{ "schemaVersion": "1.0", "item": { "slug": "instruction-anchor-guard", "name": "Instruction Anchor Guard", "source": "tencent", "type": "skill", "category": "AI 智能", "sourceUrl": "https://clawhub.ai/Dalomeve/instruction-anchor-guard", "canonicalUrl": "https://clawhub.ai/Dalomeve/instruction-anchor-guard", "targetPlatform": "OpenClaw" }, "install": { "downloadMode": "redirect", "downloadUrl": "/downloads/instruction-anchor-guard", "sourceDownloadUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=instruction-anchor-guard", "sourcePlatform": "tencent", "targetPlatform": "OpenClaw", "installMethod": "Manual import", "extraction": "Extract archive", "prerequisites": [ "OpenClaw" ], "packageFormat": "ZIP package", "includedAssets": [ "SKILL.md" ], "primaryDoc": "SKILL.md", "quickSetup": [ "Download the package from Yavira.", "Extract the archive and review SKILL.md first.", "Import or place the package into your OpenClaw setup." ], "agentAssist": { "summary": "Hand the extracted package to your coding agent with a concrete install brief instead of figuring it out manually.", "steps": [ "Download the package from Yavira.", "Extract it into a folder your agent can access.", "Paste one of the prompts below and point your agent at the extracted folder." ], "prompts": [ { "label": "New install", "body": "I downloaded a skill package from Yavira. Read SKILL.md from the extracted folder and install it by following the included instructions. Tell me what you changed and call out any manual steps you could not complete." }, { "label": "Upgrade existing", "body": "I downloaded an updated skill package from Yavira. Read SKILL.md from the extracted folder, compare it with my current installation, and upgrade it while preserving any custom configuration unless the package docs explicitly say otherwise. Summarize what changed and any follow-up checks I should run." } ] }, "sourceHealth": { "source": "tencent", "status": "healthy", "reason": "direct_download_ok", "recommendedAction": "download", "checkedAt": "2026-04-30T16:55:25.780Z", "expiresAt": "2026-05-07T16:55:25.780Z", "httpStatus": 200, "finalUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=network", "contentType": "application/zip", "probeMethod": "head", "details": { "probeUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=network", "contentDisposition": "attachment; filename=\"network-1.0.0.zip\"", "redirectLocation": null, "bodySnippet": null }, "scope": "source", "summary": "Source download looks usable.", "detail": "Yavira can redirect you to the upstream package for this source.", "primaryActionLabel": "Download for OpenClaw", "primaryActionHref": "/downloads/instruction-anchor-guard" }, "validation": { "installChecklist": [ "Use the Yavira download entry.", "Review SKILL.md after the package is downloaded.", "Confirm the extracted package contains the expected setup assets." ], "postInstallChecks": [ "Confirm the extracted package includes the expected docs or setup files.", "Validate the skill or prompts are available in your target agent workspace.", "Capture any manual follow-up steps the agent could not complete." ] }, "downloadPageUrl": "https://openagent3.xyz/downloads/instruction-anchor-guard", "agentPageUrl": "https://openagent3.xyz/skills/instruction-anchor-guard/agent", "manifestUrl": "https://openagent3.xyz/skills/instruction-anchor-guard/agent.json", "briefUrl": "https://openagent3.xyz/skills/instruction-anchor-guard/agent.md" }, "agentAssist": { "summary": "Hand the extracted package to your coding agent with a concrete install brief instead of figuring it out manually.", "steps": [ "Download the package from Yavira.", "Extract it into a folder your agent can access.", "Paste one of the prompts below and point your agent at the extracted folder." ], "prompts": [ { "label": "New install", "body": "I downloaded a skill package from Yavira. Read SKILL.md from the extracted folder and install it by following the included instructions. Tell me what you changed and call out any manual steps you could not complete." }, { "label": "Upgrade existing", "body": "I downloaded an updated skill package from Yavira. Read SKILL.md from the extracted folder, compare it with my current installation, and upgrade it while preserving any custom configuration unless the package docs explicitly say otherwise. Summarize what changed and any follow-up checks I should run." } ] }, "documentation": { "source": "clawhub", "primaryDoc": "SKILL.md", "sections": [ { "title": "Instruction Anchor Guard", "body": "Prevent loss or drift of user-critical constraints during compaction, session restart, or long multi-turn tasks." }, { "title": "When To Trigger", "body": "Trigger when any of these appear:\n\nUser marks an instruction as important, critical, must, always, never, highest priority, rule, or constraint\nA task has more than 3 steps and spans multiple turns\nCompaction happened (or is likely) and the task has non-negotiable requirements\nAgent behavior shows possible drift from prior explicit user constraints" }, { "title": "Anchor Ledger Schema", "body": "Store anchors in first available path:\n\nmemory/anchors.md\nmemory/instruction-anchors.md\n.anchors.md (workspace root fallback only)\n\nEntry schema (append-only, one section per anchor):\n\n## Anchor: \n- **source**: \n- **instruction**: \n- **verbatim**: \n- **priority**: P0 | P1 | P2\n- **scope**: global | session | task: | channel:\n- **createdAt**: \n- **expiresAt**: \n- **status**: active | paused | expired | superseded\n- **signature**: \n- **supersedes**: " }, { "title": "1) Capture", "body": "Parse latest user message for candidate anchor statements\nKeep only instruction/constraint content; remove examples/chatter\nAssign default values:\n\npriority: P1 (unless user says critical/highest -> P0)\nscope: session (unless user explicitly asks global/task scope)\nexpiresAt: session end (unless user explicitly sets never/date)" }, { "title": "2) Confirm for Broad Anchors", "body": "If scope is global or priority is P0, ask a one-line confirmation before persisting\nDo not auto-promote P2/P1 to P0 without explicit user intent" }, { "title": "3) Persist", "body": "Append anchor entry to ledger\nIf new anchor conflicts with old same-scope anchor, mark old one superseded\nNever rewrite history silently; keep audit trail" }, { "title": "4) Rehydrate (each turn and after compaction)", "body": "Load active anchors (status=active and not expired)\nBuild an in-memory ANCHOR_SET sorted by priority and recency\nInject ANCHOR_SET into planning phase before tool execution" }, { "title": "5) Drift Check (before final answer and before destructive actions)", "body": "Compare current plan against active anchors\nOn conflict:\n\nP0 conflict: stop and correct plan immediately\nP1 conflict: auto-correct and note adjustment\nP2 conflict: continue only if no user-level contradiction\n\n\nEmit DRIFT_CHECK block in response" }, { "title": "Conflict Resolution", "body": "Priority order:\n\nSystem and safety policy\nUser anchors (P0 > P1 > P2)\nCurrent-turn temporary preferences\n\nTie-breakers:\n\nMore specific scope wins (task > session > global) if same priority\nNewer anchor wins if same priority and same scope\nExplicit user override wins only when safety is not violated" }, { "title": "Output Template", "body": "When anchors exist, include this compact block:\n\nANCHORS_ACTIVE\n| ID | P | Scope | Expires | Instruction |\n|----|---|-------|---------|-------------|\n| anchor-001 | P0 | global | never | Never perform destructive data deletion without confirmation |\n\nPLAN_GUARD\n- Current action: \n- Conflicts: none | \n- Decision: clear | corrected | paused-awaiting-user\n\nDRIFT_CHECK\n- Last anchor sync: \n- Drift: no | yes\n- Fix applied: " }, { "title": "Safety Boundaries", "body": "Never store tokens, API keys, passwords, cookies, or auth headers\nNever store raw personal data unless strictly required by user instruction\nRedact sensitive literals as [REDACTED]\nStore constraints, not datasets\nDo not execute destructive commands solely because an anchor exists; still require explicit confirmation for destructive actions" }, { "title": "Expiry and Maintenance", "body": "Mark expired when expiresAt < now\nSupport control intents:\n\n/anchors list\n/anchors pause \n/anchors resume \n/anchors delete \n/anchors pin never\n\n\nRotate ledger when > 200 entries into memory/anchors-archive-YYYY-MM.md" }, { "title": "Integration Notes", "body": "Pair with memory-self-heal for retry/fallback after drift correction\nPair with task-execution-guard to enforce anchor checks at each milestone\nKeep this skill deterministic and concise; avoid free-form interpretation when conflict exists" } ], "body": "Instruction Anchor Guard\n\nPrevent loss or drift of user-critical constraints during compaction, session restart, or long multi-turn tasks.\n\nWhen To Trigger\n\nTrigger when any of these appear:\n\nUser marks an instruction as important, critical, must, always, never, highest priority, rule, or constraint\nA task has more than 3 steps and spans multiple turns\nCompaction happened (or is likely) and the task has non-negotiable requirements\nAgent behavior shows possible drift from prior explicit user constraints\nAnchor Ledger Schema\n\nStore anchors in first available path:\n\nmemory/anchors.md\nmemory/instruction-anchors.md\n.anchors.md (workspace root fallback only)\n\nEntry schema (append-only, one section per anchor):\n\n## Anchor: \n- **source**: \n- **instruction**: \n- **verbatim**: \n- **priority**: P0 | P1 | P2\n- **scope**: global | session | task: | channel:\n- **createdAt**: \n- **expiresAt**: \n- **status**: active | paused | expired | superseded\n- **signature**: \n- **supersedes**: \n\nWorkflow\n1) Capture\nParse latest user message for candidate anchor statements\nKeep only instruction/constraint content; remove examples/chatter\nAssign default values:\npriority: P1 (unless user says critical/highest -> P0)\nscope: session (unless user explicitly asks global/task scope)\nexpiresAt: session end (unless user explicitly sets never/date)\n2) Confirm for Broad Anchors\nIf scope is global or priority is P0, ask a one-line confirmation before persisting\nDo not auto-promote P2/P1 to P0 without explicit user intent\n3) Persist\nAppend anchor entry to ledger\nIf new anchor conflicts with old same-scope anchor, mark old one superseded\nNever rewrite history silently; keep audit trail\n4) Rehydrate (each turn and after compaction)\nLoad active anchors (status=active and not expired)\nBuild an in-memory ANCHOR_SET sorted by priority and recency\nInject ANCHOR_SET into planning phase before tool execution\n5) Drift Check (before final answer and before destructive actions)\nCompare current plan against active anchors\nOn conflict:\nP0 conflict: stop and correct plan immediately\nP1 conflict: auto-correct and note adjustment\nP2 conflict: continue only if no user-level contradiction\nEmit DRIFT_CHECK block in response\nConflict Resolution\n\nPriority order:\n\nSystem and safety policy\nUser anchors (P0 > P1 > P2)\nCurrent-turn temporary preferences\n\nTie-breakers:\n\nMore specific scope wins (task > session > global) if same priority\nNewer anchor wins if same priority and same scope\nExplicit user override wins only when safety is not violated\nOutput Template\n\nWhen anchors exist, include this compact block:\n\nANCHORS_ACTIVE\n| ID | P | Scope | Expires | Instruction |\n|----|---|-------|---------|-------------|\n| anchor-001 | P0 | global | never | Never perform destructive data deletion without confirmation |\n\nPLAN_GUARD\n- Current action: \n- Conflicts: none | \n- Decision: clear | corrected | paused-awaiting-user\n\nDRIFT_CHECK\n- Last anchor sync: \n- Drift: no | yes\n- Fix applied: \n\nSafety Boundaries\nNever store tokens, API keys, passwords, cookies, or auth headers\nNever store raw personal data unless strictly required by user instruction\nRedact sensitive literals as [REDACTED]\nStore constraints, not datasets\nDo not execute destructive commands solely because an anchor exists; still require explicit confirmation for destructive actions\nExpiry and Maintenance\nMark expired when expiresAt < now\nSupport control intents:\n/anchors list\n/anchors pause \n/anchors resume \n/anchors delete \n/anchors pin never\nRotate ledger when > 200 entries into memory/anchors-archive-YYYY-MM.md\nIntegration Notes\nPair with memory-self-heal for retry/fallback after drift correction\nPair with task-execution-guard to enforce anchor checks at each milestone\nKeep this skill deterministic and concise; avoid free-form interpretation when conflict exists" }, "trust": { "sourceLabel": "tencent", "provenanceUrl": "https://clawhub.ai/Dalomeve/instruction-anchor-guard", "publisherUrl": "https://clawhub.ai/Dalomeve/instruction-anchor-guard", "owner": "Dalomeve", "version": "1.0.0", "license": null, "verificationStatus": "Indexed source record" }, "links": { "detailUrl": "https://openagent3.xyz/skills/instruction-anchor-guard", "downloadUrl": "https://openagent3.xyz/downloads/instruction-anchor-guard", "agentUrl": "https://openagent3.xyz/skills/instruction-anchor-guard/agent", "manifestUrl": "https://openagent3.xyz/skills/instruction-anchor-guard/agent.json", "briefUrl": "https://openagent3.xyz/skills/instruction-anchor-guard/agent.md" } }