io.github.ashlesh-t/cognirepo

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

  pass25% of grade

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

  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. cli/cli_config.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: adapters/openai_spec.py: references a sensitive credential path (/mcp.json) — the tool-poisoning exfiltration target. HIGH: cli/init_project.py: references a sensitive credential path (/mcp.json) — the tool-poisoning exfiltration target. HIGH: cli/main.py: references a sensitive credential path (/mcp.json) — the tool-poisoning exfiltration target. HIGH: cli/wizard.py: references a sensitive credential path (/mcp.json) — the tool-poisoning exfiltration target. HIGH: tests/test_cursor_vscode.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 (cli/env_wizard.py: subprocess.* call; cli/key_probes.py: outbound HTTP/socket call). HIGH: shell/exec capability AND destructive filesystem writes together (cli/env_wizard.py: subprocess.* call; cli/init_project.py: destructive filesystem call (shutil.rmtree/os.remove/etc.)). discloses shell/exec capability: cli/env_wizard.py: subprocess.* call discloses shell/exec capability: cli/seed.py: subprocess.* call discloses shell/exec capability: cli/wizard.py: subprocess.* call discloses shell/exec capability: cli/wizard.py: os.system/os.popen/os.exec* call discloses shell/exec capability: indexer/ast_indexer.py: subprocess.* call discloses shell/exec capability: indexer/doc_ingester.py: subprocess.* call discloses shell/exec capability: server/mcp_server.py: subprocess.* call discloses shell/exec capability: tests/test_cli_main_extended.py: subprocess.* call discloses shell/exec capability: tests/test_coverage_boost.py: subprocess.* call discloses shell/exec capability: tests/test_init_seed.py: subprocess.* call discloses shell/exec capability: tests/test_post_release.py: subprocess.* call discloses shell/exec capability: tests/test_seed_extended2.py: subprocess.* call discloses shell/exec capability: tools/benchmark.py: subprocess.* call discloses shell/exec capability: tools/bg_progress.py: subprocess.* call discloses shell/exec capability: tools/git_utils.py: subprocess.* call discloses dynamic code execution: cli/main.py: uses eval/exec/__import__ discloses dynamic code execution: tests/test_benchmark_metrics.py: uses eval/exec/__import__ discloses dynamic code execution: tests/test_context_builder.py: uses eval/exec/__import__ discloses dynamic code execution: tests/test_fallback_chain.py: uses eval/exec/__import__ discloses dynamic code execution: tests/test_local_adapter.py: uses eval/exec/__import__ discloses dynamic code execution: tests/test_multilang_indexer.py: uses eval/exec/__import__ discloses dynamic code execution: tests/test_org_graph.py: uses eval/exec/__import__ discloses dynamic code execution: tests/test_tier1_dogfood.py: uses eval/exec/__import__ discloses broad/destructive filesystem access: cli/init_project.py: destructive filesystem call (shutil.rmtree/os.remove/etc.) discloses broad/destructive filesystem access: cli/main.py: destructive filesystem call (shutil.rmtree/os.remove/etc.) discloses broad/destructive filesystem access: graph/org_graph.py: open(…, 'w'/'a'/'x') write-mode file access discloses broad/destructive filesystem access: graph/org_graph.py: destructive filesystem call (shutil.rmtree/os.remove/etc.) discloses broad/destructive filesystem access: indexer/ast_indexer.py: destructive filesystem call (shutil.rmtree/os.remove/etc.) discloses broad/destructive filesystem access: memory/episodic_memory.py: open(…, 'w'/'a'/'x') write-mode file access discloses broad/destructive filesystem access: server/mcp_server.py: open(…, 'w'/'a'/'x') write-mode file access discloses broad/destructive filesystem access: server/mcp_server.py: destructive filesystem call (shutil.rmtree/os.remove/etc.) discloses broad/destructive filesystem access: tests/test_config_orgs_extended.py: open(…, 'w'/'a'/'x') write-mode file access discloses broad/destructive filesystem access: tests/test_org_graph.py: open(…, 'w'/'a'/'x') write-mode file access discloses broad/destructive filesystem access: tests/test_prune_memory_extended2.py: open(…, 'w'/'a'/'x') write-mode file access discloses broad/destructive filesystem access: tools/sync_claude_memory.py: path traversal ('../') combined with file access discloses broad/destructive filesystem access: vector_db/local_vector_db.py: open(…, 'w'/'a'/'x') write-mode file access discloses broad/destructive filesystem access: vector_db/local_vector_db.py: destructive filesystem call (shutil.rmtree/os.remove/etc.) 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.ashlesh-t/[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

  warn10% of grade

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

  manifest/source references heartbeat behavior — disclosure