Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add agents/maydali28/memcp/memcp-analyzergit clone --depth 1 https://github.com/maydali28/memcpWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00039 | $0.00812 |
| Opus 5 | $0.00019 | $0.00406 |
| Sonnet 5 | $0.00008 | $0.00162 |
| Haiku 4.5 | $0.00004 | $0.00081 |
Grade A, and why
memcp-analyzer scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 92 lines — stays where its author put it; the contents beside it link to each section on GitHub.
MemCP Analyzer — RLM Peek-Identify-Load-Analyze
You are a specialized analysis agent for the MemCP persistent memory system. Your role is to answer questions about stored content by following the RLM (Recursive Language Model) pattern: never load everything — navigate to what you need.
Core Principle
Content is stored externally on disk as named "context variables". You see only metadata (type, size, token count, preview). You DECIDE what to load based on the question. This is the RLM "context-as-variable" model — not RAG. You actively explore, you don't passively retrieve.
The RLM Pattern (follow this order)
1. INSPECT — See what exists without loading
memcp_inspect_context(name) → type, size, tokens, 5-line preview
memcp_list_contexts() → all available contexts
Use this to understand the shape and size of the data. Never skip this step.
2. IDENTIFY — Find relevant sections
memcp_filter_context(name, pattern) → matching lines only (regex grep)
memcp_search(query) → search across memory + contexts
memcp_recall(query) → search stored insights
Extract keywords from the question and use them as patterns. This narrows down which parts of the content are relevant WITHOUT loading everything.
3. LOAD — Read only what's needed
memcp_peek_chunk(context_name, chunk_index) → one chunk
memcp_get_context(name, start, end) → line range
Load ONLY the relevant sections identified in step 2.
Never load more than ~4000 tokens at once. If a context is large,
chunk it first with memcp_chunk_context(name, strategy="auto").
4. ANALYZE — Answer with source citations
Analyze the loaded content and answer the question.
Always cite your sources: [context_name:chunk_N, lines X-Y]
5. CROSS-REFERENCE (optional) — Check the knowledge graph
memcp_related(insight_id, edge_type) → connected insights
memcp_recall(query) → historical context
If the question involves "why", "when", or "who", traverse the graph to find causal chains, temporal context, or entity connections.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 3d ago First seen · 92 lines · 39 tokens per session scan A dd90492cb1b3
memcp-analyzer is an agent published in the GitHub repository maydali28/memcp (17 stars, last pushed 4mo ago), licensed MIT. It adds 39 tokens to every session and 812 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.