io.github.agentkitai/lore

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.

  examples/redaction_demo.py: contains a hardcoded OpenAI API key (sk-a…no (len 30)) — in an example/sample file; verify it is not a real key. tests/test_edge_cases.py: contains a hardcoded OpenAI API key (sk-a…no (len 30)) — in an example/sample file; verify it is not a real key. tests/test_http_store.py: assigns a literal to api_key (lore…it (len 16)) — non-provider secret literal; prefer environment intake. tests/test_http_store.py: assigns a literal to api_key (lore…a0 (len 40)) — non-provider secret literal; prefer environment intake. tests/test_http_store.py: assigns a literal to api_key (lore…ue (len 19)) — non-provider secret literal; prefer environment intake. tests/test_http_store.py: assigns a literal to api_key (lore…st (len 12)) — non-provider secret literal; prefer environment intake. tests/test_ingest_adapters.py: assigns a literal to token (test…en (len 14)) — non-provider secret literal; prefer environment intake. tests/test_ingest_adapters.py: assigns a literal to secret (webh…et (len 14)) — non-provider secret literal; prefer environment intake. tests/test_progressive.py: assigns a literal to api_key (lore…st (len 12)) — non-provider secret literal; prefer environment intake. tests/test_redact.py: embeds a PRIVATE KEY block. (in an example/sample file — lower severity, but a real key here still leaks.) tests/test_redact.py: contains a hardcoded OpenAI API key (sk-a…12 (len 27)) — in an example/sample file; verify it is not a real key. tests/test_redact.py: contains a hardcoded JWT (eyJh…8U (len 92)) — in an example/sample file; verify it is not a real key. tests/test_redact.py: assigns a literal to secret (Zk9m…fG (len 32)) — non-provider secret literal; prefer environment intake. tests/test_redact.py: assigns a literal to secret (aB3c…2u (len 31)) — non-provider secret literal; prefer environment intake. tests/test_redact.py: assigns a literal to secret (X7kR…eF (len 31)) — non-provider secret literal; prefer environment intake. tests/test_redact.py: assigns a literal to secret (c2Vj…Q= (len 40)) — non-provider secret literal; prefer environment intake. tests/test_redact_integration.py: embeds a PRIVATE KEY block. (in an example/sample file — lower severity, but a real key here still leaks.) tests/test_redact_integration.py: contains a hardcoded OpenAI API key (sk-a…12 (len 27)) — in an example/sample file; verify it is not a real key. tests/test_redact_integration.py: contains a hardcoded JWT (eyJh…8U (len 92)) — in an example/sample file; verify it is not a real key. tests/persistence/test_redact_write.py: contains a hardcoded OpenAI API key (sk-a…12 (len 27)) — in an example/sample file; verify it is not a real key. ts/tests/redact.test.ts: embeds a PRIVATE KEY block. (in an example/sample file — lower severity, but a real key here still leaks.) ts/tests/redact.test.ts: contains a hardcoded OpenAI API key (sk-a…no (len 30)) — in an example/sample file; verify it is not a real key. ts/tests/redact.test.ts: contains a hardcoded JWT (eyJh…8U (len 92)) — in an example/sample file; verify it is not a real key. ts/tests/redact.test.ts: assigns a literal to secret (Zk9m…fG (len 32)) — non-provider secret literal; prefer environment intake. Reads secrets from the environment (e.g. hooks/lore-capture-end.sh) — the recommended intake shape.

• Permission scope

  warn15% of grade

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

  has both shell/exec and network egress — capable of download-and-run, though no such path was observed (src/lore/bootstrap.py: subprocess.* call; hooks/lore-precompact.ts: outbound HTTP/socket call). has both shell/exec and destructive filesystem writes (src/lore/bootstrap.py: subprocess.* call; benchmarks/run_benchmarks.py: open(…, 'w'/'a'/'x') write-mode file access). discloses shell/exec capability: 10 files (src/lore/bootstrap.py, src/lore/setup.py, src/lore/wrap.py, +7 more) discloses broad/destructive filesystem access: 12 files (benchmarks/run_benchmarks.py, benchmarks/run_benchmarks.py, src/lore/bootstrap.py, +9 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.agentkitai/[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 tracking behavior — disclosure