sentinel-env

name	sentinel-env
description	Environment-aware pre-flight check. Use when starting work in a new environment, switching machines, or when 'works on my machine' bugs surface. Validates OS, runtime versions, installed tools, port availability, env vars, and disk space BEFORE coding starts. Like sentinel but for the environment, not the code.
metadata	{"author":"runedev","version":"0.5.0","layer":"L3","model":"haiku","group":"validation","tools":"Bash, Read, Write, Glob, Grep"}

sentinel-env

Purpose

Catch environment mismatches before they waste debugging time. Validates that the developer's machine has the right runtime versions, tools, ports, and configuration to run the project. Prevents the entire class of "works on my machine" failures that masquerade as code bugs.

This is the environment counterpart to sentinel (which checks code security) and preflight (which checks code quality). sentinel-env checks the MACHINE, not the code.

Triggers

• Called by cook Phase 0.5 — before planning, after resume check (first run in a new project only)
• Called by scaffold — after project bootstrap, verify environment matches generated config
• Called by onboard — during project onboarding, verify developer can run the project
• /rune env-check — manual environment validation
• Auto-trigger: when npm install, pip install, or similar fails during cook

Calls (outbound)

None — sentinel-env never calls another Rune skill and never modifies the developer's machine/environment (no install/update). In --agents mode it writes ONE Rune-owned advisory cache file, .rune/runtimes.json (a Rune state file, not the dev's environment — same category as the files journal/session-bridge/council write).

Called By (inbound)

• cook (L1): Phase 0.5 — first run detection (no .rune/ directory exists)
• scaffold (L1): post-bootstrap environment validation
• onboard (L2): developer onboarding verification
• User: /rune env-check direct invocation
• User: /rune env-check --agents — also run AI-Agent Fleet Detection (Step 6.6)

Data Flow

Feeds Into →

• council (L3): .rune/runtimes.json — when --agents mode ran during env pre-flight, it PRE-WARMS council's Step 1 DETECT cache (council reuses a same-session cache instead of re-probing the bridge). Schema + model-family map are OWNED by skills/council/references/dispatch-protocol.md §Detect — sentinel-env writes that exact shape, it does not define its own.

Execution

Step 1: Detect Project Type

Read project configuration files to determine what environment is needed:

1. Use Glob to check for project config files:

   • package.json → Node.js project
   • pyproject.toml / setup.py / requirements.txt → Python project
   • Cargo.toml → Rust project
   • go.mod → Go project
   • Gemfile → Ruby project
   • docker-compose.yml / Dockerfile → Docker project
   • .nvmrc / .node-version → specific Node version required
   • .python-version → specific Python version required

2. Read each detected config file to extract version constraints:

   • package.json → engines.node, engines.npm, dependency versions
   • pyproject.toml → requires-python, dependency versions
   • Cargo.toml → rust-version
   • go.mod → go directive version

3. Build an environment requirements checklist from the detected configs.

Step 2: Runtime Version Check

For each detected runtime, verify the installed version matches constraints:

# Node.js
node --version    # Compare against package.json engines.node or .nvmrc
npm --version     # Compare against package.json engines.npm
# or pnpm/yarn/bun depending on lockfile present

# Python
python --version  # Compare against pyproject.toml requires-python
pip --version

# Rust
rustc --version   # Compare against Cargo.toml rust-version
cargo --version

# Go
go version        # Compare against go.mod go directive

# Docker
docker --version
docker compose version

Version comparison logic:

• If the constraint is >=18.0.0 and installed is 20.11.1 → PASS
• If the constraint is >=18.0.0 and installed is 16.20.2 → BLOCK (wrong major version)
• If the runtime is not installed at all → BLOCK
• If no version constraint exists in config → WARN (version unconstrained)

Step 3: Required Tools Check

Detect and verify tools the project depends on:

1. Package manager: Check which lockfile exists and verify the matching tool is installed

   • package-lock.json → npm
   • pnpm-lock.yaml → pnpm
   • yarn.lock → yarn
   • bun.lockb → bun
   • poetry.lock → poetry
   • uv.lock → uv
   • Mismatched lockfile + installed tool → WARN (e.g., yarn.lock exists but only npm installed)

2. Git: git --version — required for all projects

3. Docker: Check only if Dockerfile or docker-compose.yml exists

4. Database tools: Check if prisma, drizzle, alembic, django migrations exist → verify DB client installed

5. Build tools: Check for turbo.json (turborepo), nx.json (Nx), Makefile, etc.

6. Hard dependencies — tools the project WRAPS (not just uses as dev dependency): Scan for evidence that the project wraps an external tool:

   • Grep for shutil.which(, which , command -v → project looks up an executable at runtime
   • Grep for subprocess.run(, child_process.exec(, Deno.Command( → project invokes external CLI
   • Read README/docs for "requires X installed" or "depends on X"

   For each detected hard dependency, apply the 9-tier binary detection below — checking only which/where is insufficient and produces the largest category of "works on my machine" false-negatives (user has binary installed but PATH is stale, or installed via a package manager that didn't register it, or installed as a desktop app with a bundled binary).

   9-Tier Binary Detection (stop at first hit):

   Tier	Source	Catches
   1	Explicit --<tool>-bin <path> flag	CI, automation, manual override
   2	Skill-specific env var <SKILL>_<TOOL>_BIN	Per-project pinning
   3	Tool-family env var <TOOL>_APP_BIN	Ecosystem conventions
   4	Generic tool env var <TOOL>_BIN	Legacy overrides
   5	Platform desktop-app bundle (macOS .app/Contents/Resources, Windows %LOCALAPPDATA%\Programs, Linux /opt)	Desktop app users (~40% of population)
   6	PATH lookup (which/where.exe)	Standard shell users
   7	Package manager global bin (npm config get prefix, pnpm, pipx --list, cargo install --root)	npm-global on Windows (PATH oversight)
   8	Platform common directories — Unix: ~/.local/bin, ~/.npm-global/bin, ~/.bun/bin, ~/.cargo/bin, ~/.deno/bin, ~/.volta/bin, ~/.asdf/shims, ~/.proto/bin, /opt/homebrew/bin, /usr/local/bin. Windows: %APPDATA%\npm, %USERPROFILE%\.bun\bin, %USERPROFILE%\.cargo\bin, %USERPROFILE%\.deno\bin, %LOCALAPPDATA%\Microsoft\WindowsApps, %ProgramFiles%\nodejs	Bun / Cargo / Deno / Volta / asdf / proto users + manual installers
   9	Platform release archive names (e.g., codex-x86_64-unknown-linux-musl, <tool>-aarch64-apple-darwin)	Release-tarball downloaders

   Verdict:

   • Tool found via any tier → PASS (log which tier + version)
   • Tool resolves to DIFFERENT paths across tiers (e.g. a ~/.local/bin copy AND a PATH copy, or two PATH entries) → WARN [ENV-AMBIG]: '<tool>' resolves to N paths — the shell picks <selected>, but <other> also exists. This is the "wrong binary runs" class of "works on my machine" (stale/shadowed installs, two Pythons, two npms). Don't stop at the first tier hit for the ambiguity check — after a PASS, cheaply confirm PATH (where/which -a) doesn't disagree with the winning tier. List all paths + versions so the developer can prune the shadow.
   • Tool NOT found → BLOCK with per-OS install guidance:

     [ENV-XXX] Required tool '<tool>' not found (9-tier lookup exhausted)
       → Debian/Ubuntu: sudo apt install <tool>
       → macOS: brew install <tool> (or desktop app: <URL>)
       → Windows: winget install <tool> (or choco install <tool>)
       → Any platform: npm install -g <package> (if Node tool)
       → Manual: <download URL>
       → Pin explicitly: export <TOOL>_BIN=/path/to/binary
     

   • This prevents the entire class of "it worked in CI but not locally" failures where subprocess.run() silently fails
   • Reference implementation: scripts/codex_imagen_bridge.mjs in @rune-pro/media ports this pattern

Step 4: Port Availability Check

Detect which ports the project needs and check if they're available:

1. Parse port information from:

   • package.json scripts (look for --port, -p, PORT= patterns)
   • .env / .env.example (look for PORT=, DATABASE_URL with port)
   • docker-compose.yml (ports section)
   • Common defaults: 3000 (Next.js/React), 5173 (Vite), 8000 (Django/FastAPI), 5432 (PostgreSQL), 6379 (Redis)

2. Check each port:

   # Cross-platform port check
   # Windows: netstat -ano | findstr :PORT
   # Unix: lsof -i :PORT or ss -tlnp | grep :PORT
   

3. If port is in use → WARN with the process name using it

Step 5: Environment Variables Check

Compare required env vars against actual configuration:

1. Read .env.example or .env.template if it exists
2. Read .env if it exists (DO NOT log values — only check key presence)
3. For each key in .env.example:
   • Present in .env → PASS
   • Missing from .env → WARN (with the key name, never the expected value)
4. Check for dangerous patterns:
   • .env committed to git (check .gitignore) → BLOCK (security risk)
   • Placeholder values still present (your-api-key-here, changeme, xxx) → WARN

Step 6: Disk Space and System Resources

Quick system health check:

1. Disk space: Check available space on the project drive

   • < 1 GB → WARN
   • < 500 MB → BLOCK (npm install / docker build will fail)

2. Platform-specific checks:

   • Windows: Check for long path support (git config core.longpaths for node_modules)
   • macOS: Check Xcode CLI tools if native modules detected (node-gyp in dependencies)
   • Linux: Check file watcher limit if large project (fs.inotify.max_user_watches)

Step 6.6: AI-Agent Fleet Detection (opt-in — --agents flag, or when called by council)

Skip this step unless --agents was passed or council requested it. It answers ONE question: "how many distinct AI model families can this machine reach for a decorrelated council run?" — and pre-warms council's cache so its first invocation skips a redundant probe.

Schema and model-family map are OWNED by skills/council/references/dispatch-protocol.md §Detect. Do NOT redefine them here — read that file and write its exact shape. This keeps a single source of truth; if that reference changes, this step follows it.

1. Resolve the bridge: look for 1devtool-agent on PATH, then the per-user location for the current OS (~/.1devtool/bin/1devtool-agent POSIX, %USERPROFILE%\.1devtool\bin\1devtool-agent.cmd Windows). Resolve relative to the current user's home — never hardcode a specific user's path.
2. Bridge found → 1devtool-agent list --json; for each status: detected entry map its id → model_family per the council reference table (claude→anthropic, codex→openai, gemini/agy→google, grok→xai, qwen→alibaba, wrapper CLIs cline/amp/opencode/aider→unknown). Reading ~/.1devtool/state/cli-registry.json directly is an acceptable faster path (it is the pre-scanned manifest list returns).
3. Bridge NOT found → detected: [], bridge_path: null. This is normal, not an error. Optionally 9-tier detect raw AI-agent binaries for the human report only ("5 AI CLIs installed but no 1devtool bridge → council will run subagent-only"), but do NOT put unbridged binaries in detected[] — council can't dispatch to them.
4. Write .rune/runtimes.json in the council-owned schema (checked_at session marker, bridge_path, detected[]). Only bridge-reachable status: detected entries belong in detected[].
5. Count distinct CONFIRMED families (exclude unknown wrappers) → report MULTI_FAMILY-capable (≥2) or NO_DECORRELATION-only (<2). Honesty (inherit council's caveat): "confirmed" = the CLI is vendor-dedicated by product design, NOT that a given response was verified to come from that vendor's backend (BYOK / base-URL / gateway override can collapse families invisibly). Report the count as advisory availability, never as a decorrelation guarantee — that call is council's at run time.

Step 7: Report

Produce a structured environment report:

Verdict logic:

• Any BLOCK finding → BLOCKED (environment cannot run this project)
• Any WARN finding → READY WITH WARNINGS (can run but may hit issues)
• All checks pass → READY (environment is correctly configured)

For each finding, include a specific remediation command the developer can copy-paste.

Output Format

## Environment Check: [project name]
- **Project type**: [Node.js / Python / Rust / Go / Multi]
- **Checks run**: [count]
- **Verdict**: READY | READY WITH WARNINGS | BLOCKED

### BLOCKED
- [ENV-001] Node.js 16.20.2 installed but >=18.0.0 required
  → Fix: `nvm install 18 && nvm use 18`

### WARNINGS
- [ENV-002] Port 3000 in use by process "node" (PID 12345)
  → Fix: `kill 12345` or change PORT in .env
- [ENV-003] Missing env var: DATABASE_URL (required by .env.example)
  → Fix: Copy from .env.example and fill in your database connection string

### PASSED
- [ENV-004] pnpm 9.1.0 ✓ (matches pnpm-lock.yaml)
- [ENV-005] Git 2.44.0 ✓
- [ENV-006] Docker 25.0.3 ✓
- [ENV-007] Disk space: 42 GB available ✓

### AI AGENTS (only when --agents / council mode ran)
- Bridge: 1devtool-agent ✓ (or: not found → council runs subagent-only)
- Reachable: codex (openai), grok (xai), gemini (google), agy (google) — 3 distinct confirmed families
- council capability: MULTI_FAMILY-capable (advisory — not a decorrelation guarantee; backend override can collapse families)
- Wrote .rune/runtimes.json (pre-warms council DETECT)

Constraints

1. MUST NOT install, update, or modify anything on the developer's MACHINE/environment (no apt install, no version switches, no PATH edits). Writing the Rune-owned advisory cache .rune/runtimes.json in --agents mode is the ONE exception and is not a machine modification — it is Rune state, same as journal/council run files.
2. MUST NOT log environment variable VALUES — only check key presence (security)
3. MUST provide copy-paste remediation commands for every BLOCK and WARN finding
4. MUST handle cross-platform differences (Windows/macOS/Linux) gracefully
5. MUST complete in under 10 seconds — use parallel Bash calls where possible
6. MUST NOT block on WARN findings — only BLOCK findings prevent proceeding

Sharp Edges

Failure Mode	Severity	Mitigation
False BLOCK on version — semver parsing error	HIGH	Use simple major.minor comparison, not full semver regex
Slowness on Windows — netstat/port checks are slower	MEDIUM	Timeout port checks at 3s, skip if slow
.env file contains secrets — accidentally logged	CRITICAL	NEVER read .env values, only check key existence via grep for key names
Platform detection wrong — WSL vs native Windows	MEDIUM	Check for WSL explicitly (uname -r contains "microsoft")
Over-checking — flagging optional tools as required	MEDIUM	Only check tools evidenced by config files, not speculative
Missing hard dependency — project wraps external CLI but tool not checked	HIGH	Step 3.6: scan for shutil.which, subprocess.run, child_process.exec → verify tool exists on PATH
Hard dep found but wrong version — tool exists but API changed	MEDIUM	Log version for manual review. Version compatibility is project-specific — don't guess
--agents writes .rune/runtimes.json with a schema that drifts from council's	HIGH	Schema + family map are OWNED by skills/council/references/dispatch-protocol.md §Detect — sentinel-env writes that shape, never its own; if the reference changes, this step follows it
Reporting an installed-but-unbridged AI CLI as council-reachable	HIGH	Step 6.6.3: only bridge-reachable status: detected entries go in detected[]; unbridged binaries are report-only text, never dispatchable state
Claiming a distinct-family count as a decorrelation guarantee	MEDIUM	Step 6.6.5: report as advisory availability only; "confirmed" ≠ backend-verified (BYOK/gateway override) — the real decorrelation call is council's at run time

Done When

• All detected project runtimes version-checked against constraints
• Package manager matches lockfile type
• Required ports checked for availability
• Environment variables compared against .env.example (keys only)
• Disk space verified adequate
• Structured report with READY / READY WITH WARNINGS / BLOCKED verdict
• Every BLOCK/WARN finding has a copy-paste remediation command
• In --agents mode only: .rune/runtimes.json written in council's canonical schema and a distinct-family count reported (advisory, not a decorrelation guarantee)

Cost Profile

~500-1000 tokens input, ~500-1000 tokens output. Haiku model — this is fast, cheap, read-only scanning. Runs once per new project (or on manual invoke). Sub-10-second execution target.