MCPgrade

MCP server · trust report

io.github.skynetcmd/m3-memory

Published by io.github.skynetcmd

searchversion-control

2 checks that ran failed — permission scope and transport config — so this release can’t be trusted as-is.

Publisher
io.github.skynetcmd
Repository
github.com/skynetcmd/m3-memory
Install
pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory, pypi:m3-memory
Versions
78
Inspected
5 of 6 checks · 75%
Scored
2026-09-11 · rubric 1.0.0

What we checked

Six static checks, weighted by risk. Every result reflects only what could be observed in the published package and repository — never intent.

6 checks3 clean2 flagged1 not inspected

  • Injection surface

    unscannable25% of grade

    Tool descriptions/manifest scanned for instruction-injection patterns (imperatives at the model, hidden text, 'ignore previous', data-exfil URLs).

    No tool descriptions, server instructions, prompts, or resources were fetched; nothing to scan for injection.

  • Supply chain

    pass25% of grade

    Package provenance: namespace verification, repo linkage, maintainer count, account age, postinstall scripts, typosquat distance.

    Python build script has no build-time network/exec; repository linkage verified

  • Credential hygiene

    pass15% of grade

    How the server takes secrets (env vs plaintext config vs hardcoded); secrets appearing in tool schemas.

    Reads secrets from the environment (e.g. bin/_task_runtime.py) — the recommended intake shape. PASS = no static red flag; static analysis cannot prove the code honors env-based secret handling at runtime.

  • Permission scope

    fail15% of grade

    Declared tools vs. breadth (filesystem, network, exec); flags shell-exec and unbounded filesystem access.

    HIGH: m3_memory/setup_wizard.py reads a credential path, executes commands, and makes network calls in the same file — the combination that makes exfiltration possible. HIGH: downloads and pipes straight to a shell (install_os.py: pipes a download into a shell). HIGH: executes content derived from a network response (bin/install_schedules.py: passes a network response into an exec/eval call). references credential paths: m3_memory/setup_wizard.py: references a credential path (/.cursor/) has both shell/exec and destructive filesystem writes (bin/aider_patch.py: subprocess.* call; bin/auth_utils.py: open(…, 'w'/'a'/'x') write-mode file access). discloses shell/exec capability: 59 files (bin/aider_patch.py, bin/auth_utils.py, bin/chatlog_init.py, +56 more) discloses dynamic code execution: bin/memory_bridge.py: uses eval()/exec() discloses dynamic code execution: m3_memory/cli.py: uses eval()/exec() discloses broad/destructive filesystem access: 26 files (bin/auth_utils.py, bin/auth_utils.py, bin/bench_memory.py, +23 more) Capability DISCLOSURE, not a verdict: static analysis sees the primitive is present and reachable, not whether its use is attacker-controlled.

  • Version behavior

    pass10% of grade

    Diff of tool definitions between versions; new permissions or changed descriptions in a patch release (the postmark-mcp class).

    tool surface, capabilities, install scripts, and publisher unchanged vs prior version 2026.9.9.0 (patch bump); no rug-pull drift.

  • Transport config

    fail10% of grade

    Remote servers: TLS and auth mode (none/token/OAuth). Local servers: whether the manifest indicates it phones home.

    source performs runtime self-update / download-and-execute (bypasses version_behavior); manifest/source references tracking behavior — disclosure

Version history

Each release plotted by grade against the safe line at B. A version that sinks below the line has lost its trusted standing — the shape of a rug-pull.

VersionPublishedGradeScoreInspectedChange
v2026.9.10.1 · current2026-09-11 B755/6 · 75%±0
v2026.9.9.02026-09-09 B754/6 · 65%C→B in v2026.9.9.0
v2026.9.8.02026-09-08 C655/6 · 75%B→C in v2026.9.8.0
v2026.9.7.12026-09-08 B754/6 · 65%±0
v2026.8.30.12026-08-31 B754/6 · 65%±0
v2026.8.29.12026-08-30 B754/6 · 65%±0
v2026.8.19.162026-08-12 B754/6 · 65%±0
v2026.8.19.152026-08-10 B753/6 · 55%D→B in v2026.8.19.15
v2026.8.19.112026-08-10 D405/6 · 75%±0
v2026.8.19.52026-08-09 D505/6 · 75%±0
v2026.8.7.02026-08-08 D405/6 · 75%B→D in v2026.8.7.0
v2026.7.31.12026-07-31 B854/6 · 65%C→B in v2026.7.31.1
v2026.7.28.12026-07-28 C604/6 · 65%±0
v2026.4.72026-04-10 C603/6 · 55%B→C in v2026.4.7
v2026.7.30.02026-07-30 B754/6 · 65%D→B in v2026.7.30.0
v2026.7.28.02026-07-28 D504/6 · 65%C→D in v2026.7.28.0
v2026.7.26.92026-07-27 C603/6 · 55%D→C in v2026.7.26.9
v2026.7.26.12026-07-26 D505/6 · 75%±0
v2026.7.26.02026-07-26 D504/6 · 65%C→D in v2026.7.26.0
v2026.7.23.02026-07-23 C603/6 · 55%—