一键导入
memory-compact
Trigger ContextZipper compression on Morphic-Agent's L1-L4 semantic memory. Compacts old episodic entries, promotes frequently-accessed facts.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
菜单
Trigger ContextZipper compression on Morphic-Agent's L1-L4 semantic memory. Compacts old episodic entries, promotes frequently-accessed facts.
用 Codex 或 Claude 帮你安装 复制这段 Prompt,粘贴到 Codex、Claude 或其他助手里,让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Verify a feature spec, plan, or PR against the non-negotiable principles in `.specify/memory/constitution.md`.
Extract session insights after a task completes and feed them into Morphic's self-evolution engine. Captures what worked, what didn't, and proposes prompt/routing tweaks.
Keep CLAUDE.md (thin router) and AGENTS.md (telegraph-style) in sync. Detects drift in command lists, paths, and routing tables between the two root instruction files.
Generate a monthly or on-demand cost report across LLM providers, local LLMs, and agent engines. Flags overspend and suggests downgrades.
Run live end-to-end verification across Morphic-Agent's 6 agent CLI engines. Produces a baseline comparison table with duration and cost per engine.
Inspect Morphic-Agent's fractal task graph for a given task_id. Reports bypass classifier decisions, Gate 2 skips, parallelism, and regression flags.
| name | memory-compact |
| description | Trigger ContextZipper compression on Morphic-Agent's L1-L4 semantic memory. Compacts old episodic entries, promotes frequently-accessed facts. |
| when_to_use | When semantic memory hits the Ebbinghaus forgetting threshold, when L1 context zipper ratio drops below 0.5, or proactively before long-running tasks. |
| argument-hint | [--layer L1|L2|L3|L4|all] [--dry-run] |
| allowed-tools | ["Read","Grep","Glob","Bash(morphic memory compact *)","Bash(morphic memory stats *)"] |
| model | sonnet |
You orchestrate compression of Morphic's semantic memory hierarchy.
| Layer | Scope | Default retention |
|---|---|---|
| L1 | Working / current task | session |
| L2 | Session summary | 7 days |
| L3 | Episodic / task logs | 30 days |
| L4 | Semantic / distilled knowledge | indefinite |
$ARGUMENTS for --layer and --dry-run.memory-archaeologist subagent to check stats:
morphic memory stats --json
--dry-run, report projected compression ratio without writing.morphic memory compact --layer <L> --execute
# Memory Compact — <date>
## Before
- L1: 142 entries, 98 KB
- L2: 38 entries, 210 KB
- L3: 1,204 entries, 4.1 MB
- L4: 18,422 fingerprints, 45 MB
## After
- L1: 32 entries, 22 KB (0.22× )
- L2: 12 entries, 68 KB (0.32×)
- L3: 942 entries, 2.8 MB (0.68×)
- L4: 16,108 fingerprints, 38 MB (0.84×)
## Retained in L4 (promoted from L3)
- <N> patterns
morphic memory stats reports corruption.