ln-011-agent-installer — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited ln-011-agent-installer (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Paths: File paths (references/) are relative to this skill directory.Type: L3 Worker Category: 0XX Shared
Installs or updates the two supported CLI agents: Codex CLI and Claude Code. Single pass per agent: install then immediately verify.
MANDATORY READ: Load references/coordinator_summary_contract.md, references/environment_worker_runtime_contract.md, and references/worker_runtime_contract.md
| Direction | Content |
|---|---|
| Input | OS info, disabled flags per agent, dry_run flag, optional runId, optional summaryArtifactPath |
| Output | Structured summary envelope with payload.status = completed / skipped / error, plus per-agent install outcomes in changes / detail |
If summaryArtifactPath is provided, write the same summary JSON there. If not provided, return the summary inline and remain fully standalone. If runId is not provided, generate a standalone run_id before emitting the summary envelope.
Runtime family: environment-worker-runtime
Phase profile:
PHASE_0_CONFIGPHASE_1_INSTALL_VERIFYPHASE_2_POST_CODEX_SANITY_CHECKPHASE_3_WRITE_SUMMARYPHASE_4_SELF_CHECKRuntime rules:
summary_kind=env-agent-installrun_id and write the default worker-family artifact pathrunId and summaryArtifactPath and must write the summary to the exact provided pathAlways build a structured env-agent-install summary envelope per:
references/coordinator_summary_contract.mdreferences/environment_worker_runtime_contract.mdPayload fields:
targetsagents_processedagents_skippedversionscodex_drift_detectedstatus| Agent | Install Command | Health Check | Interactive |
|---|---|---|---|
| Codex | npm i -g @openai/codex | codex --version | codex or codex "prompt" |
| Claude | npm i -g @anthropic-ai/claude-code or claude update | claude --version | claude or claude "prompt" |
Automation uses:
codex exec ...claude -p ...Note: codex / codex "prompt" are interactive TTY entrypoints. Non-interactive validation and review automation must use codex exec ..., because Codex rejects direct interactive mode when stdin is not a terminal.
For each enabled agent: Install or Update -> Verify -> RecordFor each agent in registry, apply first matching rule:
| Condition | Action | Report |
|---|---|---|
disabled: true | SKIP | disabled by user |
dry_run: true | Show planned command | dry run |
| Codex | npm install -g @openai/codex then codex --version | version or error |
| Claude installed | claude update then claude --version | version or error |
| Claude missing | npm install -g @anthropic-ai/claude-code then claude --version | version or error |
Single pass: install and verify happen atomically per agent. No separate scan phase; the install result is the state.
Error handling:
| Error | Detection | Response |
|---|---|---|
| npm not in PATH | npm --version fails | FAIL gracefully, report npm not found in PATH |
| Permission denied | stderr contains EACCES | FAIL, suggest npm install -g --prefix ~/.local {pkg} |
| Network error | stderr contains ETIMEDOUT or ENETUNREACH | FAIL, report network error |
| Unknown error | Any other non-zero exit | FAIL, include stderr |
Output table:
Agent Installation:
| Agent | Action | Version | Status |
|--------|-----------|----------|--------|
| Codex | installed | 0.124.0 | ok |
| Claude | updated | 2.1.119 | ok |After successful Codex install/update:
~/.codex/config.toml top-level approval_policy and sandbox_mode.ln-013-config-syncer.This installer does not install marketplace plugins, rewrite Codex marketplace layout, or manage Codex execution defaults. It only reports drift so install success is not mistaken for a healthy two-agent environment.
| DON'T | DO |
|---|---|
| Separate check/install/verify phases | Single pass: install then verify |
| Retry failed installs automatically | One attempt, report failure |
Use sudo npm install | Suggest --prefix for permission issues |
Install agents marked disabled | Skip with clear report |
| Configure MCP settings or marketplace plugins here | Route environment alignment to ln-013-config-syncer |
Version: 1.1.0 Last Updated: 2026-03-23
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.