{
  "schemaVersion": "1.0",
  "item": {
    "slug": "context-clean-up",
    "name": "Context Clean Up",
    "source": "tencent",
    "type": "skill",
    "category": "效率提升",
    "sourceUrl": "https://clawhub.ai/phenomenoner/context-clean-up",
    "canonicalUrl": "https://clawhub.ai/phenomenoner/context-clean-up",
    "targetPlatform": "OpenClaw"
  },
  "install": {
    "downloadMode": "redirect",
    "downloadUrl": "/downloads/context-clean-up",
    "sourceDownloadUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=context-clean-up",
    "sourcePlatform": "tencent",
    "targetPlatform": "OpenClaw",
    "installMethod": "Manual import",
    "extraction": "Extract archive",
    "prerequisites": [
      "OpenClaw"
    ],
    "packageFormat": "ZIP package",
    "includedAssets": [
      "SKILL.md",
      "references/cron-noise-checklist.md",
      "references/out-of-band-delivery.md",
      "scripts/context_cleanup_audit.py"
    ],
    "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": "2026-04-23T16:43:11.935Z",
      "expiresAt": "2026-04-30T16:43:11.935Z",
      "httpStatus": 200,
      "finalUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=4claw-imageboard",
      "contentType": "application/zip",
      "probeMethod": "head",
      "details": {
        "probeUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=4claw-imageboard",
        "contentDisposition": "attachment; filename=\"4claw-imageboard-1.0.1.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/context-clean-up"
    },
    "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/context-clean-up",
    "agentPageUrl": "https://openagent3.xyz/skills/context-clean-up/agent",
    "manifestUrl": "https://openagent3.xyz/skills/context-clean-up/agent.json",
    "briefUrl": "https://openagent3.xyz/skills/context-clean-up/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": "Context Clean Up (audit-only)",
        "body": "This skill identifies what is bloating prompt context and turns it into a safe, reversible plan."
      },
      {
        "title": "Contract",
        "body": "Audit-only by default.\nNo automatic deletions.\nNo unattended config edits.\nNo silent cron/session pruning.\nIf you ask for changes, the skill should propose:\n\nexact change,\nexpected impact,\nrollback plan,\nverification steps."
      },
      {
        "title": "Safety model",
        "body": "No exec tool usage.\nNo read tool usage.\nIf you want file-level analysis, run the bundled script manually and paste the JSON."
      },
      {
        "title": "Quick start",
        "body": "/context-clean-up → audit + actionable plan (no changes)\n\nOptional manual report generation:\n\npython3 scripts/context_cleanup_audit.py --out context-cleanup-audit.json\n\nWindows variant:\n\npy -3 scripts/context_cleanup_audit.py --out context-cleanup-audit.json"
      },
      {
        "title": "What to measure (authoritative, not vibes)",
        "body": "When available, prefer fresh-session /context json receipts over subjective claims like “it feels leaner”.\n\nHigh-signal fields:\n\neligible skills\nskills.promptChars\nprojectContextChars\nsystemPrompt.chars\npromptTokens\n\nIf exact receipts are unavailable, fall back to ranked offenders + change scope, but label confidence lower."
      },
      {
        "title": "Common offender classes",
        "body": "Tool result dumps\n\noversized exec output\nlarge read output\nlong web_fetch payloads\n\n\n\nAutomation transcript noise\n\ncron jobs that say “OK” every run\nheartbeat messages that are not alert-only\n\n\n\nBootstrap reinjection bloat\n\novergrown AGENTS.md / MEMORY.md / SOUL.md / USER.md\nlong runbooks embedded directly in SKILL.md\n\n\n\nAmbient specialist surface\n\ntoo many always-visible specialist skills that should be on-demand workers/subagents instead\n\n\n\nSummary accretion\n\nrepeated summaries that keep historical detail instead of restart-critical facts only"
      },
      {
        "title": "Phase 1 — Noise discipline",
        "body": "Make no-op automation truly silent (NO_REPLY or nothing on success).\nKeep alerts out-of-band when possible."
      },
      {
        "title": "Phase 2 — Bootstrap slimming",
        "body": "Keep always-injected files short.\nMove long guidance to references/, memory/, or external notes."
      },
      {
        "title": "Phase 3 — Ambient surface reduction",
        "body": "Remove low-frequency specialist skills from always-on prompt surface.\nPrefer worker/subagent invocation for specialist flows."
      },
      {
        "title": "Phase 4 — Higher-risk changes",
        "body": "Tool-surface or deeper runtime/config narrowing.\nOnly propose with stronger rollback and explicit approval."
      },
      {
        "title": "Step 0 — Determine scope",
        "body": "You need:\n\nworkspace dir\nstate dir (<OPENCLAW_STATE_DIR>)\n\nCommon defaults:\n\nmacOS/Linux: ~/.openclaw\nWindows: %USERPROFILE%\\.openclaw"
      },
      {
        "title": "Step 1 — Run the audit script",
        "body": "python3 scripts/context_cleanup_audit.py --workspace . --state-dir <OPENCLAW_STATE_DIR> --out context-cleanup-audit.json\n\nInterpretation cheatsheet:\n\nhuge tool outputs → transcript bloat\nmany cron/system lines → automation bloat\nlarge bootstrap docs → reinjection bloat"
      },
      {
        "title": "Step 2 — Produce a fix plan",
        "body": "Include:\n\ntop offenders\nlowest-risk fixes first\nexpected impact\nrollback notes\nverification plan"
      },
      {
        "title": "Step 3 — Verify",
        "body": "After changes:\n\nconfirm automation is silent on success\ncheck context growth flattens\nif possible, compare fresh-session /context json before/after"
      },
      {
        "title": "Important caveat",
        "body": "Many OpenClaw runtimes snapshot skills/bootstrap per session.\nSo skill/config slimming often does not fully apply to the current session.\nUse a new session for authoritative verification."
      },
      {
        "title": "References",
        "body": "references/out-of-band-delivery.md\nreferences/cron-noise-checklist.md"
      }
    ],
    "body": "Context Clean Up (audit-only)\n\nThis skill identifies what is bloating prompt context and turns it into a safe, reversible plan.\n\nContract\nAudit-only by default.\nNo automatic deletions.\nNo unattended config edits.\nNo silent cron/session pruning.\nIf you ask for changes, the skill should propose:\nexact change,\nexpected impact,\nrollback plan,\nverification steps.\nSafety model\nNo exec tool usage.\nNo read tool usage.\nIf you want file-level analysis, run the bundled script manually and paste the JSON.\nQuick start\n/context-clean-up → audit + actionable plan (no changes)\n\nOptional manual report generation:\n\npython3 scripts/context_cleanup_audit.py --out context-cleanup-audit.json\n\n\nWindows variant:\n\npy -3 scripts/context_cleanup_audit.py --out context-cleanup-audit.json\n\nWhat to measure (authoritative, not vibes)\n\nWhen available, prefer fresh-session /context json receipts over subjective claims like “it feels leaner”.\n\nHigh-signal fields:\n\neligible skills\nskills.promptChars\nprojectContextChars\nsystemPrompt.chars\npromptTokens\n\nIf exact receipts are unavailable, fall back to ranked offenders + change scope, but label confidence lower.\n\nCommon offender classes\n\nTool result dumps\n\noversized exec output\nlarge read output\nlong web_fetch payloads\n\nAutomation transcript noise\n\ncron jobs that say “OK” every run\nheartbeat messages that are not alert-only\n\nBootstrap reinjection bloat\n\novergrown AGENTS.md / MEMORY.md / SOUL.md / USER.md\nlong runbooks embedded directly in SKILL.md\n\nAmbient specialist surface\n\ntoo many always-visible specialist skills that should be on-demand workers/subagents instead\n\nSummary accretion\n\nrepeated summaries that keep historical detail instead of restart-critical facts only\nRecommended trim ladder (lowest-risk first)\nPhase 1 — Noise discipline\nMake no-op automation truly silent (NO_REPLY or nothing on success).\nKeep alerts out-of-band when possible.\nPhase 2 — Bootstrap slimming\nKeep always-injected files short.\nMove long guidance to references/, memory/, or external notes.\nPhase 3 — Ambient surface reduction\nRemove low-frequency specialist skills from always-on prompt surface.\nPrefer worker/subagent invocation for specialist flows.\nPhase 4 — Higher-risk changes\nTool-surface or deeper runtime/config narrowing.\nOnly propose with stronger rollback and explicit approval.\nWorkflow (audit → plan)\nStep 0 — Determine scope\n\nYou need:\n\nworkspace dir\nstate dir (<OPENCLAW_STATE_DIR>)\n\nCommon defaults:\n\nmacOS/Linux: ~/.openclaw\nWindows: %USERPROFILE%\\.openclaw\nStep 1 — Run the audit script\npython3 scripts/context_cleanup_audit.py --workspace . --state-dir <OPENCLAW_STATE_DIR> --out context-cleanup-audit.json\n\n\nInterpretation cheatsheet:\n\nhuge tool outputs → transcript bloat\nmany cron/system lines → automation bloat\nlarge bootstrap docs → reinjection bloat\nStep 2 — Produce a fix plan\n\nInclude:\n\ntop offenders\nlowest-risk fixes first\nexpected impact\nrollback notes\nverification plan\nStep 3 — Verify\n\nAfter changes:\n\nconfirm automation is silent on success\ncheck context growth flattens\nif possible, compare fresh-session /context json before/after\nImportant caveat\n\nMany OpenClaw runtimes snapshot skills/bootstrap per session. So skill/config slimming often does not fully apply to the current session. Use a new session for authoritative verification.\n\nReferences\nreferences/out-of-band-delivery.md\nreferences/cron-noise-checklist.md"
  },
  "trust": {
    "sourceLabel": "tencent",
    "provenanceUrl": "https://clawhub.ai/phenomenoner/context-clean-up",
    "publisherUrl": "https://clawhub.ai/phenomenoner/context-clean-up",
    "owner": "phenomenoner",
    "version": "1.0.7",
    "license": null,
    "verificationStatus": "Indexed source record"
  },
  "links": {
    "detailUrl": "https://openagent3.xyz/skills/context-clean-up",
    "downloadUrl": "https://openagent3.xyz/downloads/context-clean-up",
    "agentUrl": "https://openagent3.xyz/skills/context-clean-up/agent",
    "manifestUrl": "https://openagent3.xyz/skills/context-clean-up/agent.json",
    "briefUrl": "https://openagent3.xyz/skills/context-clean-up/agent.md"
  }
}