io.github.Intuition-Lab/personal-model

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.

• 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

  fail25% of grade

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

  Python build script executes at build time: network call in build script

• Credential hygiene

  warn15% of grade

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

  tests/test_doctor.py: assigns a literal to token (loca…n- (len 12)) — non-provider secret literal; prefer environment intake. tests/test_storage_security.py: assigns a literal to secret (zzzz…zz (len 37)) — non-provider secret literal; prefer environment intake. tests/test_storage_security.py: assigns a literal to secret (zzzz…zz (len 27)) — non-provider secret literal; prefer environment intake. Reads secrets from the environment (e.g. pyi_entrypoint.py) — the recommended intake shape.

• Permission scope

  fail15% of grade

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

  HIGH: shell/exec capability AND network egress together — download-and-execute risk (scripts/build_pypi_dist.py: subprocess.* call; resources/model_assets/viewer.js: outbound HTTP/socket call). HIGH: shell/exec capability AND destructive filesystem writes together (scripts/build_pypi_dist.py: subprocess.* call; src/persome/cli.py: destructive filesystem call (shutil.rmtree/os.remove/etc.)). discloses shell/exec capability: scripts/build_pypi_dist.py: subprocess.* call discloses shell/exec capability: src/persome/cli.py: subprocess.* call discloses shell/exec capability: src/persome/launchagent.py: subprocess.* call discloses shell/exec capability: src/persome/onboarding.py: subprocess.* call discloses shell/exec capability: src/persome/runtime_pid.py: subprocess.* call discloses shell/exec capability: src/persome/updater.py: subprocess.* call discloses shell/exec capability: src/persome/capture/ax_capture.py: subprocess.* call discloses shell/exec capability: src/persome/capture/browser_detect.py: subprocess.* call discloses shell/exec capability: src/persome/capture/ocr_subprocess.py: subprocess.* call discloses shell/exec capability: src/persome/capture/screen_state.py: subprocess.* call discloses shell/exec capability: src/persome/capture/watcher.py: subprocess.* call discloses shell/exec capability: src/persome/capture/window_meta.py: subprocess.* call discloses shell/exec capability: src/persome/capture/window_screenshot.py: subprocess.* call discloses shell/exec capability: src/persome/model/manifest.py: subprocess.* call discloses shell/exec capability: tests/test_daemon_teardown.py: subprocess.* call discloses shell/exec capability: tests/test_model_layout_asset.py: subprocess.* call discloses shell/exec capability: tests/test_ocr_subprocess.py: subprocess.* call discloses shell/exec capability: tests/test_updater.py: subprocess.* call discloses dynamic code execution: src/persome/cli.py: uses eval/exec/__import__ discloses dynamic code execution: src/persome/api/routes.py: uses eval/exec/__import__ discloses dynamic code execution: src/persome/evomem/edge_audit.py: uses eval/exec/__import__ discloses dynamic code execution: tests/test_classifier_with_fake_llm.py: uses eval/exec/__import__ discloses broad/destructive filesystem access: src/persome/cli.py: destructive filesystem call (shutil.rmtree/os.remove/etc.) discloses broad/destructive filesystem access: src/persome/doctor.py: destructive filesystem call (shutil.rmtree/os.remove/etc.) discloses broad/destructive filesystem access: src/persome/updater.py: destructive filesystem call (shutil.rmtree/os.remove/etc.) discloses broad/destructive filesystem access: src/persome/capture/ax_capture.py: destructive filesystem call (shutil.rmtree/os.remove/etc.) discloses broad/destructive filesystem access: src/persome/capture/ax_capture.py: path traversal ('../') combined with file access discloses broad/destructive filesystem access: src/persome/store/files.py: destructive filesystem call (shutil.rmtree/os.remove/etc.) discloses broad/destructive filesystem access: tests/test_cmux_source.py: destructive filesystem call (shutil.rmtree/os.remove/etc.) discloses broad/destructive filesystem access: tests/test_onboarding.py: path traversal ('../') combined with file access Capability DISCLOSURE, not a verdict: static analysis sees the primitive is present and reachable, not whether its use is attacker-controlled.

• Version behavior

  unscannable10% of grade

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

  no prior version to diff (first sight of io.github.Intuition-Lab/[email protected]); version-drift / rug-pull cannot be evaluated. Cold-start risk is covered by point-in-time checks (injection_surface, supply_chain, credential_hygiene), not here.

• 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 heartbeat behavior — disclosure