Issues

TP4 (description-behavior mismatch) never sees code embedded in markdown reference docsAffected rule: TP4 "Description-Behavior Mismatch" (MCP Tool Poisoning)

#419

· benedictkwok opened

on Aug 22, 2026

AE3/AE4 analyze binary assets the ledger already marks opaque_content (85-byte PNG -> HIGH, DO_NOT_INSTALL)

#413

· sdkkds opened

on Aug 21, 2026

Bundled hooks and settings are scanned as files but never analyzed as an execution surface

#399

· tarang-tj opened

on Aug 19, 2026

PE3 fires HIGH on credential vocabulary in prose: a SKILL.md containing only the word `keyring` is Privilege Escalation

#396

· Mark2Mac opened

on Aug 18, 2026

Coverage is computed but never affects the exit code: a partially-blind scan is indistinguishable from a complete one in CI

#389

· Mark2Mac opened

on Aug 18, 2026

SKILLSPECTOR_MAX_LLM_CONCURRENCY does not bound total in-flight LLM requests

#387

· Mark2Mac opened

on Aug 17, 2026

Feature request: detect model/provider pins in skill frontmatter

#370

· abek858 opened

on Aug 13, 2026

[SECURITY] SC8 presence check does not fix incomplete skip denylist (CWE-184 / CWE-693) (same class as pycache still applies elsewhere)

#367

· nedlir opened

on Aug 12, 2026

Oversized companion file causes SkillSpector to skip analysis of unrelated skill files (v2.8.2) (another malicious skill example that bypasses this scanner)

#363

· nedlir opened

on Aug 12, 2026

No way to scope a scan, so a skill that ships test fixtures cannot pass

#335

· shi-eric opened

on Aug 2, 2026

LLM analyzers run with no sampling controls, so the published recommendation is not reproducible

#334

· shi-eric opened

on Aug 2, 2026

Space-separated `allowed-tools` is parsed as a single tool name, so LP1 fires on fully-declared skills

#327

· shi-eric opened

on Jul 31, 2026

Navigation Menu

Uh oh!

All issues

TP4 (description-behavior mismatch) never sees code embedded in markdown reference docsAffected rule: TP4 "Description-Behavior Mismatch" (MCP Tool Poisoning)

AE3/AE4 analyze binary assets the ledger already marks opaque_content (85-byte PNG -> HIGH, DO_NOT_INSTALL)

Bundled hooks and settings are scanned as files but never analyzed as an execution surface

PE3 fires HIGH on credential vocabulary in prose: a SKILL.md containing only the word `keyring` is Privilege Escalation

Coverage is computed but never affects the exit code: a partially-blind scan is indistinguishable from a complete one in CI

SKILLSPECTOR_MAX_LLM_CONCURRENCY does not bound total in-flight LLM requests

Feature request: detect model/provider pins in skill frontmatter

[SECURITY] SC8 presence check does not fix incomplete skip denylist (CWE-184 / CWE-693) (same class as pycache still applies elsewhere)

Oversized companion file causes SkillSpector to skip analysis of unrelated skill files (v2.8.2) (another malicious skill example that bypasses this scanner)

No way to scope a scan, so a skill that ships test fixtures cannot pass

LLM analyzers run with no sampling controls, so the published recommendation is not reproducible

Space-separated `allowed-tools` is parsed as a single tool name, so LP1 fires on fully-declared skills

Uh oh!

Search results