{"data":{"skill":{"slug":"browser-act-browser-act-skill-forge","name":"browser-act-skill-forge","icon":"📦","repo":"https://github.com/browser-act/skills/tree/main/browser-act-skill-forge","status":"approved","author":"browser-act","authorVersion":null,"skillstoreRevision":null},"audit":{"id":"411da366-2bc5-48ad-9ed8-005745124f4e","skill_id":"9d70f22a-d950-4c57-8b35-68c116e6a35a","version":2,"content_hash":"a3c2acbe1c00d411fa89983dc360dbbc","risk_level":"medium","is_blocked":false,"safe_to_publish":true,"analysis_status":"ok","agent_auto_install_policy":null,"manual_install_policy":null,"summary":"Static analysis reported many command, network, and weak-crypto patterns, but review found the weak-crypto hits are Markdown and placeholder false positives. The confirmed risks are intentional: the skill generates Python wrappers, invokes shell command substitution, runs browser-side eval, captures network traffic, and writes local skill artifacts. No prompt injection attempt or confirmed malicious exfiltration was found, so this is elevated-risk automation rather than a publishing block.","remediation":[],"risk_factor_evidence":[{"factor":"external_commands","evidence":[{"file":"SKILL.md","line_end":157,"line_start":153},{"file":"SKILL.md","line_end":181,"line_start":178},{"file":"references/output_template.md","line_end":82,"line_start":75},{"file":"references/output_template.md","line_end":184,"line_start":163},{"file":"references/output_template.md","line_end":237,"line_start":217},{"file":"references/output_template.md","line_end":255,"line_start":253}]},{"factor":"network","evidence":[{"file":"SKILL.md","line_end":255,"line_start":250},{"file":"references/exploration_extraction.md","line_end":139,"line_start":123},{"file":"references/exploration_extraction.md","line_end":181,"line_start":173},{"file":"references/exploration_operation.md","line_end":63,"line_start":57},{"file":"references/output_template.md","line_end":123,"line_start":114}]},{"factor":"filesystem","evidence":[{"file":"SKILL.md","line_end":79,"line_start":78},{"file":"SKILL.md","line_end":149,"line_start":149},{"file":"SKILL.md","line_end":248,"line_start":248},{"file":"references/output_template.md","line_end":22,"line_start":9},{"file":"references/output_template.md","line_end":306,"line_start":301}]}],"critical_findings":[],"high_findings":[],"medium_findings":[{"title":"Dynamic Browser Eval and Generated Script Execution","locations":[{"file":"SKILL.md","line_end":157,"line_start":153},{"file":"references/output_template.md","line_end":82,"line_start":75},{"file":"references/output_template.md","line_end":345,"line_start":315}],"confidence":0.86,"description":"The skill instructs agents to generate Python files that print JavaScript and then execute the output through shell command substitution and browser eval. This is expected for the skill purpose, but unsafe generated code or unsafe parameter quoting could execute unintended browser-side JavaScript in an authenticated session.","confidence_reasoning":"The command-substitution and eval workflow is explicit in the instructions. The risk is elevated but partly mitigated because it is a documented, user-directed automation pattern rather than hidden behavior."},{"title":"Authenticated Browser Network and DOM Automation","locations":[{"file":"SKILL.md","line_end":18,"line_start":16},{"file":"references/exploration_extraction.md","line_end":139,"line_start":123},{"file":"references/exploration_operation.md","line_end":63,"line_start":49}],"confidence":0.83,"description":"The skill directs endpoint discovery, fetch reproduction, DOM extraction, and HAR capture inside the user browser. This can expose or manipulate data available to the logged-in user if applied to sensitive pages or overbroad tasks.","confidence_reasoning":"The browser network and DOM access is central to the skill and is clearly documented. No evidence shows external exfiltration, but authenticated-session automation is inherently sensitive."},{"title":"User-Authorized Operation Submission Automation","locations":[{"file":"references/exploration_operation.md","line_end":63,"line_start":53},{"file":"references/exploration_operation.md","line_end":153,"line_start":136},{"file":"references/output_template.md","line_end":184,"line_start":179}],"confidence":0.79,"description":"The operation workflow captures and packages form submissions and non-GET request behavior. The exploration procedure uses offline HAR capture to reduce side effects, but generated skills may later perform real actions in the user account.","confidence_reasoning":"The instructions explicitly cover form submission and request capture. The safety protocol lowers likelihood of accidental submission during exploration, but runtime side effects remain possible."}],"low_findings":[{"title":"Local Artifact and Memory File Writes","locations":[{"file":"SKILL.md","line_end":79,"line_start":78},{"file":"SKILL.md","line_end":248,"line_start":248},{"file":"references/output_template.md","line_end":306,"line_start":301}],"confidence":0.74,"description":"The skill writes generated skill directories, scripts, temporary artifacts, and optional experience notes. This is expected behavior, but users should choose an output directory carefully and review generated files.","confidence_reasoning":"The file-write behavior is explicitly described and bounded to output, tmp, and memory paths. It is a minor risk because it is visible and user-directed."},{"title":"Weak Cryptography Static Findings Dismissed","locations":[{"file":"SKILL.md","line_end":12,"line_start":3},{"file":"references/output_template.md","line_end":345,"line_start":315}],"confidence":0.9,"description":"The high-severity weak-cryptography detections appear to be false positives from Markdown text, placeholders, filenames, or template content. No cryptographic implementation, digest use, credential hashing, or security decision based on weak algorithms was found in the scanned files.","confidence_reasoning":"Manual review found documentation templates and examples, not cryptographic code. The static labels do not match the semantic content at the cited locations."}],"dangerous_patterns":[{"title":"Shell Command Substitution Around Generated Scripts","locations":[{"file":"SKILL.md","line_end":156,"line_start":155},{"file":"references/output_template.md","line_end":82,"line_start":75},{"file":"references/output_template.md","line_end":184,"line_start":163}],"confidence":0.88,"description":"Command templates use eval with shell command substitution to execute JavaScript emitted by generated Python scripts. Generated scripts and parameter quoting should be reviewed before execution.","confidence_reasoning":"The pattern is directly present in command templates and verification steps. It is dangerous by capability, even though it is part of the advertised workflow."},{"title":"Browser-Side Fetch and Network Capture","locations":[{"file":"references/exploration_extraction.md","line_end":130,"line_start":125},{"file":"references/exploration_extraction.md","line_end":265,"line_start":259},{"file":"references/exploration_operation.md","line_end":63,"line_start":57}],"confidence":0.82,"description":"The skill instructs agents to reproduce site requests with fetch and inspect network requests or HAR files. This can reveal sensitive account-scoped data when used on authenticated sites.","confidence_reasoning":"The fetch and network-capture instructions are explicit. The risk depends on the target site and user authorization, so it is not evidence of malware by itself."}],"files_scanned":4,"total_lines":1219,"audit_model":"codex","audited_at":"2026-06-28T17:44:55.386+00:00","created_at":"2026-06-28T19:08:49.659074+00:00","static_findings":[],"finding_verdicts":[],"semantic_findings":[],"subject_marketplace_commit_sha":null,"subject_content_hash":null,"subject_tree_hash":null,"subject_plugin_path":null,"audit_payload_hash":null,"confirmed_risk_level":null,"scanner_version":null,"policy_version":null,"subject":{"marketplaceCommitSha":null,"contentHash":null,"treeHash":null,"pluginPath":null,"auditPayloadHash":null},"scannerVersion":null,"policyVersion":null},"auditTranslation":null,"localization":{"requestedLocale":"en","contentLocale":"en","availableLocales":["en"],"fallbackToEnglish":false},"attestation":{"availability":"not_attestable","url":null,"status":null,"reason":"canonical install advisories are incomplete"},"trust":{"publicState":"public","auditState":"complete","auditCurrentness":null,"confirmedRiskLevel":"safe","confirmedFindingCount":0,"capabilityReviewCount":4,"needsReviewCount":0,"falsePositiveCount":1,"agentAutoInstallPolicy":null,"manualInstallPolicy":null,"artifactSignatureState":"available","attestationState":"not_attestable","verificationState":"not_verified"},"isLatest":false}}