Skip to content
Back to search

nexus-skills

MarketplaceWorks with Claude CodeGitHub

Agent skills for the NEXUS second brain: recall verified project memory before work, record outcomes after. Works with Claude Code, Codex, Antigravity, Gemini CLI and any MCP client.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/chhunsour--nexus-skills/badge.svg)](/listings/chhunsour--nexus-skills)

HTML

<a href="/listings/chhunsour--nexus-skills"><img src="/api/listings/chhunsour--nexus-skills/badge.svg" alt="findsafeskills safety rating" /></a>

Install

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

**Any agent, via the skills CLI**

npx skills add Chhunsour/nexus-skills            # pick skills and agents interactively
npx skills add Chhunsour/nexus-skills -g         # user-level, every project

**Claude Code plugin marketplace**

/plugin marketplace add Chhunsour/nexus-skills
/plugin install nexus@nexus-skills

**Manual (git clone)**

git clone https://github.com/Chhunsour/nexus-skills.git
# Then copy or symlink skills/* into your agent's skills folder:
#   Claude Code  ~/.claude/skills/            (project: .claude/skills/)
#   Codex        ~/.codex/skills/             (project: .agents/skills/)
#   Antigravity  ~/.gemini/config/skills/     (project: .agents/skills/)
#   Gemini CLI   ~/.gemini/skills/
ln -s "$PWD"/nexus-skills/skills/* ~/.claude/skills/

Or just ask your agent: *"Install the skills from github.com/Chhunsour/nexus-skills and set up NEXUS."*

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

.claude-plugin/marketplace.json · Manifest · 23 lines
  1. {
  2. "name": "nexus-skills",
  3. "owner": {
  4. "name": "Chhunsour"
  5. },
  6. "metadata": {
  7. "description": "Agent skills for the NEXUS second brain: recall verified project memory before work, record outcomes after, connect any agent.",
  8. "version": "1.0.0"
  9. },
  10. "plugins": [
  11. {
  12. "name": "nexus",
  13. "description": "Use the NEXUS second brain as long-term memory: setup and diagnostics, recall and record, and capturing project knowledge into the review queue.",
  14. "source": "./",
  15. "strict": false,
  16. "skills": [
  17. "./skills/nexus-memory",
  18. "./skills/nexus-setup",
  19. "./skills/nexus-capture"
  20. ]
  21. }
  22. ]
  23. }
README.md · README · 92 lines
  1. # NEXUS Skills
  2. Agent skills that teach AI agents (Claude Code, Codex, Antigravity, Gemini CLI, Cursor and anything else that reads `SKILL.md`) to use your **NEXUS second brain** as long-term project memory. Agents recall verified decisions and past fixes before they work, and record outcomes and lessons after.
  3. | Skill | What the agent learns |
  4. |---|---|
  5. | [`nexus-memory`](skills/nexus-memory/SKILL.md) | The core loop: pick a surface, **recall** context, decisions and solutions, follow trust rules, **record** outcomes and proposals |
  6. | [`nexus-setup`](skills/nexus-setup/SKILL.md) | Connect any agent to NEXUS (MCP registration, agent token), plus a read-only **doctor** that diagnoses and prints fixes |
  7. | [`nexus-capture`](skills/nexus-capture/SKILL.md) | Distill a repo, docs or a session into a small batch of high-quality memory proposals for your review queue |
  8. ## Install
  9. **Any agent, via the skills CLI**
  10. ```bash
  11. npx skills add Chhunsour/nexus-skills # pick skills and agents interactively
  12. npx skills add Chhunsour/nexus-skills -g # user-level, every project
  13. ```
  14. **Claude Code plugin marketplace**
  15. ```
  16. /plugin marketplace add Chhunsour/nexus-skills
  17. /plugin install nexus@nexus-skills
  18. ```
  19. **Manual (git clone)**
  20. ```bash
  21. git clone https://github.com/Chhunsour/nexus-skills.git
  22. # Then copy or symlink skills/* into your agent's skills folder:
  23. # Claude Code ~/.claude/skills/ (project: .claude/skills/)
  24. # Codex ~/.codex/skills/ (project: .agents/skills/)
  25. # Antigravity ~/.gemini/config/skills/ (project: .agents/skills/)
  26. # Gemini CLI ~/.gemini/skills/
  27. ln -s "$PWD"/nexus-skills/skills/* ~/.claude/skills/
  28. ```
  29. Or just ask your agent: *"Install the skills from github.com/Chhunsour/nexus-skills and set up NEXUS."*
  30. ## Connect NEXUS (once)
  31. Ask your agent to *"run the nexus-setup doctor"*, or run it yourself:
  32. ```bash
  33. node skills/nexus-setup/scripts/nexus-doctor.mjs
  34. ```
  35. The short version for **NEXUS Desktop**:
  36. 1. Open NEXUS → **AI Agents** → enable Claude Code and/or Codex (allow proposals if you want agents to suggest memories).
  37. 2. Mark records **verified** and **shared with agents**. Agents see nothing else.
  38. 3. Register the MCP server:
  39. ```bash
  40. claude mcp add --scope user --transport stdio nexus -- '/Applications/NEXUS.app/Contents/MacOS/nexus' --mcp claude
  41. codex mcp add nexus -- '/Applications/NEXUS.app/Contents/MacOS/nexus' --mcp codex
  42. ```
  43. Other MCP clients use `{"command": "/Applications/NEXUS.app/Contents/MacOS/nexus", "args": ["--mcp", "claude"]}`. No MCP at all? The bundled `nexus-mcp.mjs` client works from any shell.
  44. For the **NEXUS OS server** edition, put `NEXUS_API_URL` and the **agent** token in `~/.config/nexus/agent.env`. See [nexus-setup](skills/nexus-setup/SKILL.md).
  45. ## Make agents use it automatically
  46. Paste [`templates/AGENTS.snippet.md`](templates/AGENTS.snippet.md) into your project's `AGENTS.md`, `CLAUDE.md` or `GEMINI.md`.
  47. ## How it works
  48. ```mermaid
  49. flowchart LR
  50. A[Agent] -->|"skill: nexus-memory"| S{Surface}
  51. S -->|"A: MCP tools"| D["NEXUS Desktop MCP connector"]
  52. S -->|"B: nexus-mcp.mjs"| D
  53. S -->|"C: nexus-api.mjs"| O["NEXUS OS API 127.0.0.1:4100"]
  54. D -->|"verified + shared only"| R[(Your library)]
  55. A -->|"propose_memory"| Q[Review queue]
  56. A -->|"outcome / remember"| U["Unverified memories"]
  57. Q -->|"you verify"| R
  58. U -->|"you verify"| R
  59. ```
  60. Safety is built in: agents only read what you've verified and shared, every write is unverified or review-queued, the agent token can't verify, edit or delete anything, and the skills tell agents to treat memory content as data, never instructions.
  61. ## Repository layout
  62. ```
  63. skills/<name>/SKILL.md the skills (Agent Skills format)
  64. skills/*/scripts/ bundled zero-dependency Node clients (Node ≥ 18)
  65. skills/*/references/ detailed tool/API references, loaded on demand
  66. shared/scripts/ canonical client sources → tools/sync-scripts.sh copies them into skills
  67. templates/ AGENTS.md snippet
  68. .claude-plugin/ Claude Code marketplace manifest
  69. ```
  70. Development: edit `shared/scripts/*`, then run `bash tools/sync-scripts.sh && node tools/validate-skills.mjs`. CI checks both.
  71. ## License
  72. MIT

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.