harness: read Syncro keys from vault, stop hardcoding the repo root

Two defects found while running /wiki-compile, both from tooling assuming a path instead of resolving it. 1. Plaintext Syncro API keys. Mike's and Howard's live PSA keys were copy-pasted into three command files, a script, and two catalog docs -- despite both already being vaulted at msp-tools/syncro and msp-tools/syncro-howard. Replaced with reads from the SOPS vault via a new sourced helper, .claude/scripts/syncro-env.sh. Write paths fail closed; read-only paths degrade to skipped enrichment rather than a wrong key. Per-user mapping is unchanged, so Syncro attribution is too. 2. Hardcoded repo root. wiki-compile/wiki-lint/inject-standards and gen_b64.py hardcoded D:/claudetools; this machine is C:/claudetools. syncro.md also read ~/.claude/identity.json before the repo copy -- the same bug that made remediation-tool's consent-audit report a fully consented tenant as RED. Root now resolves from the script's own location, with identity.json claudetools_root as the override. get-identity.sh had a chicken-and-egg bug: it read ${CLAUDETOOLS_ROOT:-.} but never set it, so every caller had to already know the root. It now self-resolves and exports CLAUDETOOLS_ROOT + VAULT_ROOT. Verified: all three command setup blocks authenticate against live Syncro; get-identity.sh works from any cwd and honors a pre-set root; gps-rmm autoenroll resolves its key from the vault. security-review: no findings. NOTE: both keys remain valid in git history. Rotation in the Syncro portal is the required follow-up -- this commit does not resolve that exposure. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-07-10 07:56:25 -07:00
parent c34cdb3a39
commit 2f35163866
10 changed files with 125 additions and 61 deletions
--- a/.claude/commands/inject-standards.md
+++ b/.claude/commands/inject-standards.md
@@ -23,7 +23,7 @@ Follow these steps exactly when /inject-standards is invoked:

 ### Step 2 — Auto-select relevant standards

-1. Read `D:/claudetools/.claude/standards/index.yml`.
+1. Read `.claude/standards/index.yml`.
 2. Review the descriptions for all entries.
 3. Select the 2–5 standards most relevant to either:
   - The task description in $ARGUMENTS, or
@@ -35,7 +35,7 @@ Follow these steps exactly when /inject-standards is invoked:

 For each selected standard (in order of relevance):

-1. Read the file at `D:/claudetools/.claude/standards/<slug>.md`.
+1. Read the file at `.claude/standards/<slug>.md`.
 2. Display a header:
   ```
   === STANDARD: <slug> ===
@@ -87,8 +87,8 @@ Reads the recent conversation, infers the task type, selects 2–5 most relevant

 ## Standards index location

-`D:/claudetools/.claude/standards/index.yml`
+`.claude/standards/index.yml`

 ## Standards files location

-`D:/claudetools/.claude/standards/<folder>/<name>.md`
+`.claude/standards/<folder>/<name>.md`
--- a/.claude/commands/syncro.md
+++ b/.claude/commands/syncro.md
@@ -106,43 +106,28 @@ Keys are baked into the skill below. To add a new user: generate a token in Sync
 ### Get API key

 ```bash
-BASE="https://computerguru.syncromsp.com/api/v1"
+# Resolves CLAUDETOOLS_ROOT + VAULT_ROOT from identity.json and reads the caller's
+# per-user Syncro key from the SOPS vault. Never hardcode a repo path or an API key.
+#
+# The key is per-user on purpose: every action in Syncro is attributed to the key
+# owner, so using someone else's key misattributes tickets, time, and invoices.
+#
+# NOTE: identity.json lives in the REPO (.claude/identity.json), not ~/.claude/.
+# Reading ~/.claude/identity.json first is a known bug — it silently picks up a
+# stale or absent file. syncro-env.sh resolves the repo copy from its own location.
+source "$(git rev-parse --show-toplevel)/.claude/scripts/syncro-env.sh" || {
+    echo "[ERROR] Cannot resolve Syncro credentials — run onboarding / check the vault" >&2
+    exit 1
+}

-# Get repo root from identity.json (set during machine onboarding)
-# Fallback to dynamic detection for legacy machines that haven't updated identity.json yet
-IDENTITY_PATH="${HOME}/.claude/identity.json"
-if [ ! -f "$IDENTITY_PATH" ]; then
-    # Try in-repo identity.json (gitignored, machine-specific)
-    REPO_ROOT=$(git rev-parse --show-toplevel 2>/dev/null)
-    if [ -n "$REPO_ROOT" ]; then
-        IDENTITY_PATH="$REPO_ROOT/.claude/identity.json"
-    fi
-fi
+BASE="$SYNCRO_BASE"
+API_KEY="$SYNCRO_API_KEY"
+REPO_ROOT="$CLAUDETOOLS_ROOT"

-if [ ! -f "$IDENTITY_PATH" ]; then
-    echo "[ERROR] Cannot locate identity.json - run onboarding first" >&2
+if [ -z "$API_KEY" ]; then
+    echo "[ERROR] No Syncro API key for user '${SYNCRO_USER:-unknown}'" >&2
    exit 1
 fi
-
-REPO_ROOT=$(jq -r '.claudetools_root // empty' "$IDENTITY_PATH")
-if [ -z "$REPO_ROOT" ]; then
-    # Legacy fallback for machines without claudetools_root in identity.json
-    REPO_ROOT=$(git rev-parse --show-toplevel 2>/dev/null)
-    if [ -z "$REPO_ROOT" ]; then
-        echo "[ERROR] claudetools_root not set in identity.json and not in a git directory" >&2
-        echo "[ERROR] Add 'claudetools_root' field to $IDENTITY_PATH" >&2
-        exit 1
-    fi
-    echo "[WARNING] Using git-detected repo root. Add 'claudetools_root' to identity.json to avoid this." >&2
-fi
-
-# Per-user keys — actions in Syncro are attributed to the key owner
-USER_ID=$(jq -r '.user // empty' "$IDENTITY_PATH")
-case "$USER_ID" in
-  mike)   API_KEY="T259810e5c9917386b-52c2aeea7cdb5ff41c6685a73cebbeb3" ;;
-  howard) API_KEY="Tde5174a6e9e312d14-02fd5bfe0f0ee40c87d027507c680e18" ;;
-  *)      echo "[ERROR] Unknown user '$USER_ID' in identity.json — cannot select Syncro API key" >&2; exit 1 ;;
-esac
 ```

 ### Ollama drafting
--- a/.claude/commands/wiki-compile.md
+++ b/.claude/commands/wiki-compile.md
@@ -36,17 +36,18 @@ want a periodic clean rebuild. For "I just did some work, capture it," use plain
 ## Phase 0 — Setup

 ```bash
-CLAUDETOOLS_ROOT="D:/claudetools"
-VAULT="$CLAUDETOOLS_ROOT/.claude/scripts/vault.sh"
+# Resolves CLAUDETOOLS_ROOT + VAULT_ROOT from identity.json and reads the caller's
+# per-user Syncro key from the SOPS vault. Never hardcode a repo path or an API key.
+# Exports: CLAUDETOOLS_ROOT, VAULT_ROOT, SYNCRO_USER, SYNCRO_BASE, SYNCRO_API_KEY.
+source "$(git rev-parse --show-toplevel)/.claude/scripts/syncro-env.sh" || true

-# Syncro auth (read-only operations only — GET requests ONLY in this skill)
-BASE="https://computerguru.syncromsp.com/api/v1"
-USER_ID=$(jq -r '.user // empty' "$CLAUDETOOLS_ROOT/.claude/identity.json")
-case "$USER_ID" in
-  mike)   API_KEY="T259810e5c9917386b-52c2aeea7cdb5ff41c6685a73cebbeb3" ;;
-  howard) API_KEY="Tde5174a6e9e312d14-02fd5bfe0f0ee40c87d027507c680e18" ;;
-  *)      echo "[WARNING] Unknown user — Syncro enrichment skipped" ; API_KEY="" ;;
-esac
+VAULT="$CLAUDETOOLS_ROOT/.claude/scripts/vault.sh"
+BASE="$SYNCRO_BASE"          # read-only: GET requests ONLY in this skill
+API_KEY="$SYNCRO_API_KEY"    # empty if no key vaulted for this user -> skip enrichment
+
+if [ -z "$API_KEY" ]; then
+  echo "[WARNING] No Syncro key for user '${SYNCRO_USER:-unknown}' — Syncro enrichment skipped"
+fi
 ```

 **Synthesis engine:** seed/full article drafting is done by a **Sonnet subagent** (Agent tool, `model: "sonnet"`), not Ollama. The main agent gathers sources + Syncro data, delegates the draft, then reviews it before writing. No Ollama dependency.
--- a/.claude/commands/wiki-lint.md
+++ b/.claude/commands/wiki-lint.md
@@ -10,7 +10,8 @@ Scan for clients and projects that have session logs but no wiki article.

 ```bash
 # List all client slugs that have session-logs but no wiki article
-cd D:/claudetools
+# Repo root is machine-specific (C:/ on some, D:/ on others) — never hardcode it.
+cd "$(git rev-parse --show-toplevel)"
 for dir in clients/*/session-logs; do
  slug=$(echo "$dir" | sed 's|clients/||;s|/session-logs||')
  wiki="wiki/clients/$slug.md"
@@ -96,13 +97,17 @@ For every client wiki article that contains a `Syncro customer ID` line, pull li
 ### Setup

 ```bash
-BASE="https://computerguru.syncromsp.com/api/v1"
-USER_ID=$(jq -r '.user // empty' "$CLAUDETOOLS_ROOT/.claude/identity.json")
-case "$USER_ID" in
-  mike)   API_KEY="T259810e5c9917386b-52c2aeea7cdb5ff41c6685a73cebbeb3" ;;
-  howard) API_KEY="Tde5174a6e9e312d14-02fd5bfe0f0ee40c87d027507c680e18" ;;
-  *)      echo "[SYNCRO] No API key for user '$USER_ID' — skipping Step 6" ; exit 0 ;;
-esac
+# Resolves CLAUDETOOLS_ROOT + VAULT_ROOT from identity.json and reads the caller's
+# per-user Syncro key from the SOPS vault. Never hardcode a repo path or an API key.
+source "$(git rev-parse --show-toplevel)/.claude/scripts/syncro-env.sh" || true
+
+BASE="$SYNCRO_BASE"
+API_KEY="$SYNCRO_API_KEY"
+
+if [ -z "$API_KEY" ]; then
+  echo "[SYNCRO] No API key for user '${SYNCRO_USER:-unknown}' — skipping Step 6"
+  exit 0
+fi
 ```

 ### For Each Client Article
--- a/.claude/scripts/gen_b64.py
+++ b/.claude/scripts/gen_b64.py
@@ -1,5 +1,5 @@
 """Generate base64 of the WatchdogAlertsSection component."""
-import base64, sys
+import base64, os, sys

 BT = chr(96)
 BULLET = "\xb7"
@@ -195,7 +195,9 @@ lines = [
 component = "".join(lines)
 b64 = base64.b64encode(component.encode("utf-8")).decode("ascii")

-with open("D:/claudetools/.claude/scripts/component.b64", "w") as f:
+# Write next to this script — the repo root is machine-specific, never hardcode it.
+out_path = os.path.join(os.path.dirname(os.path.abspath(__file__)), "component.b64")
+with open(out_path, "w") as f:
    f.write(b64)

 print(f"Component: {len(component)} chars")
--- a/.claude/scripts/get-identity.sh
+++ b/.claude/scripts/get-identity.sh
@@ -12,7 +12,14 @@
 # (but does NOT exit, so the caller continues). This ensures skills never break on
 # missing identity - they just attribute to "Unknown".

-IDENTITY_FILE="${CLAUDETOOLS_ROOT:-.}/.claude/identity.json"
+# Resolve the repo root from this script's own location (mirrors vault.sh) so callers
+# don't have to know it. An already-set CLAUDETOOLS_ROOT still wins, and identity.json's
+# claudetools_root overrides both. Never hardcode a drive letter.
+if [ -z "${CLAUDETOOLS_ROOT:-}" ]; then
+    export CLAUDETOOLS_ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")/../.." && pwd)"
+fi
+
+IDENTITY_FILE="$CLAUDETOOLS_ROOT/.claude/identity.json"

 if [ ! -f "$IDENTITY_FILE" ]; then
    echo "[WARNING] identity.json not found at $IDENTITY_FILE - attribution will be 'Unknown'" >&2
@@ -28,5 +35,11 @@ export USER_SHORT=$(jq -r '.user // "unknown"' "$IDENTITY_FILE" 2>/dev/null || e
 export MACHINE=$(jq -r '.machine // "unknown"' "$IDENTITY_FILE" 2>/dev/null || echo "unknown")
 export USER_EMAIL=$(jq -r '.email // "unknown@unknown.com"' "$IDENTITY_FILE" 2>/dev/null || echo "unknown@unknown.com")

+# identity.json is authoritative for the root and the vault location.
+_root_override=$(jq -r '.claudetools_root // empty' "$IDENTITY_FILE" 2>/dev/null)
+[ -n "$_root_override" ] && [ -d "$_root_override" ] && export CLAUDETOOLS_ROOT="$_root_override"
+export VAULT_ROOT=$(jq -r '.vault_path // empty' "$IDENTITY_FILE" 2>/dev/null)
+unset _root_override
+
 # Success
 return 0
--- a/.claude/scripts/gps-rmm-autoenroll.sh
+++ b/.claude/scripts/gps-rmm-autoenroll.sh
@@ -13,7 +13,8 @@ TS="$(date '+%Y-%m-%d %H:%M')"
 eval "$(bash .claude/scripts/rmm-auth.sh 2>/dev/null)" >/dev/null 2>&1
 if [ -z "${TOKEN:-}" ]; then echo "$TS auth FAILED" >> "$LOG"; exit 0; fi
 SEC="$(bash .claude/scripts/vault.sh get-field msp-tools/screenconnect.sops.yaml credentials.api_secret 2>/dev/null | tr -d '\r\n')"
-SK="Tde5174a6e9e312d14-02fd5bfe0f0ee40c87d027507c680e18"
+SK="$(bash .claude/scripts/vault.sh get-field msp-tools/syncro-howard credentials.credential 2>/dev/null | tr -d '\r\n')"
+if [ -z "$SK" ]; then echo "$TS syncro key read FAILED" >> "$LOG"; exit 0; fi

 OUT="$(RMM="$RMM" TOK="$TOKEN" SK="$SK" SC_SECRET="$SEC" python projects/gps-rmm-audit/tools/rebuild-and-push.py 2>/dev/null | tail -2)"
 PUSHED="$(echo "$OUT" | grep -oE 'pushed: [0-9]+' | grep -oE '[0-9]+' | head -1 || true)"
--- a/.claude/scripts/syncro-env.sh
+++ b/.claude/scripts/syncro-env.sh
@@ -0,0 +1,57 @@
+#!/usr/bin/env bash
+# syncro-env.sh — Resolve the repo root and the caller's Syncro API key from the SOPS vault.
+#
+# Source this instead of hardcoding a repo path or an API key:
+#   source "$(dirname "${BASH_SOURCE[0]}")/syncro-env.sh"     # or by absolute path
+#   curl -s "$SYNCRO_BASE/customers?api_key=$SYNCRO_API_KEY"
+#
+# Exports: CLAUDETOOLS_ROOT, VAULT_ROOT, SYNCRO_USER, SYNCRO_BASE, SYNCRO_API_KEY
+#
+# Root resolution mirrors vault.sh: derive from this script's own location, then let
+# identity.json's claudetools_root override. Never hardcode a drive letter — the repo
+# lives at C:/claudetools on some machines and D:/claudetools on others.
+#
+# Soft-fails like get-identity.sh: on any failure it warns, leaves SYNCRO_API_KEY empty,
+# and returns 1 without exiting, so a sourcing skill degrades instead of dying.
+
+_syncro_env_dir="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
+export CLAUDETOOLS_ROOT="$(cd "$_syncro_env_dir/../.." && pwd)"
+_identity="$CLAUDETOOLS_ROOT/.claude/identity.json"
+
+export SYNCRO_BASE="https://computerguru.syncromsp.com/api/v1"
+export SYNCRO_API_KEY=""
+export SYNCRO_USER=""
+export VAULT_ROOT=""
+
+if [ ! -f "$_identity" ]; then
+    echo "[WARNING] identity.json not found at $_identity — Syncro calls will be unauthenticated" >&2
+    return 1 2>/dev/null || exit 1
+fi
+
+# identity.json is authoritative for both the repo root and the vault location.
+_root_override=$(jq -r '.claudetools_root // empty' "$_identity" 2>/dev/null)
+[ -n "$_root_override" ] && [ -d "$_root_override" ] && export CLAUDETOOLS_ROOT="$_root_override"
+export VAULT_ROOT=$(jq -r '.vault_path // empty' "$_identity" 2>/dev/null)
+export SYNCRO_USER=$(jq -r '.user // empty' "$_identity" 2>/dev/null)
+
+# Per-user token — attribution in Syncro depends on using the right one.
+case "$SYNCRO_USER" in
+    mike)   _vault_path="msp-tools/syncro" ;;
+    howard) _vault_path="msp-tools/syncro-howard" ;;
+    *)      echo "[WARNING] No Syncro key vaulted for user '$SYNCRO_USER' — enrichment skipped" >&2
+            return 1 2>/dev/null || exit 1 ;;
+esac
+
+SYNCRO_API_KEY=$(bash "$CLAUDETOOLS_ROOT/.claude/scripts/vault.sh" \
+    get-field "$_vault_path" credentials.credential 2>/dev/null | tail -1)
+export SYNCRO_API_KEY
+
+if [ -z "$SYNCRO_API_KEY" ]; then
+    echo "[ERROR] Could not read $_vault_path from vault — is SOPS/age configured?" >&2
+    bash "$CLAUDETOOLS_ROOT/.claude/scripts/log-skill-error.sh" "syncro-env" \
+        "vault read failed for $_vault_path" --context "user=$SYNCRO_USER" >/dev/null 2>&1 || true
+    return 1 2>/dev/null || exit 1
+fi
+
+unset _syncro_env_dir _identity _root_override _vault_path
+return 0 2>/dev/null || true
--- a/CATALOG_CLIENTS.md
+++ b/CATALOG_CLIENTS.md
@@ -93,7 +93,7 @@

 #### Syncro PSA/RMM
 - **API Base:** https://computerguru.syncromsp.com/api/v1
- **API Key:** T259810e5c9917386b-52c2aeea7cdb5ff41c6685a73cebbeb3
+- **API Key:** vault `msp-tools/syncro` (field `credentials.credential`). Per-user key for Howard: `msp-tools/syncro-howard`. Read via `.claude/scripts/syncro-env.sh` — never hardcode.
 - **Subdomain:** computerguru
 - **Customers:** 5,064 (29 duplicates found)

--- a/CATALOG_SHARED_DATA.md
+++ b/CATALOG_SHARED_DATA.md
@@ -406,7 +406,7 @@

 ### Syncro (PSA/RMM) - AZ Computer Guru
 - **Service:** PSA/RMM platform
- **API Key:** T259810e5c9917386b-52c2aeea7cdb5ff41c6685a73cebbeb3
+- **API Key:** vault `msp-tools/syncro` (field `credentials.credential`). Per-user key for Howard: `msp-tools/syncro-howard`. Read via `.claude/scripts/syncro-env.sh` — never hardcode.
 - **Subdomain:** computerguru
 - **API Base URL:** https://computerguru.syncromsp.com/api/v1
 - **API Docs:** https://api-docs.syncromsp.com/