Issues

Sandbox mounts parent repo .git read-only - git worktree commits fail with EROFS (0.2.66, single-repo)

#1366

· andreizimin opened

on Jul 7, 2026

Skill allow-list built from stale base-clone working tree, not the session worktree (repo skills rejected as "not in this session's skills allowlist")

#1336

· Junveloper opened

on Jun 19, 2026

Runaway self-replicating Linear agent sessions: linear_agent_session_create + unguarded parent-resume (no self/cycle/depth guard, no Linear self-author filter, no agent-session webhook idempotency)

#1328

· AbdulAlkurdi opened

on Jun 16, 2026

No way to reduce Linear agent-activity posting volume — every thought/ack is an agentActivityCreate (incl. empty body:""), exhausting the per-token rate limit

#1325

· burnashevrr opened

on Jun 14, 2026

Linear API 429 rate-limits silently break in-flight agent sessions (un-caught activity-post rejections swallowed by global unhandledRejection handler)

#1324

· burnashevrr opened

on Jun 14, 2026

Managed skill plugins can't read their own bundled files — session read allow-list omits the plugin dir (multi-step skills silently degrade)

#1323

· burnashevrr opened

on Jun 14, 2026

Feature: opt-in multi-repo catch-all — route unlabeled issues to all catch-all repositories as one multi-repo session

#1320

· mikeover opened

on Jun 12, 2026

Codex app-server backend rejects all MCP write tools — `onServerRequest` ignores `McpServerElicitation` / `RequestPermissions`

#1317

· HaiyiMei opened

on Jun 12, 2026

Proposal: `<provider>/<model>` compound Linear labels for conflict-free runner routing

#1309

· jappyjan opened

on Jun 10, 2026

Show session usage in final message

#1298

· UnknownPlatypus opened

on Jun 6, 2026

self-auth-linear silently fails to persist workspace tokens when linearWorkspaces is initialized as array

#1261

· jakejjoyner opened

on May 27, 2026

@anthropic-ai/claude-agent-sdk picks musl variant on glibc systems — "Claude Code native binary not found"

#1254

· hunter-bumpers opened

on May 24, 2026

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Sandbox mounts parent repo .git read-only - git worktree commits fail with EROFS (0.2.66, single-repo)

Skill allow-list built from stale base-clone working tree, not the session worktree (repo skills rejected as "not in this session's skills allowlist")

Runaway self-replicating Linear agent sessions: linear_agent_session_create + unguarded parent-resume (no self/cycle/depth guard, no Linear self-author filter, no agent-session webhook idempotency)

No way to reduce Linear agent-activity posting volume — every thought/ack is an agentActivityCreate (incl. empty body:""), exhausting the per-token rate limit

Linear API 429 rate-limits silently break in-flight agent sessions (un-caught activity-post rejections swallowed by global unhandledRejection handler)

Managed skill plugins can't read their own bundled files — session read allow-list omits the plugin dir (multi-step skills silently degrade)

Feature: opt-in multi-repo catch-all — route unlabeled issues to all catch-all repositories as one multi-repo session

Codex app-server backend rejects all MCP write tools — `onServerRequest` ignores `McpServerElicitation` / `RequestPermissions`

Proposal: `<provider>/<model>` compound Linear labels for conflict-free runner routing

Show session usage in final message

self-auth-linear silently fails to persist workspace tokens when linearWorkspaces is initialized as array

@anthropic-ai/claude-agent-sdk picks musl variant on glibc systems — "Claude Code native binary not found"

Uh oh!

Search results