io.github.msaad00/agent-bom

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

  fail15% of grade

  How the server takes secrets (env vs plaintext config vs hardcoded); secrets appearing in tool schemas.

  src/agent_bom/red_team.py: contains a hardcoded OpenAI API key (sk-p…tu (len 47)). Reads secrets from the environment (e.g. src/agent_bom/ai_enrich.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: src/agent_bom/cloud/aws.py: references a sensitive credential path (.aws/credentials) — the tool-poisoning exfiltration target. HIGH: src/agent_bom/constants.py: references a sensitive credential path (id_rsa) — the tool-poisoning exfiltration target. HIGH: src/agent_bom/discovery/__init__.py: references a sensitive credential path (/mcp.json) — the tool-poisoning exfiltration target. HIGH: src/agent_bom/discovery/dynamic.py: references a sensitive credential path (/mcp.json) — the tool-poisoning exfiltration target. HIGH: src/agent_bom/proxy.py: references a sensitive credential path (id_rsa) — the tool-poisoning exfiltration target. HIGH: shell/exec capability AND network egress together — download-and-execute risk (src/agent_bom/cli/_analysis.py: subprocess.* call; src/agent_bom/agent_identity.py: outbound HTTP/socket call). HIGH: shell/exec capability AND destructive filesystem writes together (src/agent_bom/cli/_analysis.py: subprocess.* call; src/agent_bom/api/audit_log.py: destructive filesystem call (shutil.rmtree/os.remove/etc.)). discloses shell/exec capability: src/agent_bom/cli/_analysis.py: subprocess.* call discloses shell/exec capability: src/agent_bom/cli/_common.py: subprocess.* call discloses shell/exec capability: src/agent_bom/cloud/coreweave.py: subprocess.* call discloses shell/exec capability: src/agent_bom/cloud/gpu_infra.py: subprocess.* call discloses shell/exec capability: src/agent_bom/cloud/nebius.py: subprocess.* call discloses shell/exec capability: src/agent_bom/discovery/__init__.py: subprocess.* call discloses shell/exec capability: src/agent_bom/filesystem.py: subprocess.* call discloses shell/exec capability: src/agent_bom/guard.py: subprocess.* call discloses shell/exec capability: src/agent_bom/image.py: subprocess.* call discloses shell/exec capability: src/agent_bom/integrity.py: subprocess.* call discloses shell/exec capability: src/agent_bom/k8s.py: subprocess.* call discloses shell/exec capability: src/agent_bom/parsers/os_parsers.py: subprocess.* call discloses shell/exec capability: src/agent_bom/parsers/python_parsers.py: subprocess.* call discloses shell/exec capability: src/agent_bom/resolvers/runtime_resolver.py: subprocess.* call discloses shell/exec capability: src/agent_bom/sast.py: subprocess.* call discloses dynamic code execution: src/agent_bom/cli/_inventory.py: uses eval/exec/__import__ discloses dynamic code execution: src/agent_bom/cloud/clickhouse.py: uses eval/exec/__import__ discloses dynamic code execution: src/agent_bom/cloud/gcp.py: uses eval/exec/__import__ discloses dynamic code execution: src/agent_bom/cloud/gcp_cis_benchmark.py: uses eval/exec/__import__ discloses broad/destructive filesystem access: src/agent_bom/api/audit_log.py: destructive filesystem call (shutil.rmtree/os.remove/etc.) discloses broad/destructive filesystem access: src/agent_bom/api/fleet_store.py: destructive filesystem call (shutil.rmtree/os.remove/etc.) discloses broad/destructive filesystem access: src/agent_bom/api/routes/scan.py: path traversal ('../') combined with file access discloses broad/destructive filesystem access: src/agent_bom/cli/_history.py: path traversal ('../') combined with file access discloses broad/destructive filesystem access: src/agent_bom/cli/_runtime.py: open(…, 'w'/'a'/'x') write-mode file access discloses broad/destructive filesystem access: src/agent_bom/cli/agents/__init__.py: open(…, 'w'/'a'/'x') write-mode file access discloses broad/destructive filesystem access: src/agent_bom/cloud/azure.py: path traversal ('../') combined with file access discloses broad/destructive filesystem access: src/agent_bom/discovery/__init__.py: path traversal ('../') combined with file access discloses broad/destructive filesystem access: src/agent_bom/enrichment.py: destructive filesystem call (shutil.rmtree/os.remove/etc.) discloses broad/destructive filesystem access: src/agent_bom/logging_config.py: destructive filesystem call (shutil.rmtree/os.remove/etc.) discloses broad/destructive filesystem access: src/agent_bom/oci_parser.py: path traversal ('../') combined with file access discloses broad/destructive filesystem access: src/agent_bom/proxy_configure.py: destructive filesystem call (shutil.rmtree/os.remove/etc.) discloses broad/destructive filesystem access: src/agent_bom/red_team.py: path traversal ('../') combined with file access discloses broad/destructive filesystem access: src/agent_bom/watch.py: open(…, 'w'/'a'/'x') write-mode 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

  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.msaad00/[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

  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); sends telemetry to third-party SDK (datadog) — disclosure, verify opt-in; manifest/source references analytics behavior — disclosure