ham
HAM (Hierarchical Agent Memory) is a Claude Code skill that reduces AI context/token usage by replacing a single large CLAUDE.md with smaller, directory-scoped CLAUDE.md files and local memory files (e.g., .memory/decisions.md, patterns.md). It auto-scans a project and routes the agent to read only relevant context for the directory being worked in, and provides commands for setup, updating, routing, and local dashboard/analytics.
Score Breakdown
⚙ Agent Friendliness
🔒 Security
Based on provided content only: no authentication model is present, and the tool operates by writing context/decision files and reading local Claude Code session data. Risk considerations cannot be verified from supplied README (e.g., whether it handles secrets safely in generated memory, dependency CVEs, or the security posture of any local dashboard/front-end).
⚡ Reliability
Best When
You use Claude Code and your repo has a large or growing root CLAUDE.md/context that is repeatedly re-read across many coding prompts.
Avoid When
You cannot use/modify Claude Code skill files or prefer a model/tool-agnostic memory system rather than Claude Code-specific context routing.
Use Cases
- • Lower token usage and cost for Claude Code coding sessions by scoping context to the active directory
- • Improve agent speed by reducing irrelevant context loading
- • Maintain structured, hierarchical project conventions (root/global, per-directory, and per-subdirectory patterns)
- • Track estimated token/cost savings and directory context health via a local dashboard
Not For
- • Agents/tools other than Claude Code (e.g., Cursor/Copilot/Windsurf) are not supported per README
- • General-purpose memory storage or an API service for arbitrary applications (this is a local skill/workflow)
- • Teams needing centralized, authenticated multi-user memory storage as described under 'HAM Pro' (not assessed here)
Interface
Authentication
No authentication described in README; it appears to be a local skill that reads/writes files on the user's machine and reads Claude Code session JSONL locally.
Pricing
README does not describe pricing for the OSS skill. It mentions 'HAM Pro' with broader multi-agent support, but pricing details are not provided in the supplied content.
Agent Metadata
Known Gotchas
- ⚠ Skill is explicitly Claude Code-only; other coding assistants may not work
- ⚠ Installation involves cloning into ~/.claude/skills/ham or per-project paths; mismatched paths may prevent discovery
- ⚠ Commands like dashboard launch and launch.js auto-build/install dependencies on first run, which may affect automation
Alternatives
Full Evaluation Report
Comprehensive deep-dive: security analysis, reliability audit, agent experience review, cost modeling, competitive positioning, and improvement roadmap for ham.
AI-powered analysis · PDF + markdown · Delivered within 30 minutes
Package Brief
Quick verdict, integration guide, cost projections, gotchas with workarounds, and alternatives comparison.
Delivered within 10 minutes
Score Monitoring
Get alerted when this package's AF, security, or reliability scores change significantly. Stay ahead of regressions.
Continuous monitoring
Scores are editorial opinions as of 2026-03-30.