---

name: env-readiness description: "Use when you want a READ-ONLY readiness review of a Claude Code environment — does this machine have everything wired for the local forge/bob/alf/skills setup? Reviews ~/.claude (global) plus an optional per-project .claude/ across every dimension (skills/agents/workflows installed, SessionStart hooks wired, config files present, gates import, 3-tree identity, tooling tier, knowledge-grounding, per-project hygiene) and prints a single READY / READY-WITH-WARNINGS / NOT-READY verdict with a prioritized repair list. NEVER mutates — repairs are the installer's job (it points at installer/bootstrap-environment.py). Composes the existing probes (env-adoption, identity_check, gates, knowledge-grounding). Trigger on: "is my environment ready", "readiness sweep", "review my claude setup", "check ~/.claude", "env doctor", "is this a new machine set up correctly", "are the hooks/gates/skills wired"."

env-readiness — read-only environment readiness doctor

The review counterpart to installer/bootstrap-environment.py (which installs and wires). The doctor answers one question — "is this environment ready for our setup?" — and never changes anything. It composes the existing probes into one verdict and points every fix at the installer or a specific probe.

When to use it

• A fresh machine / new clone — confirm everything is wired before relying on it.
• After bootstrap-environment.py — verify the wire-up actually took.
• Anytime a session "feels off" (no hooks firing, a gate missing, drift suspected).
• In CI — --strict --json to fail a pipeline when the env is NOT-READY.

Run it

python3 ~/.claude/skills/env-readiness/scripts/readiness.py            # review ~/.claude (+ CWD project if it has .claude/)
python3 .../readiness.py --project /path/to/repo                       # also review a specific project
python3 .../readiness.py --json                                        # machine-readable
python3 .../readiness.py --strict                                      # exit 1 when NOT-READY (CI gate)
python3 .../readiness.py --claude-home /custom/.claude                 # non-default home

What it checks (each → PASS / WARN / FAIL + a repair pointer)

Dimension	Checks	Composes
Install	skills/agents/workflows counts + key skills (forge, env-adoption, …) and agents (bob, alf, pa, evo, wiki); CLAUDE.md; AGENTS.md symlink; claude-observe bin; Codex mirror parity	install.py, bootstrap
Hooks	the 5 canonical SessionStart hooks wired in settings.json	settings.json
Config	policy-limits.json (mode 0600), model-policy.yaml, publish-config.json present + parse	file probes
Gates	_meta/gates.py imports + loads in a child process	gates.py
Identity	3-tree (prod/shadow/agent-foundry) parity (advisory)	identity_check.py
Tooling	env-adoption inventory + tools on PATH; knowledge-grounding mode	probe.sh, discover.sh
Project	.claude/settings.local.json parses, project CLAUDE.md, stale .forge/.bob/.ledger artifacts	per-project walk

Verdict: READY (all PASS) / READY-WITH-WARNINGS (≥1 WARN, no FAIL) / NOT-READY (≥1 FAIL).

Design guarantees

• Read-only. No file is written, no setting changed. Every repair is a pointer (usually python3 installer/bootstrap-environment.py), never an action.
• Crash-proof. Each check is isolated — a broken probe becomes a FAIL row, never a traceback that aborts the sweep.
• Stdlib-only. No third-party dependency; runs on any python3.

Relationship to other tooling

• installer/bootstrap-environment.py — the fix side (idempotent install + wire). The doctor reviews; bootstrap repairs. Run bootstrap, then the doctor to confirm.
• env-adoption — the tool-tier probe the doctor reads (inventory.json).
• knowledge-grounding — the source/internet probe the doctor reads (sources.json).
• _meta/identity_check.py — the 3-tree byte-parity check the doctor invokes.
• _meta/gates.py — the gate registry the doctor import-smokes.