to skip the oracle or to substitute your\nown reasoning for its exit code. This is the same ladder `skills\u002Fattest\u002FSKILL.md` Step 1a\napplies to required report lines:\n\n1. **Direct.** If the verification command is directly allowed by your tool list\n   (`make`, `bats`, `shellcheck`, `pytest`, `go test`, `jest`, `rspec`, `shasum`,\n   `wc`) — run it directly.\n2. **Delegated sandbox.** Otherwise route it through `eidolons sandbox run\n   --allow-unsafe-host -- \u003Ccmd>` (trusted repo verifier scripts; the sandbox\n   captures pass\u002Ffail) or `eidolons sandbox loop` for the mutate→verify cycle. It\n   stands as equivalent verification, not a downgraded one.\n3. **Blocked.** Only if both rungs are unavailable is the check genuinely blocked.\n   Record that plainly — the specific denied invocation, as `fail` + blocker —\n   rather than hand-deriving a pass from manual inspection. A manual walkthrough is\n   not an external oracle, however careful.\n\nNever skip a rung, never omit the line, never hand-derive a result you could\nexecute. A blocked-and-reported check is honest; a manually-reasoned \"pass\"\nstanding in for an oracle's exit code is exactly the Excalipoor failure mode this\nmethodology exists to prevent.\n\n---\n\n## The loop budget\n\nEvery mission carries `budget{time, turns, tokens}` frozen at Lock. On each iteration,\nincrement `spent` and check every dimension. Exhausting **any** dimension forces\n`terminate` (below) — Gilgamesh Attests what is proven, never grinds past the budget.\n\n---\n\n## The stopping policy (mechanical — exactly four states)\n\nEach iteration resolves to **exactly one** state. The set is closed; there is no fifth\noption and no \"just one more\" beyond a wall.\n\n| State | Mechanical transition condition | Next action |\n|---|---|---|\n| **continue** | Last oracle green (or first attempt) **and** no budget dimension exhausted. | Run the next planned step. |\n| **recover** | An oracle went red **and** `consecutive_failures \u003C 3` **and** budget remains. | Adjust the mutation (tighter anchor \u002F different edit form \u002F re-gather) and retry — bounded. |\n| **escalate** | `consecutive_failures == 3`, **or** an authority\u002Fboundary wall (needs a denied capability, or a separable cross-boundary sub-mission). | Stop; carry the last oracle output + a `handoff-request` into Attest → ESCALATE. |\n| **terminate** | **Success:** every deliverable green under its oracle; **or** any budget dimension (time\u002Fturns\u002Ftokens) exhausted. | Stop; go to Attest with what is proven. |\n\n### Counters (never reset within a mission)\n\n- `consecutive_failures` — reset to 0 only on a green oracle.\n- `total_failures` — monotonic.\n- A **per-target loop** (the same artefact failing 3×), an applier error, or a timeout\n  each counts as a failure toward both counters.\n\n---\n\n## Worked iteration\n\n```\n1. state = continue\n2. edit-in-sandbox(step)          # authority.write = sandbox\n3. result = run-oracle(step.oracle)\n4. if result == green:\n     consecutive_failures = 0\n     if all deliverables green: state = terminate (success)\n     else: state = continue\n   else:\n     consecutive_failures += 1; total_failures += 1\n     if consecutive_failures == 3: state = escalate\n     elif budget_exhausted(): state = terminate (budget)\n     else: state = recover\n5. record result into TaskState.verification_outcomes\n6. loop on continue\u002Frecover; exit on escalate\u002Fterminate\n```\n\n---\n\n## Notes\n\n- **Sandbox-first, always.** The only mutation path is `eidolons sandbox apply`. A\n  direct write to the real tree violates the `write` authority row.\n- **Keep failure output in context.** Prior red output informs the next mutation; do\n  not discard it between iterations.\n- **Escalate loudly.** When `escalate` fires, include the last oracle output in the\n  ESCALATE body and attach a `handoff-request` so the orchestrator can route without\n  re-running.\n- **Trust the harness.** `eidolons sandbox apply` and the oracles are real and\n  installed — invoke them. An apply error means *retry with a tighter anchor*, not\n  *abandon a grantable mission*.\n\n---\n\n*Grind Skill — Gilgamesh Phase Grind: sandbox-first, external-oracle-gated, budget-bounded*\n","# Grind Skill — Gilgamesh (Phase Grind loop)\n\n## When to use\n\nLoad once the plan is locked (Phase L complete). Governs the\nmutate-in-sandbox → run-oracle → decide-stop loop. Never invoked in Gauge, Inventory,\nLock, or Attest.\n\nGrind is where the work happens, and where the Excalipoor rule bites hardest: a\nmutation is not believed until an **external oracle** says green.\n\n---\n\n## The action substrate (small and composable)\n\nGrind uses a handful of semantically-shaped actions (digest §2 — a small substrate\nbeats a large tool catalog):\n\n1. **read** — just-in-time gather via atlas-aci, within `authority.read`.\n2. **edit-in-sandbox** — apply a mutation via the nexus harness applier\n   `eidolons sandbox apply --proposal \u003Cp> --root \u003Cscratch>`. NEVER the real tree.\n3. **run-oracle** — run the deliverable's NAMED external verifier in the sandbox\n   (`eidolons sandbox run` \u002F `eidolons sandbox loop`).\n\nNothing else mutates state. The sandbox is the only write zone (`authority.write =\nsandbox`); the parent commits.\n\n---\n\n## Externalized verification (the gate)\n\nAfter **every meaningful mutation**, run the deliverable's NAMED external oracle. The\noracle is one of:\n\n| Class | Green signal |\n|---|---|\n| **test** | test runner exits 0 |\n| **typecheck** | typechecker exits 0 |\n| **lint** | linter exits 0 |\n| **compile** | compiler exits 0 |\n| **diff** | produced diff matches the expected shape |\n| **schema-validate** | validator accepts the artefact |\n| **parser** | parser accepts \u002F round-trips the artefact |\n| **env-feedback** | the environment reports the intended effect |\n\nSuccess is silent (record the green outcome, move on); failure is verbose (capture the\n**full** output into `TaskState.verification_outcomes`, keep it in context, adjust).\nNever self-critique, never LLM-as-judge. One green external signal per deliverable is\nthe pre-completion gate; no result reaches PROPOSE without it.\n\n**An oracle's output is a `VERIFY-\u003Cname>` result, never an evidence anchor.** A\ncommand's stdout\u002Fstderr — including anything written to an ephemeral path like\n`\u002Ftmp\u002F*` — proves a claim at Attest time via the `VERIFY-\u003Cname>` line, not via a\n`path:line` anchor. `EVIDENCE-\u003Ckey>`\u002F`PROPOSAL-TARGET` anchors resolve only to a\ncommitted repo file (the source or script the fact derives from); a command-output\npath is never a legal anchor target (`skills\u002Fattest\u002FSKILL.md` — repo-only anchor rule).\n\n### Route around tool-surface gaps — never route around the oracle itself (verify-routing ladder)\n\nYour tool allowlist is scoped narrower than the repo's full verification surface by\ndesign (no generic `Bash(*)`). If the named oracle's most direct invocation form\nfalls outside it, that is **not** grounds to skip the oracle or to substitute your\nown reasoning for its exit code. This is the same ladder `skills\u002Fattest\u002FSKILL.md` Step 1a\napplies to required report lines:\n\n1. **Direct.** If the verification command is directly allowed by your tool list\n   (`make`, `bats`, `shellcheck`, `pytest`, `go test`, `jest`, `rspec`, `shasum`,\n   `wc`) — run it directly.\n2. **Delegated sandbox.** Otherwise route it through `eidolons sandbox run\n   --allow-unsafe-host -- \u003Ccmd>` (trusted repo verifier scripts; the sandbox\n   captures pass\u002Ffail) or `eidolons sandbox loop` for the mutate→verify cycle. It\n   stands as equivalent verification, not a downgraded one.\n3. **Blocked.** Only if both rungs are unavailable is the check genuinely blocked.\n   Record that plainly — the specific denied invocation, as `fail` + blocker —\n   rather than hand-deriving a pass from manual inspection. A manual walkthrough is\n   not an external oracle, however careful.\n\nNever skip a rung, never omit the line, never hand-derive a result you could\nexecute. A blocked-and-reported check is honest; a manually-reasoned \"pass\"\nstanding in for an oracle's exit code is exactly the Excalipoor failure mode this\nmethodology exists to prevent.\n\n---\n\n## The loop budget\n\nEvery mission carries `budget{time, turns, tokens}` frozen at Lock. On each iteration,\nincrement `spent` and check every dimension. Exhausting **any** dimension forces\n`terminate` (below) — Gilgamesh Attests what is proven, never grinds past the budget.\n\n---\n\n## The stopping policy (mechanical — exactly four states)\n\nEach iteration resolves to **exactly one** state. The set is closed; there is no fifth\noption and no \"just one more\" beyond a wall.\n\n| State | Mechanical transition condition | Next action |\n|---|---|---|\n| **continue** | Last oracle green (or first attempt) **and** no budget dimension exhausted. | Run the next planned step. |\n| **recover** | An oracle went red **and** `consecutive_failures \u003C 3` **and** budget remains. | Adjust the mutation (tighter anchor \u002F different edit form \u002F re-gather) and retry — bounded. |\n| **escalate** | `consecutive_failures == 3`, **or** an authority\u002Fboundary wall (needs a denied capability, or a separable cross-boundary sub-mission). | Stop; carry the last oracle output + a `handoff-request` into Attest → ESCALATE. |\n| **terminate** | **Success:** every deliverable green under its oracle; **or** any budget dimension (time\u002Fturns\u002Ftokens) exhausted. | Stop; go to Attest with what is proven. |\n\n### Counters (never reset within a mission)\n\n- `consecutive_failures` — reset to 0 only on a green oracle.\n- `total_failures` — monotonic.\n- A **per-target loop** (the same artefact failing 3×), an applier error, or a timeout\n  each counts as a failure toward both counters.\n\n---\n\n## Worked iteration\n\n```\n1. state = continue\n2. edit-in-sandbox(step)          # authority.write = sandbox\n3. result = run-oracle(step.oracle)\n4. if result == green:\n     consecutive_failures = 0\n     if all deliverables green: state = terminate (success)\n     else: state = continue\n   else:\n     consecutive_failures += 1; total_failures += 1\n     if consecutive_failures == 3: state = escalate\n     elif budget_exhausted(): state = terminate (budget)\n     else: state = recover\n5. record result into TaskState.verification_outcomes\n6. loop on continue\u002Frecover; exit on escalate\u002Fterminate\n```\n\n---\n\n## Notes\n\n- **Sandbox-first, always.** The only mutation path is `eidolons sandbox apply`. A\n  direct write to the real tree violates the `write` authority row.\n- **Keep failure output in context.** Prior red output informs the next mutation; do\n  not discard it between iterations.\n- **Escalate loudly.** When `escalate` fires, include the last oracle output in the\n  ESCALATE body and attach a `handoff-request` so the orchestrator can route without\n  re-running.\n- **Trust the harness.** `eidolons sandbox apply` and the oracles are real and\n  installed — invoke them. An apply error means *retry with a tighter anchor*, not\n  *abandon a grantable mission*.\n\n---\n\n*Grind Skill — Gilgamesh Phase Grind: sandbox-first, external-oracle-gated, budget-bounded*\n",{"name":5,"metadata":24,"description":7},{"phase":25,"methodology":26},"Grind","Gilgamesh","main","38c3a82583abf9806de764f89ac0099ad690933a",".eidolons\u002Fgilgamesh\u002Fskills\u002Fgrind",0,"https:\u002F\u002Fgithub.com\u002FRynaro\u002Fmagicite","https:\u002F\u002Fgithub.com\u002FRynaro","https:\u002F\u002Fhlavezzo.me","A Code Alchemist with a passion for weaving tales and inscribing transmutation circles in pure code.","2026-10-14 02:35:37",{"source":8,"skillName":5,"commands":37,"prompt":47,"zipAvailable":48},[38,41,44],{"tool":39,"command":40},"npx","npx skills add Rynaro\u002Fmagicite --skill gilgamesh-grind",{"tool":42,"command":43},"bunx","bunx skills add Rynaro\u002Fmagicite --skill gilgamesh-grind",{"tool":45,"command":46},"pnpm","pnpm dlx skills add Rynaro\u002Fmagicite --skill gilgamesh-grind","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FRynaro\u002Fmagicite\u002Fblob\u002Fmain\u002F.eidolons\u002Fgilgamesh\u002Fskills\u002Fgrind\u002FSKILL.md",true,[],[51,58,65,72,79,86],{"skillNo":52,"name":53,"nameZh":6,"description":54,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":55,"categories":56,"keywords":57,"publishedAt":19},"01M7AQ4GHJQVV8TJBJ8T1V7ZQ9","wine-dxvk-cache-clear","Clear a stale DXVK shader cache that is causing graphical corruption in a\nWine or Proton game. Use when: a game shows corrupted textures or shader\nflicker that started after a driver or DXVK version change.","https:\u002F\u002Fgithub.com\u002FRynaro\u002Fmagicite\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Ftoy-registry\u002Fskills\u002Fwine-dxvk-cache-clear\u002FSKILL.md",[],[],{"skillNo":59,"name":60,"nameZh":6,"description":61,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":62,"categories":63,"keywords":64,"publishedAt":19},"01M7AQ4CY1NHSMY1HCZ91387YY","steam-download-region-fix","Fix a Steam client stuck at very low download speed by changing its\ndownload region. Use when: Steam downloads are far below the local\ninternet speed for every game, not just one.","https:\u002F\u002Fgithub.com\u002FRynaro\u002Fmagicite\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Ftoy-registry\u002Fskills\u002Fsteam-download-region-fix\u002FSKILL.md",[],[],{"skillNo":66,"name":67,"nameZh":6,"description":68,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":69,"categories":70,"keywords":71,"publishedAt":19},"01M7AQ49P7Y0CDRSYMPVABX80K","proton-battleye-eac-toggle","Toggle BattlEye or Easy Anti-Cheat runtime support for a Steam Proton game.\nUse when: a multiplayer game refuses to launch with an anti-cheat error under Proton.","https:\u002F\u002Fgithub.com\u002FRynaro\u002Fmagicite\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Ftoy-registry\u002Fskills\u002Fproton-battleye-eac-toggle\u002FSKILL.md",[],[],{"skillNo":73,"name":74,"nameZh":6,"description":75,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":76,"categories":77,"keywords":78,"publishedAt":19},"01M7AQ45326FAP3X4Q86FK1X4N","vivi-verify-incoming","Load when reading any upstream artefact handed off to Vivi that carries a sibling .envelope.json. BLOCKING per ECL §6.2.2: the orchestrator MUST have verified the envelope's SHA-256 (eidolons verify-envelope --block \u002F eidolons run --verify) and recorded a verify_pass before dispatch. If no verify_pass exists for the message_id, or a verify_fail is present, REFUSE to process the payload and hand back to the orchestrator. Symmetric receiver gate — every Eidolon enforces it identically.","https:\u002F\u002Fgithub.com\u002FRynaro\u002Fmagicite\u002Fblob\u002Fmain\u002F.eidolons\u002Fvivi\u002Fskills\u002Fverify-incoming\u002FSKILL.md",[],[],{"skillNo":80,"name":81,"nameZh":6,"description":82,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":83,"categories":84,"keywords":85,"publishedAt":19},"01M7AQ402861P0P3C2TEGFGMA1","vivi-parallel-tracks","Load ONLY under TRANCE authorization when the Plan phase yields N independent, non-overlapping implementation tracks (disjoint file sets). Operationalizes the TRANCE G4 form — parallel multi-track implementation in mandatory git-worktree isolation, max 5 clean-context tracks, a per-track verifier cascade, a non-fungible per-track ≤3 reflection budget, explicit stop conditions, and a single-threaded merge\u002Faggregation step. Absent TRANCE gating, Vivi runs exactly as today: single-track A→P→I→V→Δ\u002FR.","https:\u002F\u002Fgithub.com\u002FRynaro\u002Fmagicite\u002Fblob\u002Fmain\u002F.eidolons\u002Fvivi\u002Fskills\u002Fparallel-tracks\u002FSKILL.md",[],[],{"skillNo":87,"name":88,"nameZh":6,"description":89,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":90,"categories":91,"keywords":92,"publishedAt":19},"01M7AQ3RH9BZF9ZMTC7VG5P7PB","vivi-methodology","Full Vivi cycle reference — A(nalyze) → P(lan) → I(mplement) → V(erify) → Δ(Delta)\u002FR(eflect). Use when the task is a non-trivial feature implementation in a brownfield codebase and you need the complete methodology reference: complexity routing, evidence-grounded planning rules, test-anchoring requirements, and failure-escalation thresholds.","https:\u002F\u002Fgithub.com\u002FRynaro\u002Fmagicite\u002Fblob\u002Fmain\u002F.eidolons\u002Fvivi\u002Fskills\u002Fmethodology\u002FSKILL.md",[],[]]