io.github.n24q02m/better-code-review-graph

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_embeddings.py: assigns a literal to api_key (expl…ey (len 12)) — non-provider secret literal; prefer environment intake. Reads secrets from the environment (e.g. migrations/versions/005_temporal_columns.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: tests/test_xpia.py: references a sensitive credential path (/.ssh/) — the tool-poisoning exfiltration target. HIGH: tests/test_xpia_adversarial.py: references a sensitive credential path (/.ssh/) — the tool-poisoning exfiltration target. HIGH: shell/exec capability AND network egress together — download-and-execute risk (scripts/preserve-diacritics.py: subprocess.* call; src/better_code_review_graph/relay_setup.py: outbound HTTP/socket call). HIGH: shell/exec capability AND destructive filesystem writes together (scripts/preserve-diacritics.py: subprocess.* call; src/better_code_review_graph/resolver/rust.py: path traversal ('../') combined with file access). discloses shell/exec capability: scripts/preserve-diacritics.py: subprocess.* call discloses shell/exec capability: src/better_code_review_graph/federation.py: subprocess.* call discloses shell/exec capability: src/better_code_review_graph/incremental.py: subprocess.* call discloses shell/exec capability: src/better_code_review_graph/tools.py: subprocess.* call discloses shell/exec capability: src/better_code_review_graph/security/semgrep_engine.py: subprocess.* call discloses shell/exec capability: tests/conftest_e2e.py: subprocess.* call discloses shell/exec capability: tests/test_commits_migration.py: subprocess.* call discloses shell/exec capability: tests/test_coverage_gaps.py: subprocess.* call discloses shell/exec capability: tests/test_e2e.py: subprocess.* call discloses shell/exec capability: tests/test_federation_smoke.py: subprocess.* call discloses shell/exec capability: tests/test_full_live.py: subprocess.* call discloses shell/exec capability: tests/test_heuristic.py: subprocess.* call with shell=True (command-injection prone) discloses shell/exec capability: tests/test_incremental.py: subprocess.* call discloses shell/exec capability: tests/test_incremental_extra.py: subprocess.* call discloses shell/exec capability: tests/test_live_mcp.py: subprocess.* call discloses shell/exec capability: tests/test_phase_3_smoke.py: subprocess.* call discloses shell/exec capability: tests/test_renamed_in_diff.py: subprocess.* call discloses shell/exec capability: tests/test_renamed_in_diff_exception.py: subprocess.* call discloses shell/exec capability: tests/test_reviewer_summary.py: subprocess.* call discloses shell/exec capability: tests/test_server.py: subprocess.* call discloses shell/exec capability: tests/test_stdio_direct.py: subprocess.* call discloses shell/exec capability: tests/test_temporal_migration.py: subprocess.* call discloses shell/exec capability: tests/test_v17_coverage_gaps.py: subprocess.* call discloses dynamic code execution: tests/test_heuristic.py: uses eval/exec/__import__ discloses dynamic code execution: tests/test_incremental_extra.py: uses eval/exec/__import__ discloses broad/destructive filesystem access: src/better_code_review_graph/resolver/rust.py: path traversal ('../') combined with file access discloses broad/destructive filesystem access: tests/conftest_e2e.py: destructive filesystem call (shutil.rmtree/os.remove/etc.) discloses broad/destructive filesystem access: tests/test_import_graph_dispatch.py: path traversal ('../') combined with file access discloses broad/destructive filesystem access: tests/test_resolver_go.py: path traversal ('../') combined with file access discloses broad/destructive filesystem access: tests/test_resolver_rust.py: path traversal ('../') combined with file access discloses broad/destructive filesystem access: tests/test_tools_full.py: destructive filesystem call (shutil.rmtree/os.remove/etc.) discloses broad/destructive filesystem access: tests/test_v17_coverage_gaps.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 3.20.0 (minor bump): new capability primitive(s) in source: dynamic-eval, http-client, subprocess; new hardcoded outbound host/BCC destination(s): 127.0.0.1, aistudio.google.com, better-code-review-graph.n24q02m.dev, cloud.google.com, crg.example, crg.example.com, crg.example.test, dashboard.cohere.com, env-embed, env-llm, example.com, gateway.example, github.com, graphml.graphdrawing.org, gw-a, gw-b, jina.ai, json.schemastore.org, maven.apache.org, platform.openai.com, proxy.example, relay.example.com.

• 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