{ "schemaVersion": "1.0", "item": { "slug": "shed", "name": "Shed", "source": "tencent", "type": "skill", "category": "开发工具", "sourceUrl": "https://clawhub.ai/compass-soul/shed", "canonicalUrl": "https://clawhub.ai/compass-soul/shed", "targetPlatform": "OpenClaw" }, "install": { "downloadMode": "redirect", "downloadUrl": "/downloads/shed", "sourceDownloadUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=shed", "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": "2026-05-07T17:22:31.273Z", "expiresAt": "2026-05-14T17:22:31.273Z", "httpStatus": 200, "finalUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=afrexai-annual-report", "contentType": "application/zip", "probeMethod": "head", "details": { "probeUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=afrexai-annual-report", "contentDisposition": "attachment; filename=\"afrexai-annual-report-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/shed" }, "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/shed", "agentPageUrl": "https://openagent3.xyz/skills/shed/agent", "manifestUrl": "https://openagent3.xyz/skills/shed/agent.json", "briefUrl": "https://openagent3.xyz/skills/shed/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": "Shed — Context Hygiene for Agents", "body": "Shed what you don't need. Keep what matters.\n\nNamed for molting — the process of shedding an outer layer to grow. Your context window is your skin. When it gets too heavy, shed the dead weight." }, { "title": "Core Principle", "body": "Tool outputs are 84% of your context growth but the lowest-value tokens you carry. (Lindenbauer et al., NeurIPS 2025 DL4C workshop, measured on SWE-agent). Everything flows from this." }, { "title": "After Every Tool Call", "body": "Extract, don't accumulate. When a tool returns large output (file contents, search results, logs, API responses), immediately write the key facts to a file or compress into bullets. The raw output is now disposable.\nAsk: \"Will I need this verbatim later?\" Almost never. The answer you extracted is what matters, not the 500 lines that contained it." }, { "title": "When Context Reaches ~70%", "body": "Trigger condensation. Don't wait for the platform to compact you — that's losing control of your own memory. At 70%, actively shed.\nMask old tool outputs first (free, no LLM calls). Keep your reasoning and action history intact — you need your decision chain, not the raw ls -la from 20 turns ago.\nSummarize reasoning only as backup. If masking isn't enough, compress old reasoning turns. But this is lossy and costs an LLM call — use sparingly.\nNever re-summarize a summary. If you've already condensed once and context is growing again, switch context or spawn a sub-agent. Recursive summarization compounds errors." }, { "title": "When Completing a Task", "body": "Write results to file, then switch context immediately. Stale completed-task context is anti-signal for your next task. Don't carry it.\nLeave breadcrumbs. Before switching: write what you did, what's next, and where the files are to memory/YYYY-MM-DD.md. Future-you needs a trailhead, not a transcript." }, { "title": "When Delegating Work", "body": "Spawn fresh-context sub-agents for complex sub-tasks. Your context is noise for their work. Give them a clean prompt with just what they need.\nDon't inherit parent context into children. The AutoGen pattern: each agent gets its own token budget. Inherited bloat = inherited degradation." }, { "title": "Architecture (For Agent Builders)", "body": "Structure context into typed blocks with hard size limits. Every production framework converges here — Letta uses labeled blocks (human, persona, knowledge) with character caps. A monolithic context is unmanageable.\nSeparate working memory (in-context) from reference memory (file/DB). Your effective context is much smaller than your window size. Models lose information in the middle of long contexts.\nPlace critical information at the beginning or end of context, never the middle. Positional attention bias underweights middle content by up to 15 percentage points (Hsieh et al., 2024, \"Found in the Middle\")." }, { "title": "The Complexity Trap", "body": "Don't assume sophisticated compression (LLM summarization) beats simple approaches (observation masking). The JetBrains \"Complexity Trap\" paper (2025) tested both across 5 model configurations on SWE-bench Verified:\n\nSimple masking halved cost relative to raw agent\nMasking matched or exceeded LLM summarization solve rates\nExample: Qwen3-Coder went from 53.8% → 54.8% with masking alone\n\nThe lesson: start simple. Mask tool outputs. Only add summarization if masking alone isn't enough." }, { "title": "Cost Model", "body": "Without intervention, cost per turn scales quadratically (each turn adds tokens AND reprocesses all previous tokens). Periodic condensation converts this to linear scaling. OpenHands measured 2x cost reduction with their condenser." }, { "title": "Quick Reference", "body": "SituationActionTool returned big outputExtract facts → file → discard rawContext at ~70%Mask old tool outputsContext still growing after maskingSummarize oldest reasoning turnsTask completeWrite results → switch contextComplex sub-task neededSpawn fresh sub-agentAlready condensed, still growingSwitch context or spawnCritical info to preservePut at start or end, not middle" }, { "title": "Sources", "body": "Lindenbauer et al., \"The Complexity Trap\" (NeurIPS 2025 DL4C): https://arxiv.org/abs/2508.21433\nOpenHands Context Condensation (2025): https://openhands.dev/blog/openhands-context-condensensation-for-more-efficient-ai-agents\nLetta/MemGPT Memory Blocks: https://www.letta.com/blog/memory-blocks\nLLMLingua-2 (ACL 2024): https://aclanthology.org/2024.acl-long.91/\nLiu et al., \"Lost in the Middle\" (2023): https://arxiv.org/abs/2307.03172\nHsieh et al., \"Found in the Middle\" (2024): https://arxiv.org/abs/2406.16008\nMEM1 Dynamic State Management (2025): https://arxiv.org/abs/2506.15841" } ], "body": "Shed — Context Hygiene for Agents\n\nShed what you don't need. Keep what matters.\n\nNamed for molting — the process of shedding an outer layer to grow. Your context window is your skin. When it gets too heavy, shed the dead weight.\n\nCore Principle\n\nTool outputs are 84% of your context growth but the lowest-value tokens you carry. (Lindenbauer et al., NeurIPS 2025 DL4C workshop, measured on SWE-agent). Everything flows from this.\n\nThe Rules\nAfter Every Tool Call\nExtract, don't accumulate. When a tool returns large output (file contents, search results, logs, API responses), immediately write the key facts to a file or compress into bullets. The raw output is now disposable.\nAsk: \"Will I need this verbatim later?\" Almost never. The answer you extracted is what matters, not the 500 lines that contained it.\nWhen Context Reaches ~70%\nTrigger condensation. Don't wait for the platform to compact you — that's losing control of your own memory. At 70%, actively shed.\nMask old tool outputs first (free, no LLM calls). Keep your reasoning and action history intact — you need your decision chain, not the raw ls -la from 20 turns ago.\nSummarize reasoning only as backup. If masking isn't enough, compress old reasoning turns. But this is lossy and costs an LLM call — use sparingly.\nNever re-summarize a summary. If you've already condensed once and context is growing again, switch context or spawn a sub-agent. Recursive summarization compounds errors.\nWhen Completing a Task\nWrite results to file, then switch context immediately. Stale completed-task context is anti-signal for your next task. Don't carry it.\nLeave breadcrumbs. Before switching: write what you did, what's next, and where the files are to memory/YYYY-MM-DD.md. Future-you needs a trailhead, not a transcript.\nWhen Delegating Work\nSpawn fresh-context sub-agents for complex sub-tasks. Your context is noise for their work. Give them a clean prompt with just what they need.\nDon't inherit parent context into children. The AutoGen pattern: each agent gets its own token budget. Inherited bloat = inherited degradation.\nArchitecture (For Agent Builders)\nStructure context into typed blocks with hard size limits. Every production framework converges here — Letta uses labeled blocks (human, persona, knowledge) with character caps. A monolithic context is unmanageable.\nSeparate working memory (in-context) from reference memory (file/DB). Your effective context is much smaller than your window size. Models lose information in the middle of long contexts.\nPlace critical information at the beginning or end of context, never the middle. Positional attention bias underweights middle content by up to 15 percentage points (Hsieh et al., 2024, \"Found in the Middle\").\nThe Complexity Trap\n\nDon't assume sophisticated compression (LLM summarization) beats simple approaches (observation masking). The JetBrains \"Complexity Trap\" paper (2025) tested both across 5 model configurations on SWE-bench Verified:\n\nSimple masking halved cost relative to raw agent\nMasking matched or exceeded LLM summarization solve rates\nExample: Qwen3-Coder went from 53.8% → 54.8% with masking alone\n\nThe lesson: start simple. Mask tool outputs. Only add summarization if masking alone isn't enough.\n\nCost Model\n\nWithout intervention, cost per turn scales quadratically (each turn adds tokens AND reprocesses all previous tokens). Periodic condensation converts this to linear scaling. OpenHands measured 2x cost reduction with their condenser.\n\nQuick Reference\nSituation\tAction\nTool returned big output\tExtract facts → file → discard raw\nContext at ~70%\tMask old tool outputs\nContext still growing after masking\tSummarize oldest reasoning turns\nTask complete\tWrite results → switch context\nComplex sub-task needed\tSpawn fresh sub-agent\nAlready condensed, still growing\tSwitch context or spawn\nCritical info to preserve\tPut at start or end, not middle\nSources\nLindenbauer et al., \"The Complexity Trap\" (NeurIPS 2025 DL4C): https://arxiv.org/abs/2508.21433\nOpenHands Context Condensation (2025): https://openhands.dev/blog/openhands-context-condensensation-for-more-efficient-ai-agents\nLetta/MemGPT Memory Blocks: https://www.letta.com/blog/memory-blocks\nLLMLingua-2 (ACL 2024): https://aclanthology.org/2024.acl-long.91/\nLiu et al., \"Lost in the Middle\" (2023): https://arxiv.org/abs/2307.03172\nHsieh et al., \"Found in the Middle\" (2024): https://arxiv.org/abs/2406.16008\nMEM1 Dynamic State Management (2025): https://arxiv.org/abs/2506.15841" }, "trust": { "sourceLabel": "tencent", "provenanceUrl": "https://clawhub.ai/compass-soul/shed", "publisherUrl": "https://clawhub.ai/compass-soul/shed", "owner": "compass-soul", "version": "1.0.0", "license": null, "verificationStatus": "Indexed source record" }, "links": { "detailUrl": "https://openagent3.xyz/skills/shed", "downloadUrl": "https://openagent3.xyz/downloads/shed", "agentUrl": "https://openagent3.xyz/skills/shed/agent", "manifestUrl": "https://openagent3.xyz/skills/shed/agent.json", "briefUrl": "https://openagent3.xyz/skills/shed/agent.md" } }