io.github.swapnanil/vectr

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_proactive_proxy.py: assigns a literal to secret (sk-a…23 (len 19)) — non-provider secret literal; prefer environment intake. Reads secrets from the environment (e.g. agent/hook_cli.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: api.py: references a sensitive credential path (/mcp.json) — the tool-poisoning exfiltration target. HIGH: integrations/vscode_bridge.py: references a sensitive credential path (/mcp.json) — the tool-poisoning exfiltration target. HIGH: main.py: references a sensitive credential path (/mcp.json) — the tool-poisoning exfiltration target. HIGH: tests/test_chunk_quality.py: references a sensitive credential path (/.cursor/) — the tool-poisoning exfiltration target. HIGH: tests/test_main.py: references a sensitive credential path (/mcp.json) — the tool-poisoning exfiltration target. HIGH: tests/test_mcp_stdio.py: references a sensitive credential path (/mcp.json) — the tool-poisoning exfiltration target. HIGH: shell/exec capability AND network egress together — download-and-execute risk (main.py: subprocess.* call; agent/instance_registry.py: outbound HTTP/socket call). HIGH: shell/exec capability AND destructive filesystem writes together (main.py: subprocess.* call; agent/chunk_quality.py: path traversal ('../') combined with file access). discloses shell/exec capability: main.py: subprocess.* call discloses shell/exec capability: tests/test_hint_language_live.py: subprocess.* call discloses shell/exec capability: tests/test_hook_import_cost.py: subprocess.* call discloses shell/exec capability: tests/test_mcp_server.py: subprocess.* call discloses shell/exec capability: tests/test_mcp_stdio.py: subprocess.* call discloses shell/exec capability: tests/test_prompt_templates.py: subprocess.* call discloses dynamic code execution: tests/test_integration.py: uses eval/exec/__import__ discloses dynamic code execution: tests/test_proactive_settings.py: uses eval/exec/__import__ discloses broad/destructive filesystem access: agent/chunk_quality.py: path traversal ('../') combined with file access discloses broad/destructive filesystem access: agent/instance_registry.py: destructive filesystem call (shutil.rmtree/os.remove/etc.) discloses broad/destructive filesystem access: main.py: open(…, 'w'/'a'/'x') write-mode file access discloses broad/destructive filesystem access: tests/test_chunk_quality.py: path traversal ('../') combined with file access discloses broad/destructive filesystem access: tests/test_watcher.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

  fail10% of grade

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

  vs prior 1.2.0 (minor bump): new capability primitive(s) in source: fs-write, http-client, net, subprocess; new hardcoded outbound host/BCC destination(s): 127.0.0.1, api.anthropic.com, central, central-host, daemon, h, host, proxy, upstream.

• Transport config

  warn10% of grade

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

  manifest/source references tracking behavior — disclosure