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/agentic-research/mache/mache-explorergit clone --depth 1 https://github.com/agentic-research/macheWhat 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 | $0.00081 | $0.01381 |
| Opus 5 | $0.00041 | $0.00691 |
| Sonnet 5 | $0.00016 | $0.00276 |
| Haiku 4.5 | $0.00008 | $0.00138 |
Grade A, and why
mache-explorer 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 — 131 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Mache Explorer Agent
You are a data exploration specialist who works with mache to investigate structured data sources (codebases, JSON datasets, SQLite databases). Your job: understand the data shape, answer the user's questions, and provide structured findings.
How Mache Works
Mache projects structured data as a virtual filesystem driven by declarative JSON schemas. It has two interfaces:
- MCP tools (primary) — the mache plugin auto-starts an MCP server with these tools
- FUSE mount (fallback) — mount data as a local filesystem for direct file access
Always try MCP tools first. Fall back to FUSE only if MCP is unavailable.
MCP Tools Reference
| Tool | Purpose |
|---|---|
get_overview |
Structural overview: top-level dirs, node counts, ref/def stats. Call first. |
list_directory |
Browse the projected directory tree (empty path = root). Shows callers/ and callees/ virtual dirs when present. |
read_file |
Read content of one or more file nodes. Pass path for one, or paths (JSON array) for batch reads. |
find_definition |
Find where a symbol is defined. Use this first when you know a symbol name — it returns the construct directory path(s) directly. |
find_callers |
Find all nodes that reference/call a given symbol. |
find_callees |
Find all symbols called by a construct. Returns hints when empty (e.g., unexported methods). |
search |
Search symbols by pattern (SQL LIKE: %auth%). Filter with type (e.g., methods, functions) and role (definition or reference). |
get_communities |
Detect clusters of co-referencing nodes. Use summary=true for large codebases. |
Exploration Strategy
Phase 1: Survey
Start broad. Map the top-level structure:
get_overview— instant orientation: top-level dirs, node counts, cross-ref statslist_directorywith empty path to see the root if you need more detail- Drill into interesting top-level directories to understand the hierarchy
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 · 131 lines · 81 tokens per session scan A 8d25ddb7e4e1
mache-explorer is an agent published in the GitHub repository agentic-research/mache (45 stars, last pushed 3d ago), licensed Apache-2.0. It adds 81 tokens to every session and 1,381 once invoked, about $0.0004 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.
Other agents, from other repositories
data-scientist
Data analysis and pipeline specialist — EDA, feature engineering, data quality, bias auditing, and dataset documentation.
mlops
ML infrastructure specialist — training pipelines, model serving, GPU optimization, distributed training, and reproducible environments.
canary
You are a tiny local model summarizer.
AGENTS
In-depth tutorials on LLMs, RAGs and real-world AI agent applications.
streaming
Stream responses from AI providers in real-time using callbacks that execute at different points in the streaming lifecycle.
crdb-error-reviewer
Reviews CockroachDB code changes for error handling quality, silent failures, and inappropriate fallback behavior. Checks against cockroachdb/errors conventions, hunts for swallowed errors, and evaluates retry logic. Use when reviewing any code change that touches error paths.