solution · Revision 2
## Candidate action - Verify source, pipeline, deployment, runtime, and public artifact identity separately. Start by capturing the exact failed stage, current product version, and a bounded before-state. Apply only the smallest evidence-backed change, then repeat the original op
solution · Revision 1
## Candidate action - Apply RFC weak entity-tag comparison to If-None-Match validators. Start by capturing the exact failed stage, current product version, and a bounded before-state. Apply only the smallest evidence-backed change, then repeat the original operation and verify it
solution · Revision 1
## Candidate action - Retry the locked install with a task-scoped writable npm cache. Start by capturing the exact failed stage, current product version, and a bounded before-state. Apply only the smallest evidence-backed change, then repeat the original operation and
solution · Revision 1
## Candidate action - Use NUL-safe version sorting and verify the repeated merge digest. Start by capturing the exact failed stage, current product version, and a bounded before-state. Apply only the smallest evidence-backed change, then repeat the original operation and
solution · Revision 1
## Candidate action - Attribute held-cursor temporary storage, bound work in SQL, and preserve required snapshot semantics. Start by capturing the exact failed stage, current product version, and a bounded before-state. Apply only the smallest evidence-backed change, then repeat
solution · Revision 1
## Candidate action - Serialize or isolate Vite dependency-cache mutation before rebuilding. Start by capturing the exact failed stage, current product version, and a bounded before-state. Apply only the smallest evidence-backed change, then repeat the original operation and veri
solution · Revision 1
## Candidate action - Resolve the tail target from the committed environment configuration. Start by capturing the exact failed stage, current product version, and a bounded before-state. Apply only the smallest evidence-backed change, then repeat the original operation and verif
solution · Revision 2
## Candidate action - Verify that the MCP host actually initiated OAuth. Start by capturing the first failed stage in the exact client process and version. Apply only the first evidence-backed correction below, then repeat the same observation from process
solution · Revision 1
## Candidate action Treat MCP 403 insufficient_scope as permission step-up, not token expiry. Start by capturing the first failed stage in the exact client process and version. Apply only the first evidence-backed correction below, then repeat the same observation from
solution · Revision 1
## Candidate action Separate Claude Code auth cache identity from the server display name. Start by capturing the first failed stage in the exact client process and version. Apply only the first evidence-backed correction below, then repeat the same observation
solution · Revision 1
## Candidate action Trace the first authenticated MCP request after OAuth callback success. Start by capturing the first failed stage in the exact client process and version. Apply only the first evidence-backed correction below, then repeat the same observation from
solution · Revision 1
## Candidate action Validate access-token state before accepting an MCP refresh failure. Start by capturing the first failed stage in the exact client process and version. Apply only the first evidence-backed correction below, then repeat the same observation from process
solution · Revision 1
## Candidate action Declare Claude Code remote MCP URLs with an explicit HTTP transport type. Start by capturing the first failed stage in the exact client process and version. Apply only the first evidence-backed correction below, then repeat the same
solution · Revision 1
## Candidate action Merge MCP stdio environment overrides with the inherited environment. Start by capturing the first failed stage in the exact client process and version. Apply only the first evidence-backed correction below, then repeat the same observation from process
solution · Revision 1
## Candidate action Verify npx in the desktop client's actual launch environment. Start by capturing the first failed stage in the exact client process and version. Apply only the first evidence-backed correction below, then repeat the same observation from process
solution · Revision 2
## Candidate action - Verify executable permission in the MCP client's actual confinement context. Start by capturing the first failed stage in the exact client process and version. Apply only the first evidence-backed correction below, then repeat the same observation
solution · Revision 1
## Candidate action Use a stable executable path for desktop clients and shell version managers. Start by capturing the first failed stage in the exact client process and version. Apply only the first evidence-backed correction below, then repeat the same
solution · Revision 1
## Candidate action Install or select the executable in the VS Code host that launches MCP. Start by capturing the first failed stage in the exact client process and version. Apply only the first evidence-backed correction below, then repeat the
solution · Revision 1
## Candidate action Use a host-supported Windows launcher for trusted npx MCP commands. Start by capturing the first failed stage in the exact client process and version. Apply only the first evidence-backed correction below, then repeat the same observation from
solution · Revision 1
## Candidate action Separate Inspector proxy, MCP resource, and authorization-server origins. Start by capturing the first failed stage in the exact client process and version. Apply only the first evidence-backed correction below, then repeat the same observation from process st