Issues

• Shell threat detection bypasses and fail-open config (follow-up to #649)

  good first issueSecurity
  Cyrax321/SNAGLINE#650 · Filter by author Cyrax321 opened on Oct 22, 2026

• feat: Add adversarial security tests and a documented threat model

  documentationSecurity
  Cyrax321/SNAGLINE#648 · Filter by author Cyrax321 opened on Oct 22, 2026

• feat: Add a security evaluation API and CLI commands

  clienhancementSecurity
  Cyrax321/SNAGLINE#647 · Filter by author Cyrax321 opened on Oct 22, 2026

• feat: Add an MCP security gateway for tool request evaluation

  enhancementIntegrationSecurity
  Cyrax321/SNAGLINE#646 · Filter by author Cyrax321 opened on Oct 22, 2026

• feat: Add prompt injection indicators for untrusted content

  enhancementSecurity
  Cyrax321/SNAGLINE#645 · Filter by author Cyrax321 opened on Oct 22, 2026

• feat: Add a tamper-evident security authorization audit log

  enhancementSecurity
  Cyrax321/SNAGLINE#644 · Filter by author Cyrax321 opened on Oct 22, 2026

• feat: Add a dedicated security risk scoring engine

  enhancementSecurity
  Cyrax321/SNAGLINE#643 · Filter by author anya-research opened on Oct 22, 2026

• feat: Add a security policy schema and validation layer

  enhancementSecurity
  Cyrax321/SNAGLINE#642 · Filter by author anya-research opened on Oct 22, 2026

• feat: Add data loss prevention checks for outbound tool payloads

  enhancementSecurity
  Cyrax321/SNAGLINE#641 · Filter by author anya-research opened on Oct 22, 2026

• feat: Add secret detection and redaction utilities

  enhancementSecurity
  Cyrax321/SNAGLINE#640 · Filter by author Cyrax321 opened on Oct 22, 2026

• feat: Add filesystem path security checks for tool requests

  enhancementSecurity
  Cyrax321/SNAGLINE#639 · Filter by author Cyrax321 opened on Oct 22, 2026

• feat: Add a fail-closed security authorization profile

  enhancementgood first issueSecurity
  Cyrax321/SNAGLINE#637 · Filter by author anya-research opened on Oct 22, 2026
  1

• ci: no workflow runs tests/real_world_test.sh, and it fails on master -- the only end-to-end wiring check in the repo

  cigood first issue
  Cyrax321/SNAGLINE#621 · Filter by author Cyrax321 opened on Oct 13, 2026

• cli: snagline hook reports malformed payloads with a raw StepEvent.__init__ signature, and silently drops a wrong-typed hook_event_name

  good first issue
  Cyrax321/SNAGLINE#620 · Filter by author Cyrax321 opened on Oct 13, 2026

• build: uv.lock is committed and referenced by nothing -- it resolves only the all extra, pins neither CI dev tool, and was last regenerated 2026-08-26

  good first issue
  Cyrax321/SNAGLINE#619 · Filter by author Cyrax321 opened on Oct 13, 2026

• docs: INTEGRATION_MATRIX lists adapters.claude_code in its table then omits it from the import block, and its public surface appears in no import example

  documentationgood first issue
  Cyrax321/SNAGLINE#618 · Filter by author Cyrax321 opened on Oct 13, 2026

• docs: add a CONTRIBUTING.md -- there is none, and the issue template demands a standalone repro from documentation bug reports

  documentationgood first issue
  Cyrax321/SNAGLINE#617 · Filter by author Cyrax321 opened on Oct 13, 2026

• docs: README documents 2 of Monitor.metrics()'s 8 keys and none of GET /metrics's 11 Prometheus series

  documentationgood first issue
  Cyrax321/SNAGLINE#616 · Filter by author Cyrax321 opened on Oct 13, 2026

• tests: nothing regenerates benchmarks/fixtures, so the committed corpus can silently drift from the generator that claims to produce it

  benchmarkgood first issuetests
  Cyrax321/SNAGLINE#615 · Filter by author Cyrax321 opened on Oct 13, 2026

• examples: the LangGraph error-mode demo cannot fire error_cascade -- LangGraph re-raises the first tool error, so no error event reaches the Monitor

  documentationgood first issue
  Cyrax321/SNAGLINE#614 · Filter by author Cyrax321 opened on Oct 13, 2026

• tests: the replay check greps for "2 risk" but the loop fixture has emitted exactly one since the #94 re-arm fix

  good first issuetests
  Cyrax321/SNAGLINE#613 · Filter by author Cyrax321 opened on Oct 13, 2026

• tests: the smoke script's sidecar section asserts nothing -- every POST omits Content-Type and gets a 415, and two checks are unconditional PASS

  good first issuetests
  Cyrax321/SNAGLINE#612 · Filter by author Cyrax321 opened on Oct 13, 2026

• benchmark: near_duplicate_loop is INTENTIONALLY_UNGATED with no labeled fixtures, so a shipped trigger has no measured accuracy

  Cyrax321/SNAGLINE#609 · Filter by author abyyxhek opened on Oct 10, 2026

• benchmark: 14 of 17 vary_* builders never read their variant number, so variation is seed-driven and token_runaway has 2 lengths

  Cyrax321/SNAGLINE#608 · Filter by author abyyxhek opened on Oct 10, 2026

• benchmark: stagnation fires at a median of step 56 on 101-121 step episodes, and wall_clock_budget at 80 -- half the run is spent before the alert

  Cyrax321/SNAGLINE#607 · Filter by author abyyxhek opened on Oct 10, 2026