# Reported issues for nexus-mcp Pod holds 8 of 8 problems reported by people outside the maintainer team. Issues filed by the project's own owners, members and collaborators are excluded entirely — a maintainer's release checklist is not a warning to a prospective user. Back to [nexus-mcp](/mcp/nexus-mcp). ## Most discussed ### ci: bump shared-workflows refs to v3.0.2 and add fork PR safety gate ## What Bump every `j7an/shared-workflows` reusable workflow reference in this repo to `v3.0.2`, pinned to the dereferenced commit SHA: `29bcd576f856e85042ab699dce66523711a44b48 # v3.0.2` Also add a status-only `pull_request_target` companion workflow for non-Dependabot PRs, using the existing `dep-rank` workflow split as the reference pattern. ## Why `shared-workflows` v3.0.2 includes the fork-PR read-only-token handling from j7an/shared-workflows#80. That makes the reusable dependency-saf [Read the thread](https://github.com/j7an/nexus-mcp/issues/213) · 2026-06-06 · closed · 1 comment ### Track: deps: bump the all-actions group with 3 updates ## Dependency Update Tracking **PR:** #150 **Package:** `all-actions-d4d48f0125` **Ecosystem:** GitHub Actions **Eligible for merge:** 2026-04-10 (after 5 business days) ### Security Links Monitor these sources for advisories before the cooldown expires: | Source | Link | |--------|------| | GitHub Advisory (GHSA) | [Search advisories](https://github.com/advisories?query=ecosystem%3Aactions+all-actions-d4d48f0125) | | OSV.dev | [Search vulnerabilities](https://osv.dev/list?ecosystem=GitHub+A [Read the thread](https://github.com/j7an/nexus-mcp/issues/151) · 2026-04-04 · closed · outside contributor · 1 comment ### Remove Gemini CLI support from committed repo files ## Background Google announced that Gemini CLI is transitioning to Antigravity CLI. On **June 18, 2026**, Gemini CLI stops serving requests for Google AI Pro, Ultra, and free Gemini Code Assist for individuals users. Enterprise and paid API-key access may continue, but consumer Gemini CLI support is no longer a stable baseline for Nexus MCP. Sources: - https://developers.googleblog.com/an-important-update-transitioning-gemini-cli-to-antigravity-cli/ - https://github.com/google-gemini/gemini-cl [Read the thread](https://github.com/j7an/nexus-mcp/issues/201) · 2026-05-29 · closed · 0 comments ### Configure uv exclude-newer for one week ## Context uv supports delaying dependency resolution away from freshly uploaded package artifacts with `exclude-newer`. This helps avoid pulling newly released package versions immediately after publication. ## Scope Add a one-week cutoff for uv-managed dependency resolution. ## Tasks - Add `[tool.uv]` to `pyproject.toml` if it does not already exist. - Set `exclude-newer = "1 week"`. - Run `uv lock` so the lockfile is resolved under the cutoff. - Verify with `uv sync --locked --all-extras -- [Read the thread](https://github.com/j7an/nexus-mcp/issues/197) · 2026-05-25 · closed · 0 comments ### ci: add OSV-Scanner and dependency-review-action to security pipeline ## What Add standard supply-chain checks alongside the existing `dependency-safety` workflow: 1. **OSV-Scanner** in `security.yml` - `osv-scan`: full recursive scan on `push` to `main` and weekly `schedule` - `osv-scan-pr`: PR/merge-queue scan with annotations for newly introduced vulnerabilities 2. **Dependency Review Action** in `ci.yml` - Runs on pull requests - Fails on moderate-or-higher vulnerability severity - Blocks incompatible copyleft licenses: GPL-2.0, GPL-3.0, AGPL [Read the thread](https://github.com/j7an/nexus-mcp/issues/176) · 2026-04-22 · closed · 0 comments ### bug: `_try_extract_error` picks terminal session summary instead of real API error # bug: `_try_extract_error` picks terminal session summary instead of real API error **Labels:** `bug` `error-handling` `dx` **Affects:** `src/nexus_mcp/runners/gemini.py` --- ## Description ### What When the Gemini CLI exits with a real API error (e.g., HTTP 429 `MODEL_CAPACITY_EXHAUSTED`), `_try_extract_error` surfaces a misleading `"Gemini API error 1: [object Object] ()"` instead of the actual reason. The caller — whether a human reading an MCP error or code checking the exception mess [Read the thread](https://github.com/j7an/nexus-mcp/issues/14) · 2026-02-18 · closed · external user · 0 comments ### feat: automatic model fallback on retryable Gemini API errors # feat: automatic model fallback on retryable Gemini API errors **Labels:** `feat` `error-handling` `resilience` **Depends on:** `#003` (structured CLI error extraction must land first) **Affects:** `src/nexus_mcp/exceptions.py`, `src/nexus_mcp/config.py`, `src/nexus_mcp/runners/base.py`, `src/nexus_mcp/runners/gemini.py` --- ## Description ### What When the Gemini CLI returns a retryable error — such as `MODEL_CAPACITY_EXHAUSTED` (HTTP 429) or `UNAVAILABLE` (HTTP 503) — the system currentl [Read the thread](https://github.com/j7an/nexus-mcp/issues/11) · 2026-02-17 · closed · external user · 0 comments ### bug: Gemini CLI error details silently discarded — MCP client sees only "exit code 1" # bug: Gemini CLI error details silently discarded — MCP client sees only "exit code 1" **Labels:** `bug` `error-handling` `dx` **Affects:** `src/nexus_mcp/runners/gemini.py`, `src/nexus_mcp/runners/base.py`, `src/nexus_mcp/server.py`, `src/nexus_mcp/exceptions.py` --- ## Description ### What When the Gemini CLI exits with a non-zero code (HTTP 429 rate limit, 401 unauthorized, quota exhausted, model capacity, etc.), the MCP client receives only: ``` CLI command failed with exit code 1 ``` [Read the thread](https://github.com/j7an/nexus-mcp/issues/10) · 2026-02-17 · closed · external user · 0 comments ## Most recent The remaining reports are on [the project's issue tracker](https://github.com/j7an/nexus-mcp/issues).