io.github.oxbshw/watch-skill

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/health/test_doctor.py: assigns a literal to secret (gsk-…ue (len 21)) — non-provider secret literal; prefer environment intake. tests/live/test_browser_policy.py: contains a hardcoded JWT (eyJh…VP (len 49)) — in an example/sample file; verify it is not a real key. Reads secrets from the environment (e.g. benchmarks/cost/run_cost_benchmark.py) — the recommended intake shape.

• Permission scope

  warn15% of grade

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

  references credential paths: src/watch_skill/operate/fixture_site.py: references a credential path (/.ssh/) installer script pipes a download into a shell — conventional, but worth reading before running (scripts/install.sh: pipes a download into a shell). has both shell/exec and network egress — capable of download-and-run, though no such path was observed (benchmarks/cost/run_cost_benchmark.py: subprocess.* call; app/src/preview.ts: outbound HTTP/socket call). has both shell/exec and destructive filesystem writes (benchmarks/cost/run_cost_benchmark.py: subprocess.* call; src/watch_skill/config.py: destructive filesystem call (shutil.rmtree/os.remove/etc.)). discloses shell/exec capability: 42 files (benchmarks/cost/run_cost_benchmark.py, examples/08-loop-types/monitor_loop.py, examples/08-loop-types/render_title_card.py, +39 more) discloses dynamic code execution: src/watch_skill/vision/worker_main.py: uses eval()/exec() discloses broad/destructive filesystem access: 9 files (src/watch_skill/config.py, src/watch_skill/identity.py, src/watch_skill/acquire/cache.py, +6 more) 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.oxbshw/[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); sends telemetry to third-party SDK (amplitude) — disclosure, verify opt-in; manifest/source references Telemetry behavior — disclosure