io.github.Patdolitse/piia-engram

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_anchor_live_smoke_runner.py: assigns a literal to api_key (quot…et (len 13)) — non-provider secret literal; prefer environment intake. tests/test_continuity_report.py: assigns a literal to secret (ZZ_C…DY (len 25)) — non-provider secret literal; prefer environment intake. tests/test_continuity_report.py: assigns a literal to secret (ZZ_R…ET (len 23)) — non-provider secret literal; prefer environment intake. tests/test_core.py: assigns a literal to secret (ZZ_C…ET (len 25)) — non-provider secret literal; prefer environment intake. tests/test_core.py: assigns a literal to secret (ZZ_D…ET (len 24)) — non-provider secret literal; prefer environment intake. tests/test_crypto.py: assigns a literal to secret (test…se (len 15)) — non-provider secret literal; prefer environment intake. tests/test_crypto.py: assigns a literal to secret (rota…st (len 13)) — non-provider secret literal; prefer environment intake. tests/test_crypto.py: assigns a literal to secret (unic…st (len 12)) — non-provider secret literal; prefer environment intake. tests/test_dock_resume.py: assigns a literal to secret (DOCK…ET (len 23)) — non-provider secret literal; prefer environment intake. tests/test_export_redaction.py: embeds a PRIVATE KEY block. (in an example/sample file — lower severity, but a real key here still leaks.) tests/test_governance_readpath_matrix.py: assigns a literal to SECRET (ZZSE…nt (len 47)) — non-provider secret literal; prefer environment intake. tests/test_governance_runtime.py: contains a hardcoded OpenAI API key (sk-p…u0 (len 39)) — in an example/sample file; verify it is not a real key. tests/test_hook_content_digest.py: embeds a PRIVATE KEY block. (in an example/sample file — lower severity, but a real key here still leaks.) tests/test_lifecycle_apply.py: assigns a literal to SECRET (ZZ_L…DY (len 30)) — non-provider secret literal; prefer environment intake. tests/test_management_actions.py: assigns a literal to SECRET (ZZ_M…ET (len 27)) — non-provider secret literal; prefer environment intake. tests/test_management_surface_readonly.py: assigns a literal to SECRET (ZZ_S…EN (len 23)) — non-provider secret literal; prefer environment intake. tests/test_management_view.py: assigns a literal to SECRET (ZZ_M…EN (len 26)) — non-provider secret literal; prefer environment intake. tests/test_mcp_tools.py: assigns a literal to secret (ZZ_M…ET (len 27)) — non-provider secret literal; prefer environment intake. tests/test_mcp_tools.py: assigns a literal to secret (ZZ_M…ET (len 24)) — non-provider secret literal; prefer environment intake. tests/test_mcp_tools.py: assigns a literal to secret (ZZ_M…ET (len 28)) — non-provider secret literal; prefer environment intake. tests/test_mcp_tools.py: assigns a literal to secret (ZZ_M…ET (len 21)) — non-provider secret literal; prefer environment intake. tests/test_merge_apply.py: assigns a literal to SECRET (ZZ_M…DY (len 26)) — non-provider secret literal; prefer environment intake. tests/test_reconcile_apply.py: assigns a literal to SECRET (ZZ_R…DY (len 30)) — non-provider secret literal; prefer environment intake. tests/test_release_auth_preflight.py: contains a hardcoded OpenAI API key (sk-S…nt (len 37)) — in an example/sample file; verify it is not a real key. tests/test_release_flow_recovery.py: assigns a literal to secret (gho_…NT (len 23)) — non-provider secret literal; prefer environment intake. tests/test_release_orchestrator.py: contains a hardcoded OpenAI API key (sk-S…nt (len 27)) — in an example/sample file; verify it is not a real key. tests/test_sanitize_check.py: embeds a PRIVATE KEY block. (in an example/sample file — lower severity, but a real key here still leaks.) tests/test_sanitize_check.py: assigns a literal to token (ghp_…34 (len 38)) — non-provider secret literal; prefer environment intake. tests/test_sensitivity.py: embeds a PRIVATE KEY block. (in an example/sample file — lower severity, but a real key here still leaks.) tests/test_sensitivity.py: contains a hardcoded JWT (eyJh…mN (len 65)) — in an example/sample file; verify it is not a real key. tests/test_setup_wizard.py: assigns a literal to secret (ZZST…DY (len 20)) — non-provider secret literal; prefer environment intake. tests/test_setup_wizard.py: assigns a literal to secret (ZZST…ET (len 20)) — non-provider secret literal; prefer environment intake. tests/test_setup_wizard.py: assigns a literal to secret (SECR…SE (len 20)) — non-provider secret literal; prefer environment intake. tests/test_setup_wizard.py: assigns a literal to secret (ZZ_C…ET (len 24)) — non-provider secret literal; prefer environment intake. tests/test_staging_review.py: assigns a literal to SECRET (ZZ_S…ET (len 24)) — non-provider secret literal; prefer environment intake. tests/test_user_portrait.py: assigns a literal to secret (SECR…23 (len 37)) — non-provider secret literal; prefer environment intake. Reads secrets from the environment (e.g. demos/cross_tool_continuity_demo.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/piia_engram/reconcile.py: references a credential path (/.cursor/) has both shell/exec and network egress — capable of download-and-run, though no such path was observed (src/piia_engram/cli_commands.py: subprocess.* call; src/piia_engram/dock_ui/static/app.js: outbound HTTP/socket call). has both shell/exec and destructive filesystem writes (src/piia_engram/cli_commands.py: subprocess.* call; demos/backup_roundtrip_harness.py: destructive filesystem call (shutil.rmtree/os.remove/etc.)). discloses shell/exec capability: 8 files (src/piia_engram/cli_commands.py, src/piia_engram/doctor.py, src/piia_engram/freshness_anchors.py, +5 more) discloses broad/destructive filesystem access: 8 files (demos/backup_roundtrip_harness.py, demos/client_ab_evidence_harness.py, demos/cross_tool_continuity_demo.py, +5 more) 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 4.19.1 (minor bump): new capability primitive(s) in source: child_process, dynamic-eval, fs-write, http-client, subprocess; new hardcoded outbound host/BCC destination(s): 0.0.0.0, 1.2.3.4, 10.0.0.1, 10.255.255.255, 127.0.0.1, 169.254.169.254, 172.16.0.1, 172.31.255.255, 192.168.0.100, 192.168.1.1, api.example.com, custom.example.com, docs.python.org, evil.example.com, evil.test, example.com, example.invalid, example.test, feedback.example.test, github.com, img.shields.io, internal.attacker.com, metadata-proxy.evil.com, metadata.google.internal, my-relay.example.test, nonexistent.invalid, nope.invalid, public.example.com, pypi.org, pypistats.org, python.org, registry.example, telemetry.example.test, x, x.com, youtu.be.

• 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