WavescopeSAFE
Wavelet-based multi-resolution context management for LLMs via MCP
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
[](https://safeskill.dev/scan/yogthos-wavescope-mcp) Wavelet-based multi-resolution context management for LLMs via MCP.
Provides a zoomable view of code files — the model can look at high-level structure, then drill down to specific regions as needed, without loading entire files into context.
How it works
- Structural signal — each line gets an importance score from the
keywords it declares (class, def, export, etc.), with comments and blank lines scoring zero. Indentation attenuates the score rather than adding to it: a line nested eight levels deep is less of a landmark than a top-level declaration, and a line that declares nothing stays near zero at any depth.
- Ricker wavelet transform — the signal is convolved with Ricker (Mexican
hat) wavelets at 8 scales (1–128), detecting structural boundaries at multiple resolutions.
- Multi-scale peaks — coefficient maxima are extracted and sorted by
magnitude, identifying the most important structural transitions. A coefficient is an integral over a window, so its maximum often falls between two declarations rather than on either; reported positions are snapped to the nearest line that is both a local signal maximum and a meaningful share of the structure around it, so they land on code you can navigate to rather than on a blank line or a stray }.
- Band assembly — three zoom levels, sized as a fraction of the query
radius (default 300):
- Fine (scales 1–2): raw lines in ±
radius/5(≥10 lines) - Medium (scales 4–16): function/class signatures in ±
radius/2 - Coarse (scales 32–128): section-level structural summary across ±
radius
See here for a more detailed explanation.
Installation
From npm (recommended)
npm install -g waves
d89af6fb44dfOBSERVED · 2026-10-08Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control.
claude mcp add wavescope-mcp -- npx -y [email protected]
{
"mcpServers": {
"wavescope-mcp": {
"command": "npx",
"args": [
"-y",
"[email protected]"
]
}
}
}Exposed tools (13)
11 read · 2 write · 0 destructive.
| Tool | Risk | Description |
|---|---|---|
diff_wavelet_context | read | |
get_complexity_heatmap | read | |
get_cursor_context | read | |
get_cursor_important_positions | read | |
get_entropy_bands | read | |
get_important_positions | read | |
get_summary_at_scale | read | |
get_wavelet_coefficients | read | |
query_wavelet_context | read | |
stream_close | read | |
stream_poll | read | |
stream_start | write | |
update_cursor_position | write |
Trust audit
SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | PASS |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | PASS |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- none-observed
- Network
- none-observed
- Shell
- none-observed
- Dependencies
- not all pinned
- Secrets in source
- none-found
Findings (3)
tryReadFileAtRef(repoRoot, "/etc/passwd", "HEAD"),
@modelcontextprotocol/sdk, zod, @types/node, @vitest/coverage-v8, tsx, typescript, vite, vitest
Gates applied: no_behavioural_pass, no_license.
d89af6fb44dffull audit observations/trust-audit/mcp-server/yogthos__wavescope.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-08 | d89af6fb44df | SAFE | B | 89 | first audit |
Questions
What is the Wavescope MCP server?
Wavelet-based multi-resolution context management for LLMs via MCP
What tools does Wavescope expose?
13 in total: 11 read-only, 2 that write, and 0 that can delete or overwrite. Every one is listed on this page with its risk.
Is Wavescope safe to connect to an agent?
The audit found nothing in the source that contradicts what it says it does, and graded it B (89/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean server reads B.
What credentials does Wavescope need?
No credential environment variables were found in its source, so it appears to need none.
How does Wavescope run?
It speaks stdio, so it runs as a local process your client starts. It is published on npm as wavescope-mcp at 1.0.14.
How current is this page?
The grade is for one exact copy of the source (d89af6fb44df), read on 2026-10-08. The repository is watched and re-audited when it changes.