[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M73C1P75B7GZXNTVDRV5VBFM":3,"skill-same:01M73C1P75B7GZXNTVDRV5VBFM":47},{"skillNo":4,"name":5,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":17,"publishedAt":18,"linkOnly":19,"markdownContent":20,"markdownBody":21,"frontMatter":22,"refName":23,"commitSha":24,"skillDirectory":25,"forksCount":26,"repositoryHtmlUrl":27,"repositoryHomepageUrl":28,"ownerHtmlUrl":29,"ownerBlogUrl":30,"ownerDescription":31,"updatedAt":32,"install":33,"resources":46},"01M73C1P75B7GZXNTVDRV5VBFM","hermes-context-compressor",null,"Horonomos-laps\u002FYana-AI","Horonomos-laps","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F355299122?v=4","01M76B3QPXE88ET2SQPDPVATZT"," Horonomos-laps","01M72ZQRJPCYPSSWT4D00VJR2A",13,"Apache-2.0","https:\u002F\u002Fgithub.com\u002FHoronomos-laps\u002FYana-AI\u002Fblob\u002Fmain\u002Fcore\u002Fskills\u002Fhermes-context-compressor\u002FSKILL.md",[],[],"2026-11-11 15:11:14",false,"---\nname: hermes-context-compressor\ndescription: Multi-phase context compression — prune old tool results, protect head\u002Ftail, LLM summarize with structured template, iterative updates, graceful fallback. Pluggable engine, focus_topic, anti-thrash guard. Source: NousResearch\u002Fhermes-agent (MIT).\norigin: NousResearch\u002Fhermes-agent (MIT) — agent\u002Fcontext_compressor.py\nlicense: MIT\nversion: 1.0.0\ncompatibility: yana-ai >= 1.3.54\n---\n\n## Implementation (real, runnable — added 2026-06-19)\n\nThis skill is no longer prose-only. The real algorithm was read from\nhermes-agent's actual source (commit `5378b941209d8f62a65455041658ce8ce8144cc9`)\nand ported — not copied verbatim, since the original extends hermes'\n`ContextEngine` and calls hermes' multi-provider client — into:\n\n- Module: `core\u002Flib\u002Fhermes_adapted\u002Fcontext_compressor.py`\n  - Tool-pair safety + tail anchoring (fixes for incidents #10896, #29824 in\n    the original — keeps the active task and the last visible reply out of\n    the compressed region, and cleans up orphaned tool_call\u002Fresult pairs)\n    split into `core\u002Flib\u002Fhermes_adapted\u002Fcontext_compressor_pairs.py`\n- Tests:  `tests\u002Ftest_hermes_context_compressor.py` (6) +\n  `tests\u002Ftest_hermes_context_compressor_pairs.py` (9) — 15 passing\n- Provenance: `vendor\u002Fhermes-agent\u002F_upstream\u002Fcontext_compressor.py` (original, for reference)\n\nUse `ContextCompressor(context_length, cfg, summarize_fn)` — pass your own\n`summarize_fn: Callable[[str], Optional[str]]` to wire in whatever model\nYana AI uses for the auxiliary summary call.\n\n# \u002Fhermes-context-compressor\n\n## When to Use\n\n- Conversation context approaching model limit (75%+ of token window)\n- Agent sessions with many tool calls producing large outputs\n- Need to preserve continuity across compression without losing task state\n- Building a context engine for Yana AI or long-running mission runs\n\n## Do NOT use for\n\n- Single-turn completions (no prior context to compress)\n- Conversations under 10K tokens (overhead not worth it)\n- See also: [[hermes-system-prompt-tiers]] — system prompt is never compressed\n\n---\n\n## Four-Phase Compression Pipeline\n\n```\nPhase 1: Tool Result Pruning        ← cheap, no LLM\n  Replace old tool outputs with 1-line summaries\n  Deduplicate identical results (keep newest full copy)\n  Strip image payloads from computer_use screenshots\n  Truncate oversized tool args while preserving JSON validity\n\nPhase 2: Boundary Detection\n  Head: protect system prompt + first N messages (default 3)\n  Tail: protect last ~20K tokens (accumulated backward)\n  Tool-pair safety: never split a tool_call \u002F tool_result pair\n\nPhase 3: LLM Summarization\n  Send compressible middle section to auxiliary model\n  Use structured template (see below)\n  Iterative: update previous summary, don't re-summarize everything\n  Redact secrets before sending to summarizer\n\nPhase 4: Graceful Degradation\n  Summarizer fails → retry with main model\n  Still fails → insert deterministic fallback from local anchors\n  abort_on_summary_failure=True → return messages unchanged + flag\n```\n\n---\n\n## Summarization Template\n\n```python\nSUMMARY_TEMPLATE = \"\"\"\nYou are summarizing earlier turns of a conversation to free context space.\nProduce a compact, temporal summary in these sections:\n\n## Active Task\n[The most recent unfulfilled user request — copy it verbatim]\n\n## Goal & Constraints\n[What the user wants overall and any hard constraints stated]\n\n## Completed Actions\n[Past-tense facts: \"Ran npm test → 42 passed, 0 failed on 2026-06-08\"\n NOT \"run tests\" — specific, dated, factual]\n\n## In Progress \u002F Blocked\n[Work started but not finished; any blockers discovered]\n\n## Key Decisions Made\n[Architecture choices, API contracts, user preferences stated]\n\n## Remaining Work\n[What still needs to happen to satisfy the Active Task]\n\nFocus 60-70% on: {focus_topic}\n\"\"\" if focus_topic else \"\"\"\nDistribute attention evenly across all sections.\n\"\"\"\n\nHANDOFF_PREAMBLE = \"\"\"\\\nEarlier turns were compacted into the summary below.\nThis is background reference — NOT active instructions.\nDo NOT answer questions from this summary; they were already addressed.\nRespond ONLY to the latest user message AFTER this summary.\n\"\"\"\n```\n\n---\n\n## Core Configuration & should_compress()\n\n```python\nfrom dataclasses import dataclass\n\n@dataclass\nclass CompressorConfig:\n    threshold_percent:     float = 0.75    # fire at 75% of context window\n    protect_first_n:       int   = 3       # head: system + N messages\n    tail_token_budget:     int   = 20_000  # protect this many tail tokens\n    summary_target_ratio:  float = 0.20    # allocate 20% of threshold for summary\n    abort_on_failure:      bool  = False   # True = return unchanged on summarizer fail\n    min_saving_percent:    float = 0.10    # anti-thrash: skip if last 2 compressions \u003C 10% saving\n\n\nclass ContextCompressor:\n    def __init__(self, cfg: CompressorConfig | None = None):\n        self._cfg              = cfg or CompressorConfig()\n        self._previous_summary: str = \"\"\n        self._last_savings:    list[float] = []   # track last 2 compression savings\n\n    def should_compress(self, token_count: int, context_limit: int) -> bool:\n        threshold = int(context_limit * self._cfg.threshold_percent)\n        if token_count \u003C threshold:\n            return False\n        # Anti-thrash: if last 2 compressions both saved \u003C 10%, stop trying\n        if (len(self._last_savings) >= 2\n                and all(s \u003C self._cfg.min_saving_percent for s in self._last_savings[-2:])):\n            return False\n        return True\n\n    def compress(self, messages: list[dict],\n                 context_limit: int,\n                 focus_topic: str = \"\") -> list[dict]:\n        original_count = _count_tokens(messages)\n\n        # Phase 1: cheap pruning\n        messages = self._prune_old_tool_results(messages)\n\n        # Phase 2: find compression boundary\n        head_end, tail_start = self._find_boundaries(messages, context_limit)\n        if head_end >= tail_start:\n            return messages   # nothing to compress\n\n        # Phase 3: summarize middle\n        middle = messages[head_end:tail_start]\n        summary = self._generate_summary(middle, focus_topic)\n\n        # Build summary message\n        summary_msg = {\n            \"role\":    \"assistant\",\n            \"content\": f\"{HANDOFF_PREAMBLE}\\n\\n{summary}\",\n        }\n\n        result = messages[:head_end] + [summary_msg] + messages[tail_start:]\n\n        # Track savings for anti-thrash\n        new_count   = _count_tokens(result)\n        saving_pct  = 1 - (new_count \u002F original_count)\n        self._last_savings.append(saving_pct)\n        if len(self._last_savings) > 5:\n            self._last_savings.pop(0)\n\n        return result\n```\n\n---\n\n## Iterative summary updates\n\n```python\ndef _generate_summary(self, middle: list[dict], focus_topic: str) -> str:\n    prompt = SUMMARY_TEMPLATE.format(focus_topic=focus_topic) if focus_topic \\\n             else SUMMARY_TEMPLATE.replace(\"{focus_topic}\", \"\")\n\n    # Iterative: update previous summary rather than re-summarize\n    if self._previous_summary:\n        prompt = (\n            f\"Previous summary:\\n{self._previous_summary}\\n\\n\"\n            \"Update the summary above with the new turns below. \"\n            \"Keep completed work; add new findings.\\n\\n\" + prompt\n        )\n\n    summary = call_llm(prompt, messages=middle)\n    self._previous_summary = summary\n    return summary\n```\n\n---\n\n## Tool result pruning (Phase 1)\n\n```python\ndef _prune_old_tool_results(self, messages: list[dict]) -> list[dict]:\n    \"\"\"Replace old tool outputs with 1-line summaries. Deduplicate.\"\"\"\n    seen_hashes: dict[str, int] = {}   # result_hash → last index with full content\n    pruned = list(messages)\n\n    for i, msg in enumerate(pruned):\n        if msg.get(\"role\") != \"tool\": continue\n        content = msg.get(\"content\", \"\")\n        h = hashlib.sha256(content.encode()).hexdigest()[:16]\n\n        if h in seen_hashes:\n            # Duplicate — replace older with back-reference\n            prev_idx = seen_hashes[h]\n            lines    = content.count(\"\\n\") + 1\n            tool_name = msg.get(\"name\", \"tool\")\n            pruned[prev_idx] = {**pruned[prev_idx],\n                                \"content\": f\"[{tool_name}] (same as result #{i})\"}\n        seen_hashes[h] = i\n\n        # Also summarize large old results (not in tail protection zone)\n        if len(content) > 2000:\n            tool_name = msg.get(\"name\", \"tool\")\n            exit_hint = \"exit 0\" if \"exit 0\" in content else \"\"\n            lines     = content.count(\"\\n\") + 1\n            pruned[i] = {**msg,\n                         \"content\": f\"[{tool_name}] ran → {exit_hint} {lines} lines (pruned)\"}\n\n    return pruned\n```\n\n---\n\n## Anti-Fake-Pass Checklist\n\n```\n❌ No anti-thrash guard — compressor loops on already-minimal context, wasting tokens\n❌ Handoff preamble not injected — model re-executes stale instructions from summary\n❌ Tool call\u002Fresult pairs split by boundary detection — API rejects orphaned tool_call\n❌ Secrets sent to auxiliary summarizer model — summary leaks credentials\n❌ Previous summary not preserved — each compression loses historical context\n❌ threshold_percent set to 0.99 — compression fires too late, already truncating\n```\n","## Implementation (real, runnable — added 2026-06-19)\n\nThis skill is no longer prose-only. The real algorithm was read from\nhermes-agent's actual source (commit `5378b941209d8f62a65455041658ce8ce8144cc9`)\nand ported — not copied verbatim, since the original extends hermes'\n`ContextEngine` and calls hermes' multi-provider client — into:\n\n- Module: `core\u002Flib\u002Fhermes_adapted\u002Fcontext_compressor.py`\n  - Tool-pair safety + tail anchoring (fixes for incidents #10896, #29824 in\n    the original — keeps the active task and the last visible reply out of\n    the compressed region, and cleans up orphaned tool_call\u002Fresult pairs)\n    split into `core\u002Flib\u002Fhermes_adapted\u002Fcontext_compressor_pairs.py`\n- Tests:  `tests\u002Ftest_hermes_context_compressor.py` (6) +\n  `tests\u002Ftest_hermes_context_compressor_pairs.py` (9) — 15 passing\n- Provenance: `vendor\u002Fhermes-agent\u002F_upstream\u002Fcontext_compressor.py` (original, for reference)\n\nUse `ContextCompressor(context_length, cfg, summarize_fn)` — pass your own\n`summarize_fn: Callable[[str], Optional[str]]` to wire in whatever model\nYana AI uses for the auxiliary summary call.\n\n# \u002Fhermes-context-compressor\n\n## When to Use\n\n- Conversation context approaching model limit (75%+ of token window)\n- Agent sessions with many tool calls producing large outputs\n- Need to preserve continuity across compression without losing task state\n- Building a context engine for Yana AI or long-running mission runs\n\n## Do NOT use for\n\n- Single-turn completions (no prior context to compress)\n- Conversations under 10K tokens (overhead not worth it)\n- See also: [[hermes-system-prompt-tiers]] — system prompt is never compressed\n\n---\n\n## Four-Phase Compression Pipeline\n\n```\nPhase 1: Tool Result Pruning        ← cheap, no LLM\n  Replace old tool outputs with 1-line summaries\n  Deduplicate identical results (keep newest full copy)\n  Strip image payloads from computer_use screenshots\n  Truncate oversized tool args while preserving JSON validity\n\nPhase 2: Boundary Detection\n  Head: protect system prompt + first N messages (default 3)\n  Tail: protect last ~20K tokens (accumulated backward)\n  Tool-pair safety: never split a tool_call \u002F tool_result pair\n\nPhase 3: LLM Summarization\n  Send compressible middle section to auxiliary model\n  Use structured template (see below)\n  Iterative: update previous summary, don't re-summarize everything\n  Redact secrets before sending to summarizer\n\nPhase 4: Graceful Degradation\n  Summarizer fails → retry with main model\n  Still fails → insert deterministic fallback from local anchors\n  abort_on_summary_failure=True → return messages unchanged + flag\n```\n\n---\n\n## Summarization Template\n\n```python\nSUMMARY_TEMPLATE = \"\"\"\nYou are summarizing earlier turns of a conversation to free context space.\nProduce a compact, temporal summary in these sections:\n\n## Active Task\n[The most recent unfulfilled user request — copy it verbatim]\n\n## Goal & Constraints\n[What the user wants overall and any hard constraints stated]\n\n## Completed Actions\n[Past-tense facts: \"Ran npm test → 42 passed, 0 failed on 2026-06-08\"\n NOT \"run tests\" — specific, dated, factual]\n\n## In Progress \u002F Blocked\n[Work started but not finished; any blockers discovered]\n\n## Key Decisions Made\n[Architecture choices, API contracts, user preferences stated]\n\n## Remaining Work\n[What still needs to happen to satisfy the Active Task]\n\nFocus 60-70% on: {focus_topic}\n\"\"\" if focus_topic else \"\"\"\nDistribute attention evenly across all sections.\n\"\"\"\n\nHANDOFF_PREAMBLE = \"\"\"\\\nEarlier turns were compacted into the summary below.\nThis is background reference — NOT active instructions.\nDo NOT answer questions from this summary; they were already addressed.\nRespond ONLY to the latest user message AFTER this summary.\n\"\"\"\n```\n\n---\n\n## Core Configuration & should_compress()\n\n```python\nfrom dataclasses import dataclass\n\n@dataclass\nclass CompressorConfig:\n    threshold_percent:     float = 0.75    # fire at 75% of context window\n    protect_first_n:       int   = 3       # head: system + N messages\n    tail_token_budget:     int   = 20_000  # protect this many tail tokens\n    summary_target_ratio:  float = 0.20    # allocate 20% of threshold for summary\n    abort_on_failure:      bool  = False   # True = return unchanged on summarizer fail\n    min_saving_percent:    float = 0.10    # anti-thrash: skip if last 2 compressions \u003C 10% saving\n\n\nclass ContextCompressor:\n    def __init__(self, cfg: CompressorConfig | None = None):\n        self._cfg              = cfg or CompressorConfig()\n        self._previous_summary: str = \"\"\n        self._last_savings:    list[float] = []   # track last 2 compression savings\n\n    def should_compress(self, token_count: int, context_limit: int) -> bool:\n        threshold = int(context_limit * self._cfg.threshold_percent)\n        if token_count \u003C threshold:\n            return False\n        # Anti-thrash: if last 2 compressions both saved \u003C 10%, stop trying\n        if (len(self._last_savings) >= 2\n                and all(s \u003C self._cfg.min_saving_percent for s in self._last_savings[-2:])):\n            return False\n        return True\n\n    def compress(self, messages: list[dict],\n                 context_limit: int,\n                 focus_topic: str = \"\") -> list[dict]:\n        original_count = _count_tokens(messages)\n\n        # Phase 1: cheap pruning\n        messages = self._prune_old_tool_results(messages)\n\n        # Phase 2: find compression boundary\n        head_end, tail_start = self._find_boundaries(messages, context_limit)\n        if head_end >= tail_start:\n            return messages   # nothing to compress\n\n        # Phase 3: summarize middle\n        middle = messages[head_end:tail_start]\n        summary = self._generate_summary(middle, focus_topic)\n\n        # Build summary message\n        summary_msg = {\n            \"role\":    \"assistant\",\n            \"content\": f\"{HANDOFF_PREAMBLE}\\n\\n{summary}\",\n        }\n\n        result = messages[:head_end] + [summary_msg] + messages[tail_start:]\n\n        # Track savings for anti-thrash\n        new_count   = _count_tokens(result)\n        saving_pct  = 1 - (new_count \u002F original_count)\n        self._last_savings.append(saving_pct)\n        if len(self._last_savings) > 5:\n            self._last_savings.pop(0)\n\n        return result\n```\n\n---\n\n## Iterative summary updates\n\n```python\ndef _generate_summary(self, middle: list[dict], focus_topic: str) -> str:\n    prompt = SUMMARY_TEMPLATE.format(focus_topic=focus_topic) if focus_topic \\\n             else SUMMARY_TEMPLATE.replace(\"{focus_topic}\", \"\")\n\n    # Iterative: update previous summary rather than re-summarize\n    if self._previous_summary:\n        prompt = (\n            f\"Previous summary:\\n{self._previous_summary}\\n\\n\"\n            \"Update the summary above with the new turns below. \"\n            \"Keep completed work; add new findings.\\n\\n\" + prompt\n        )\n\n    summary = call_llm(prompt, messages=middle)\n    self._previous_summary = summary\n    return summary\n```\n\n---\n\n## Tool result pruning (Phase 1)\n\n```python\ndef _prune_old_tool_results(self, messages: list[dict]) -> list[dict]:\n    \"\"\"Replace old tool outputs with 1-line summaries. Deduplicate.\"\"\"\n    seen_hashes: dict[str, int] = {}   # result_hash → last index with full content\n    pruned = list(messages)\n\n    for i, msg in enumerate(pruned):\n        if msg.get(\"role\") != \"tool\": continue\n        content = msg.get(\"content\", \"\")\n        h = hashlib.sha256(content.encode()).hexdigest()[:16]\n\n        if h in seen_hashes:\n            # Duplicate — replace older with back-reference\n            prev_idx = seen_hashes[h]\n            lines    = content.count(\"\\n\") + 1\n            tool_name = msg.get(\"name\", \"tool\")\n            pruned[prev_idx] = {**pruned[prev_idx],\n                                \"content\": f\"[{tool_name}] (same as result #{i})\"}\n        seen_hashes[h] = i\n\n        # Also summarize large old results (not in tail protection zone)\n        if len(content) > 2000:\n            tool_name = msg.get(\"name\", \"tool\")\n            exit_hint = \"exit 0\" if \"exit 0\" in content else \"\"\n            lines     = content.count(\"\\n\") + 1\n            pruned[i] = {**msg,\n                         \"content\": f\"[{tool_name}] ran → {exit_hint} {lines} lines (pruned)\"}\n\n    return pruned\n```\n\n---\n\n## Anti-Fake-Pass Checklist\n\n```\n❌ No anti-thrash guard — compressor loops on already-minimal context, wasting tokens\n❌ Handoff preamble not injected — model re-executes stale instructions from summary\n❌ Tool call\u002Fresult pairs split by boundary detection — API rejects orphaned tool_call\n❌ Secrets sent to auxiliary summarizer model — summary leaks credentials\n❌ Previous summary not preserved — each compression loses historical context\n❌ threshold_percent set to 0.99 — compression fires too late, already truncating\n```\n",{},"main","481dcd7dd2e288b9d627a2bd017db70e64e6c6fa","core\u002Fskills\u002Fhermes-context-compressor",2,"https:\u002F\u002Fgithub.com\u002FHoronomos-laps\u002FYana-AI","https:\u002F\u002Fyana.vutam.link\u002F","https:\u002F\u002Fgithub.com\u002FHoronomos-laps","https:\u002F\u002Fyana.vutam.link","An open-source governed runtime & execution environment for multi-agent AI systems.","2026-11-14 13:24:09",{"source":7,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add Horonomos-laps\u002FYana-AI --skill hermes-context-compressor",{"tool":39,"command":40},"bunx","bunx skills add Horonomos-laps\u002FYana-AI --skill hermes-context-compressor",{"tool":42,"command":43},"pnpm","pnpm dlx skills add Horonomos-laps\u002FYana-AI --skill hermes-context-compressor","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FHoronomos-laps\u002FYana-AI\u002Fblob\u002Fmain\u002Fcore\u002Fskills\u002Fhermes-context-compressor\u002FSKILL.md",true,[],[48,56,63,71,79,87],{"skillNo":49,"name":50,"nameZh":6,"description":51,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":52,"licenseFileUrl":6,"sourceHtmlUrl":53,"categories":54,"keywords":55,"publishedAt":18},"01M735GKET5YVYKAY62W2P5YXB","openai--sentry--sentry","Use when the user asks to inspect Sentry issues or events, summarize recent production errors, or pull basic Sentry health data via the Sentry API; perform read-only queries with the bundled script and require `SENTRY_AUTH_TOKEN`.","MIT","https:\u002F\u002Fgithub.com\u002FHoronomos-laps\u002FYana-AI\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fopenai--sentry--sentry\u002FSKILL.md",[],[],{"skillNo":57,"name":58,"nameZh":6,"description":59,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":60,"categories":61,"keywords":62,"publishedAt":18},"01M73DCAP83T17HY7SMGK7MGFE","ito-basket-compare","Compare Itô prediction-market baskets against a user's knowledge base, portfolio notes, financial context, watchlist, or research thesis. Use for read-only basket comparison and gap analysis without investment advice or live trading.","https:\u002F\u002Fgithub.com\u002FHoronomos-laps\u002FYana-AI\u002Fblob\u002Fmain\u002Fcore\u002Fskills\u002Fito-basket-compare\u002FSKILL.md",[],[],{"skillNo":64,"name":65,"nameZh":6,"description":66,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":67,"sourceHtmlUrl":68,"categories":69,"keywords":70,"publishedAt":18},"01M734CYEMZFDR3KH6SW0JP8TD","implementing-network-traffic-baselining","Build network traffic baselines from NetFlow\u002FIPFIX data using Python pandas for statistical analysis, z-score anomaly detection, and hourly\u002Fdaily traffic pattern profiling","https:\u002F\u002Fgithub.com\u002Fyanacuti1121\u002FYana-AI\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fimplementing-network-traffic-baselining\u002FLICENSE","https:\u002F\u002Fgithub.com\u002FHoronomos-laps\u002FYana-AI\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fimplementing-network-traffic-baselining\u002FSKILL.md",[],[],{"skillNo":72,"name":73,"nameZh":6,"description":74,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":75,"sourceHtmlUrl":76,"categories":77,"keywords":78,"publishedAt":18},"01M73ASZ6NJRJM44Z2P84NGQ3Q","detecting-evasion-techniques-in-endpoint-logs","Detects defense evasion techniques used by adversaries in endpoint logs including log tampering, timestomping, process injection, and security tool disabling. Use when investigating suspicious endpoint behavior, building detection rules for evasion tactics, or conducting threat hunting for stealthy adversary activity. Activates for requests involving evasion detection, defense evasion analysis, log tampering detection, or MITRE ATT&CK TA0005.","https:\u002F\u002Fgithub.com\u002Fyanacuti1121\u002FYana-AI\u002Fblob\u002Fmain\u002Fcore\u002Fskills\u002Fdetecting-evasion-techniques-in-endpoint-logs\u002FLICENSE","https:\u002F\u002Fgithub.com\u002FHoronomos-laps\u002FYana-AI\u002Fblob\u002Fmain\u002Fcore\u002Fskills\u002Fdetecting-evasion-techniques-in-endpoint-logs\u002FSKILL.md",[],[],{"skillNo":80,"name":81,"nameZh":6,"description":82,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":83,"sourceHtmlUrl":84,"categories":85,"keywords":86,"publishedAt":18},"01M734AFNXD06Z78QRK0VKGT7Q","implementing-microsegmentation-with-guardicore","Implementing microsegmentation using Akamai Guardicore Segmentation to map application dependencies, create granular network policies, visualize east-west traffic flows, and enforce least-privilege communication between workloads across data centers and cloud.","https:\u002F\u002Fgithub.com\u002Fyanacuti1121\u002FYana-AI\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fimplementing-microsegmentation-with-guardicore\u002FLICENSE","https:\u002F\u002Fgithub.com\u002FHoronomos-laps\u002FYana-AI\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fimplementing-microsegmentation-with-guardicore\u002FSKILL.md",[],[],{"skillNo":88,"name":89,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":7,"ownerLogin":8,"ownerAvatarUrl":9,"ownerAccountNo":10,"ownerDisplayName":11,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":90,"categories":91,"keywords":92,"publishedAt":18},"01M736TK27QMCXK0DVCH25SJN5","ramdisk-tmpfs-patterns","https:\u002F\u002Fgithub.com\u002FHoronomos-laps\u002FYana-AI\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Framdisk-tmpfs-patterns\u002FSKILL.md",[],[]]