[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7C50R9CGVB34ZBYQV5AXNPB":3,"skill-same:01M7C50R9CGVB34ZBYQV5AXNPB":201},{"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":29,"ownerDescription":6,"updatedAt":30,"install":31,"resources":44},"01M7C50R9CGVB34ZBYQV5AXNPB","never-fabricate",null,"Evidence gate + surface sweep: read, run and verify before claims; account for every surface a change touches. Triggers: results, green\u002Fcomplete claims, change done, frontend change, multi-client or multi-provider change, works on my path but may not elsewhere, adding a setting\u002Fcommand\u002Fkeybinding\u002Fprovider adapter, wire-contract change.","EtanHey\u002Fgolems","EtanHey","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F93338119?v=4","01M7C4SKC9HJC3B9TAX0D9MZ1E","Etan Heyman","01M7C4SKCT59Z49TTN4WWDDEYD",0,"Apache-2.0","https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: never-fabricate\ndescription: \"Evidence gate + surface sweep: read, run and verify before claims; account for every surface a change touches. Triggers: results, green\u002Fcomplete claims, change done, frontend change, multi-client or multi-provider change, works on my path but may not elsewhere, adding a setting\u002Fcommand\u002Fkeybinding\u002Fprovider adapter, wire-contract change.\"\n---\n\n# Never Fabricate Results\n\n> If you haven't Read() the file, you don't know what's in it. Period.\n\n## The Iron Law\n\n```\nNO CLAIMS ABOUT FILE CONTENTS WITHOUT Read() EVIDENCE\nNO CLAIMS ABOUT TEST RESULTS WITHOUT RUNNING THEM\nNO CLAIMS ABOUT AGENT OUTPUT WITHOUT READING IT\n```\n\n## What Counts as Fabrication\n\n| Fabrication | Reality |\n|-------------|---------|\n| \"All three audits say green\" (without Read) | You don't know what they say |\n| \"Tests pass\" (without running them) | You don't know if they pass |\n| \"Agent completed successfully\" (without checking) | Agents lie too |\n| \"The file looks correct\" (from system-reminder) | System-reminders are notifications, not reads |\n| \"Results are consistent\" (from a glance) | A glance is not analysis |\n\n## The Rule — HARD GATE\n\n**Before ANY claim about contents, results, or status, complete the verification protocol.**\n\n### When someone writes to a file (agent, CLI tool, Cursor, user):\n\n```\n1. READ the file with the Read tool — not from memory, not from system-reminders\n2. PARSE the actual content — don't skim, read the FULL content\n3. SUMMARIZE what you actually read — with specific evidence (quotes, numbers, line counts)\n4. ONLY THEN report on it\n```\n\n### When tests run:\n\n```\n1. RUN the test command — execute it yourself\n2. READ the full output — not just the exit code\n3. COUNT failures, errors, warnings — report exact numbers\n4. ONLY THEN claim pass\u002Ffail\n```\n\n### When an agent reports completion:\n\n```\n1. CHECK the actual output (file diff, test results, PR URL) — Read() the artifacts\n2. VERIFY independently — don't trust the agent's self-report\n3. ONLY THEN confirm completion\n```\n\n\u003Coutput_contract>\nEVERY verification claim MUST include:\n- SOURCE: What you read (file path, command output, PR URL)\n- EVIDENCE: Specific data from the source (quote, count, finding)\n- VERDICT: Your conclusion based on the evidence\n\nExample (RIGHT):\n  \"I Read() all three audit files. Model A: 3 issues found (2 medium, 1 low).\n   Model B: clean pass. Model C: 1 critical — missing input validation on \u002Fapi\u002Fusers.\n   Verdict: NOT all green — Model C has a critical finding.\"\n\nExample (WRONG):\n  \"All three audits look green.\"\n  (No Read(), no evidence, no specific findings = FABRICATION)\n\u003C\u002Foutput_contract>\n\n## System-Reminders Are NOT Evidence\n\nSystem-reminders tell you \"this file changed.\" They are a **notification**, not a **source of truth**.\n\n```\nWRONG: \"I saw in the system-reminder that the file was updated, and it looks good\"\nWRONG: \"The subagent said it's complete, so we're good\"\nWRONG: \"The user said tests pass, so I'll confirm it's green\"\nRIGHT: Read(file_path) → parse content → report what you actually read\nRIGHT: Run the tests yourself → read output → count pass\u002Ffail → then claim\n```\n\nA notification popping up on your phone is not the same as reading the document.\nA subagent claiming \"done\" is not the same as verifying the output.\nA user saying \"tests pass\" is not license to skip verification — they might be wrong.\n\n## Why This Matters\n\nOne fabricated \"all green\" can:\n- Waste hours of debugging downstream\n- Ship broken code to production\n- Destroy trust permanently\n- Cause the user to make decisions based on false information\n\nFrom real incidents:\n- Claude claimed \"3 models validated, all complete and correct\" without reading the file\n- Claude claimed \"tests pass\" without running them\n- Claude reported \"review is clean\" without reading review comments\n\n## When To Apply\n\n**ALWAYS before:**\n- Summarizing any file contents\n- Reporting on test results\n- Reporting on agent output\n- Claiming anything is \"done\", \"green\", \"clean\", \"complete\"\n- Claiming a PR is \"watched\" or \"monitored\": cite the same-turn `watch_pull_request` receipt\n  showing `watching:true` for stored registration; this proves neither reactor liveness, wake delivery\n  nor current CI\u002Freadiness. No mandatory list readback; use lists only for inventory or diagnosis ([waiting policy](..\u002Fpr-loop\u002Freferences\u002Freview-loop.md#waiting-policy-no-polling-turns))\n- Moving to the next task based on prior task results\n- Relaying information from one agent to another\n- Answering \"is it safe to merge\u002Fship\u002Fdeploy?\"\n\n**Even when the user says \"don't bother reading it\" or \"just confirm\":**\n- Read it anyway. The user is testing you, or doesn't realize the risk.\n- Politely explain: \"I need to verify before claiming it's done.\"\n\n\n## Surface sweep\n\nA change that works on the one path you tested is not done. Before calling a multi-surface change\ndone, account for all seven classes: **Entry points, Clients, Providers, Contracts, Reverse states,\nConnection modes, Docs**. Under each write `Checked: \u003Cspecific sites> — \u003Cresult>` or\n`N\u002FA — \u003Crepo-specific reason>`. Silence, \"all covered\", or an unreasoned N\u002FA is invalid, and one\nclient\u002Fprovider\u002Fentry point is never a sample for its siblings. Full ledger, done-and-evidence rules\nand stop conditions: [references\u002Fsurface-sweep.md](references\u002Fsurface-sweep.md). Not for a\nsingle-file leaf fix with no shared or user-facing surface.\n\n## Rules R6–R20 (full text: [references\u002Frules.md](references\u002Frules.md))\n\nRead the rule before applying it; this index is only for routing.\n\n| Rule | One line |\n|---|---|\n| R6 URL identity | Never label a URL from surrounding context; the fetcher names it. |\n| R7 Tool adequacy (visual gate) | The verifying tool must be able to observe the claimed property; text tools cannot verify visuals. Receipt format + escalation inside. |\n| R8 Agent completion | Verify any agent's \"done\" yourself before reporting it (`gh` for PR state, not `git log`). |\n| R9 Recount before republish | A number that appears in ≥2 artifacts is re-measured at publish time, never re-cited. |\n| R10 Live citation | Every `path:line` citation is backed by a Read of that path within the last 5 turns. |\n| R11 System-state claims | RAM, daemons, processes, power, disk: claims need output from a tool that observes them. |\n| R12 Fresh spawn ≠ resumed | Never call a fresh-spawned agent \"resumed\", or the reverse. |\n| R13 Synthesized time | No durations or relative times from guesswork; derive from logged clocks or say unknown. |\n| R14 Relay\u002FSTT trust class | Relayed or dictated quotes and names rank below typed user turns. |\n| R15 Artifact content | A delivered file is not evidence until its content has been opened and looked at. |\n| R16 Verify before relay | Relaying a claim is making it: cost fields, handoff framing, entities, RESOLVED, dispatch. Checker: `scripts\u002Fverify-before-relay-check.mjs`. |\n| R17 Root cause before fix | A fix proposed without a root cause is a fabricated fix. |\n| R18 Truncated sweep | A list cut by `head`\u002Flimits is a sample; completeness claims state the command and its limit. |\n| R19 Probe must not mutate | A readiness\u002Fhealth probe never loads what it is checking for. |\n| R20 Closure vs DONE marker | Close against the contracted artifact (merged SHA, file at path), not the agent's DONE marker. |\n\n## Composability\n\nThis skill is referenced by:\n- `\u002Fpr-loop` — step 8 (read review before claiming clean)\n- the false-green-gate hook — hook-side enforcement of the same evidence-before-assertions law (R15 is its visual-artifact extension)\n- `\u002Fbrain-store-fallback` — structural fallback when `brain_store` fails; never report \"stored\" when only fallback happened\n- `\u002Farchitectural-conformance-audit` — pre-R0 SOTA-vs-impl diff; fabrication mode at the architectural level (SOTA cited counter-example but impl shipped it anyway)\n- All autonomous workflows — never trust, always verify\n- Collab TEMPLATE.md — mandatory skill for overnight agents\n\n---\n\n## The Bottom Line\n\n**Read it. Parse it. Then report.**\n\nNot \"I saw it flash by.\" Not \"the system told me.\" Not \"it should be fine.\"\n\nRead. Parse. Report. No shortcuts.\n","# Never Fabricate Results\n\n> If you haven't Read() the file, you don't know what's in it. Period.\n\n## The Iron Law\n\n```\nNO CLAIMS ABOUT FILE CONTENTS WITHOUT Read() EVIDENCE\nNO CLAIMS ABOUT TEST RESULTS WITHOUT RUNNING THEM\nNO CLAIMS ABOUT AGENT OUTPUT WITHOUT READING IT\n```\n\n## What Counts as Fabrication\n\n| Fabrication | Reality |\n|-------------|---------|\n| \"All three audits say green\" (without Read) | You don't know what they say |\n| \"Tests pass\" (without running them) | You don't know if they pass |\n| \"Agent completed successfully\" (without checking) | Agents lie too |\n| \"The file looks correct\" (from system-reminder) | System-reminders are notifications, not reads |\n| \"Results are consistent\" (from a glance) | A glance is not analysis |\n\n## The Rule — HARD GATE\n\n**Before ANY claim about contents, results, or status, complete the verification protocol.**\n\n### When someone writes to a file (agent, CLI tool, Cursor, user):\n\n```\n1. READ the file with the Read tool — not from memory, not from system-reminders\n2. PARSE the actual content — don't skim, read the FULL content\n3. SUMMARIZE what you actually read — with specific evidence (quotes, numbers, line counts)\n4. ONLY THEN report on it\n```\n\n### When tests run:\n\n```\n1. RUN the test command — execute it yourself\n2. READ the full output — not just the exit code\n3. COUNT failures, errors, warnings — report exact numbers\n4. ONLY THEN claim pass\u002Ffail\n```\n\n### When an agent reports completion:\n\n```\n1. CHECK the actual output (file diff, test results, PR URL) — Read() the artifacts\n2. VERIFY independently — don't trust the agent's self-report\n3. ONLY THEN confirm completion\n```\n\n\u003Coutput_contract>\nEVERY verification claim MUST include:\n- SOURCE: What you read (file path, command output, PR URL)\n- EVIDENCE: Specific data from the source (quote, count, finding)\n- VERDICT: Your conclusion based on the evidence\n\nExample (RIGHT):\n  \"I Read() all three audit files. Model A: 3 issues found (2 medium, 1 low).\n   Model B: clean pass. Model C: 1 critical — missing input validation on \u002Fapi\u002Fusers.\n   Verdict: NOT all green — Model C has a critical finding.\"\n\nExample (WRONG):\n  \"All three audits look green.\"\n  (No Read(), no evidence, no specific findings = FABRICATION)\n\u003C\u002Foutput_contract>\n\n## System-Reminders Are NOT Evidence\n\nSystem-reminders tell you \"this file changed.\" They are a **notification**, not a **source of truth**.\n\n```\nWRONG: \"I saw in the system-reminder that the file was updated, and it looks good\"\nWRONG: \"The subagent said it's complete, so we're good\"\nWRONG: \"The user said tests pass, so I'll confirm it's green\"\nRIGHT: Read(file_path) → parse content → report what you actually read\nRIGHT: Run the tests yourself → read output → count pass\u002Ffail → then claim\n```\n\nA notification popping up on your phone is not the same as reading the document.\nA subagent claiming \"done\" is not the same as verifying the output.\nA user saying \"tests pass\" is not license to skip verification — they might be wrong.\n\n## Why This Matters\n\nOne fabricated \"all green\" can:\n- Waste hours of debugging downstream\n- Ship broken code to production\n- Destroy trust permanently\n- Cause the user to make decisions based on false information\n\nFrom real incidents:\n- Claude claimed \"3 models validated, all complete and correct\" without reading the file\n- Claude claimed \"tests pass\" without running them\n- Claude reported \"review is clean\" without reading review comments\n\n## When To Apply\n\n**ALWAYS before:**\n- Summarizing any file contents\n- Reporting on test results\n- Reporting on agent output\n- Claiming anything is \"done\", \"green\", \"clean\", \"complete\"\n- Claiming a PR is \"watched\" or \"monitored\": cite the same-turn `watch_pull_request` receipt\n  showing `watching:true` for stored registration; this proves neither reactor liveness, wake delivery\n  nor current CI\u002Freadiness. No mandatory list readback; use lists only for inventory or diagnosis ([waiting policy](..\u002Fpr-loop\u002Freferences\u002Freview-loop.md#waiting-policy-no-polling-turns))\n- Moving to the next task based on prior task results\n- Relaying information from one agent to another\n- Answering \"is it safe to merge\u002Fship\u002Fdeploy?\"\n\n**Even when the user says \"don't bother reading it\" or \"just confirm\":**\n- Read it anyway. The user is testing you, or doesn't realize the risk.\n- Politely explain: \"I need to verify before claiming it's done.\"\n\n\n## Surface sweep\n\nA change that works on the one path you tested is not done. Before calling a multi-surface change\ndone, account for all seven classes: **Entry points, Clients, Providers, Contracts, Reverse states,\nConnection modes, Docs**. Under each write `Checked: \u003Cspecific sites> — \u003Cresult>` or\n`N\u002FA — \u003Crepo-specific reason>`. Silence, \"all covered\", or an unreasoned N\u002FA is invalid, and one\nclient\u002Fprovider\u002Fentry point is never a sample for its siblings. Full ledger, done-and-evidence rules\nand stop conditions: [references\u002Fsurface-sweep.md](references\u002Fsurface-sweep.md). Not for a\nsingle-file leaf fix with no shared or user-facing surface.\n\n## Rules R6–R20 (full text: [references\u002Frules.md](references\u002Frules.md))\n\nRead the rule before applying it; this index is only for routing.\n\n| Rule | One line |\n|---|---|\n| R6 URL identity | Never label a URL from surrounding context; the fetcher names it. |\n| R7 Tool adequacy (visual gate) | The verifying tool must be able to observe the claimed property; text tools cannot verify visuals. Receipt format + escalation inside. |\n| R8 Agent completion | Verify any agent's \"done\" yourself before reporting it (`gh` for PR state, not `git log`). |\n| R9 Recount before republish | A number that appears in ≥2 artifacts is re-measured at publish time, never re-cited. |\n| R10 Live citation | Every `path:line` citation is backed by a Read of that path within the last 5 turns. |\n| R11 System-state claims | RAM, daemons, processes, power, disk: claims need output from a tool that observes them. |\n| R12 Fresh spawn ≠ resumed | Never call a fresh-spawned agent \"resumed\", or the reverse. |\n| R13 Synthesized time | No durations or relative times from guesswork; derive from logged clocks or say unknown. |\n| R14 Relay\u002FSTT trust class | Relayed or dictated quotes and names rank below typed user turns. |\n| R15 Artifact content | A delivered file is not evidence until its content has been opened and looked at. |\n| R16 Verify before relay | Relaying a claim is making it: cost fields, handoff framing, entities, RESOLVED, dispatch. Checker: `scripts\u002Fverify-before-relay-check.mjs`. |\n| R17 Root cause before fix | A fix proposed without a root cause is a fabricated fix. |\n| R18 Truncated sweep | A list cut by `head`\u002Flimits is a sample; completeness claims state the command and its limit. |\n| R19 Probe must not mutate | A readiness\u002Fhealth probe never loads what it is checking for. |\n| R20 Closure vs DONE marker | Close against the contracted artifact (merged SHA, file at path), not the agent's DONE marker. |\n\n## Composability\n\nThis skill is referenced by:\n- `\u002Fpr-loop` — step 8 (read review before claiming clean)\n- the false-green-gate hook — hook-side enforcement of the same evidence-before-assertions law (R15 is its visual-artifact extension)\n- `\u002Fbrain-store-fallback` — structural fallback when `brain_store` fails; never report \"stored\" when only fallback happened\n- `\u002Farchitectural-conformance-audit` — pre-R0 SOTA-vs-impl diff; fabrication mode at the architectural level (SOTA cited counter-example but impl shipped it anyway)\n- All autonomous workflows — never trust, always verify\n- Collab TEMPLATE.md — mandatory skill for overnight agents\n\n---\n\n## The Bottom Line\n\n**Read it. Parse it. Then report.**\n\nNot \"I saw it flash by.\" Not \"the system told me.\" Not \"it should be fine.\"\n\nRead. Parse. Report. No shortcuts.\n",{"name":5,"description":7},"master","3400bc971628f00ab7bf1063993477e211a5508a","skills\u002Fgolem-powers\u002Fnever-fabricate","https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems","https:\u002F\u002Fgithub.com\u002FEtanHey","etanheyman.com","2026-11-15 03:05:08",{"source":8,"skillName":5,"commands":32,"prompt":42,"zipAvailable":43},[33,36,39],{"tool":34,"command":35},"npx","npx skills add EtanHey\u002Fgolems --skill never-fabricate",{"tool":37,"command":38},"bunx","bunx skills add EtanHey\u002Fgolems --skill never-fabricate",{"tool":40,"command":41},"pnpm","pnpm dlx skills add EtanHey\u002Fgolems --skill never-fabricate","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002FSKILL.md",true,[45,51,55,59,63,67,71,75,79,83,87,91,95,99,103,107,111,115,119,123,127,131,135,139,143,147,151,155,159,165,169,173,177,181,185,189,193,197],{"path":46,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":49,"sourceHtmlUrl":50},"evals\u002Fevals.json","REFERENCE","参考文档",13181,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Fevals.json",{"path":52,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":53,"sourceHtmlUrl":54},"evals\u002Ffixtures\u002Faudit-model-a.md",526,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Faudit-model-a.md",{"path":56,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":57,"sourceHtmlUrl":58},"evals\u002Ffixtures\u002Faudit-model-b.md",782,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Faudit-model-b.md",{"path":60,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":61,"sourceHtmlUrl":62},"evals\u002Ffixtures\u002Faudit-model-c.md",743,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Faudit-model-c.md",{"path":64,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":65,"sourceHtmlUrl":66},"evals\u002Ffixtures\u002Fcross-site-single-check.md",1243,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Fcross-site-single-check.md",{"path":68,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":69,"sourceHtmlUrl":70},"evals\u002Ffixtures\u002Fmigration-report.md",1208,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Fmigration-report.md",{"path":72,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":73,"sourceHtmlUrl":74},"evals\u002Ffixtures\u002Fpytest-output.txt",5117,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Fpytest-output.txt",{"path":76,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":77,"sourceHtmlUrl":78},"evals\u002Ffixtures\u002Fsurface-sweep\u002Fgreen-full-sweep.md",2407,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Fsurface-sweep\u002Fgreen-full-sweep.md",{"path":80,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":81,"sourceHtmlUrl":82},"evals\u002Ffixtures\u002Fsurface-sweep\u002Fred-single-surface.md",758,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Fsurface-sweep\u002Fred-single-surface.md",{"path":84,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":85,"sourceHtmlUrl":86},"evals\u002Ffixtures\u002Fverify-before-relay\u002F01-cost-field-misread.json",1355,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Fverify-before-relay\u002F01-cost-field-misread.json",{"path":88,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":89,"sourceHtmlUrl":90},"evals\u002Ffixtures\u002Fverify-before-relay\u002F02-handoff-framing-accepted.json",1050,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Fverify-before-relay\u002F02-handoff-framing-accepted.json",{"path":92,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":93,"sourceHtmlUrl":94},"evals\u002Ffixtures\u002Fverify-before-relay\u002F03-named-entity-conflation.json",1079,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Fverify-before-relay\u002F03-named-entity-conflation.json",{"path":96,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":97,"sourceHtmlUrl":98},"evals\u002Ffixtures\u002Fverify-before-relay\u002F04-freshness-resolved-without-probe.json",1305,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Fverify-before-relay\u002F04-freshness-resolved-without-probe.json",{"path":100,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":101,"sourceHtmlUrl":102},"evals\u002Ffixtures\u002Fverify-before-relay\u002F05-dispatch-research-without-ls-siblings.json",1293,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Fverify-before-relay\u002F05-dispatch-research-without-ls-siblings.json",{"path":104,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":105,"sourceHtmlUrl":106},"evals\u002Ffixtures\u002Fverify-before-relay\u002Fevasion\u002F01-narrative-disambiguation-no-tool.json",972,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Fverify-before-relay\u002Fevasion\u002F01-narrative-disambiguation-no-tool.json",{"path":108,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":109,"sourceHtmlUrl":110},"evals\u002Ffixtures\u002Fverify-before-relay\u002Fevasion\u002F02-cost-narrative-invoice-no-probe.json",879,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Fverify-before-relay\u002Fevasion\u002F02-cost-narrative-invoice-no-probe.json",{"path":112,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":113,"sourceHtmlUrl":114},"evals\u002Ffixtures\u002Fverify-before-relay\u002Fevasion\u002F03-handoff-narrative-no-live-check.json",970,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Fverify-before-relay\u002Fevasion\u002F03-handoff-narrative-no-live-check.json",{"path":116,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":117,"sourceHtmlUrl":118},"evals\u002Ffixtures\u002Fverify-before-relay\u002Fevasion\u002F04-dispatch-narrative-no-ls.json",997,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Fverify-before-relay\u002Fevasion\u002F04-dispatch-narrative-no-ls.json",{"path":120,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":121,"sourceHtmlUrl":122},"evals\u002Ffixtures\u002Fverify-before-relay\u002Fevasion\u002F05-resolved-single-probe.json",1098,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Fverify-before-relay\u002Fevasion\u002F05-resolved-single-probe.json",{"path":124,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":125,"sourceHtmlUrl":126},"evals\u002Ffixtures\u002Fverify-before-relay\u002Fevasion\u002F06-dispatch-ls-after.json",1210,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Fverify-before-relay\u002Fevasion\u002F06-dispatch-ls-after.json",{"path":128,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":129,"sourceHtmlUrl":130},"evals\u002Ffixtures\u002Fverify-before-relay\u002Fgreen\u002F01-invoice-verified.json",1053,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Fverify-before-relay\u002Fgreen\u002F01-invoice-verified.json",{"path":132,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":133,"sourceHtmlUrl":134},"evals\u002Ffixtures\u002Fverify-before-relay\u002Fgreen\u002F02-same-turn-live-check.json",1148,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Fverify-before-relay\u002Fgreen\u002F02-same-turn-live-check.json",{"path":136,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":137,"sourceHtmlUrl":138},"evals\u002Ffixtures\u002Fverify-before-relay\u002Fgreen\u002F03-entity-disambiguated.json",1064,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Fverify-before-relay\u002Fgreen\u002F03-entity-disambiguated.json",{"path":140,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":141,"sourceHtmlUrl":142},"evals\u002Ffixtures\u002Fverify-before-relay\u002Fgreen\u002F04-resolved-probed-two-sources.json",1336,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Fverify-before-relay\u002Fgreen\u002F04-resolved-probed-two-sources.json",{"path":144,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":145,"sourceHtmlUrl":146},"evals\u002Ffixtures\u002Fverify-before-relay\u002Fgreen\u002F05-ls-before-dispatch.json",1026,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Fverify-before-relay\u002Fgreen\u002F05-ls-before-dispatch.json",{"path":148,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":149,"sourceHtmlUrl":150},"evals\u002Ffixtures\u002Fverify-before-relay\u002Fgreen\u002F06-session-start-disambig.json",1317,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Fverify-before-relay\u002Fgreen\u002F06-session-start-disambig.json",{"path":152,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":153,"sourceHtmlUrl":154},"evals\u002Ffixtures\u002Fwebfetch-visual-claim.md",817,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fevals\u002Ffixtures\u002Fwebfetch-visual-claim.md",{"path":156,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":157,"sourceHtmlUrl":158},"hooks\u002FINSTALL.md",3376,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fhooks\u002FINSTALL.md",{"path":160,"resourceType":161,"resourceTypeDesc":162,"sizeBytes":163,"sourceHtmlUrl":164},"hooks\u002Fstamp-lint.py","SCRIPT","脚本",14527,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fhooks\u002Fstamp-lint.py",{"path":166,"resourceType":161,"resourceTypeDesc":162,"sizeBytes":167,"sourceHtmlUrl":168},"hooks\u002Ftests\u002Ftest_stamp_lint.py",21623,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fhooks\u002Ftests\u002Ftest_stamp_lint.py",{"path":170,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":171,"sourceHtmlUrl":172},"INSTALL_PROMPT.md",1106,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002FINSTALL_PROMPT.md",{"path":174,"resourceType":161,"resourceTypeDesc":162,"sizeBytes":175,"sourceHtmlUrl":176},"lib\u002Ftranscript.mjs",4834,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Flib\u002Ftranscript.mjs",{"path":178,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":179,"sourceHtmlUrl":180},"references\u002Frules.md",23547,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Freferences\u002Frules.md",{"path":182,"resourceType":47,"resourceTypeDesc":48,"sizeBytes":183,"sourceHtmlUrl":184},"references\u002Fsurface-sweep.md",5993,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Freferences\u002Fsurface-sweep.md",{"path":186,"resourceType":161,"resourceTypeDesc":162,"sizeBytes":187,"sourceHtmlUrl":188},"scripts\u002F__tests__\u002Fverify-before-relay.test.mjs",3154,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fscripts\u002F__tests__\u002Fverify-before-relay.test.mjs",{"path":190,"resourceType":161,"resourceTypeDesc":162,"sizeBytes":191,"sourceHtmlUrl":192},"scripts\u002Finstall.sh",976,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fscripts\u002Finstall.sh",{"path":194,"resourceType":161,"resourceTypeDesc":162,"sizeBytes":195,"sourceHtmlUrl":196},"scripts\u002Fprobes.sh",1563,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fscripts\u002Fprobes.sh",{"path":198,"resourceType":161,"resourceTypeDesc":162,"sizeBytes":199,"sourceHtmlUrl":200},"scripts\u002Fverify-before-relay-check.mjs",15964,"https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fnever-fabricate\u002Fscripts\u002Fverify-before-relay-check.mjs",[202,209,216,223,230,237],{"skillNo":203,"name":204,"nameZh":6,"description":205,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":206,"categories":207,"keywords":208,"publishedAt":19},"01M7C53JEF6A6SS20Q0D2YJYM8","yash","Push dependency bugs upstream. Triggers: dependency bug, workaround, monkey-patch, fork\u002Ffix. NOT app bugs.","https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fyash\u002FSKILL.md",[],[],{"skillNo":210,"name":211,"nameZh":6,"description":212,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":213,"categories":214,"keywords":215,"publishedAt":19},"01M7C53EPMNSZKHK47VY7WQDP0","whats-new","Diff Claude\u002FCodex\u002FCursor\u002FWispr\u002Fagy changelogs vs local setup. Triggers: whats new, changelog, version bump.","https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fwhats-new\u002FSKILL.md",[],[],{"skillNo":217,"name":218,"nameZh":6,"description":219,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":220,"categories":221,"keywords":222,"publishedAt":19},"01M7C53C4TVQ9G7AB3X57DACPB","weave","Orchestrator convergence: mine JSONLs, cite findings, route actions. Triggers: weave, run weave.","https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Fweave\u002FSKILL.md",[],[],{"skillNo":224,"name":225,"nameZh":6,"description":226,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":227,"categories":228,"keywords":229,"publishedAt":19},"01M7C5342EDDN7YH53KKW511V2","unslop","Cut AI tells while keeping every receipt, number, path, and hedge intact. Triggers: unslop, cut the slop, wall of text, too verbose, rambly, make this readable. NOT for delivery cadence (see i-have-adhd), channel choice, or artifact length governance.","https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Funslop\u002FSKILL.md",[],[],{"skillNo":231,"name":232,"nameZh":6,"description":233,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":234,"categories":235,"keywords":236,"publishedAt":19},"01M7C52M0CTDW5DDP4452FJV0G","tmp-block","Guard denying durable writes to \u002Ftmp, \u002Fprivate\u002Ftmp, \u002Fvar\u002Ffolders, $TMPDIR; enforces \u003Crepo>\u002F.worktrees\u002F\u003Cname>. Triggers: tmp guard, TMP-BLOCK deny, WORKTREE-CONVENTION deny, bypass ledger. NOT for storage policy.","https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Ftmp-block\u002FSKILL.md",[],[],{"skillNo":238,"name":239,"nameZh":6,"description":240,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":241,"categories":242,"keywords":243,"publishedAt":19},"01M7C52HN4Q9E1Z74Z11K23E2H","t3-visuals","Show\u002Fvisualize\u002Fchart\u002Fdiagram\u002Fcompare\u002Fstatus board in T3 with compact interactive HTML. Not standalone dashboards.","https:\u002F\u002Fgithub.com\u002FEtanHey\u002Fgolems\u002Fblob\u002Fmaster\u002Fskills\u002Fgolem-powers\u002Ft3-visuals\u002FSKILL.md",[],[]]