CASAN/AINative_OKR_CASAN5/packages/casan-harness/scripts/bash/telemetry-integrity.sh

#!/usr/bin/env bash
set -uo pipefail

# CASAN H5 — Telemetry integrity proof (Track A, V9).
#
# Token/cost telemetry (provider-usage.jsonl, cost/metrics.jsonl) previously sat
# OUTSIDE the signed audit chain, so a forger could rewrite token counts to hide
# cost abuse and nothing would detect it. This binds those files to a signed
# manifest: any byte change flips the manifest head hash, and because the head
# is RSA-signed with an off-repo key, a forger cannot re-sign a rewritten head.
#
# Usage:
#   telemetry-integrity.sh sign     — hash telemetry files, write + sign manifest head
#   telemetry-integrity.sh verify   — recompute, compare head, verify signature
#
# Key resolution (sign): CASAN_AUDIT_PRIV, else level5/central-governance/audit-private.pem
# Key resolution (verify): CASAN_AUDIT_PUB, else level5/central-governance/audit-public.pem
#
# Outputs (under .specify/logs/level5/):
#   telemetry-manifest.json — {basename: sha256} for each telemetry file
#   telemetry-head.txt      — sha256 over the canonical manifest text
#   telemetry-head.sig      — RSA signature of telemetry-head.txt (when a key exists)
#
# Exit: 0 ok, 1 tamper/mismatch/invalid-signature, 64 usage.

SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
source "$SCRIPT_DIR/casan-paths.sh"
PROJECT_ROOT="$CASAN_APP_ROOT"
L5_DIR="$CASAN_STATE_ROOT/logs/level5"
COST_DIR="$CASAN_STATE_ROOT/logs/cost"
mkdir -p "$L5_DIR"

CMD="${1:-}"
MANIFEST="$L5_DIR/telemetry-manifest.json"
HEAD_FILE="$L5_DIR/telemetry-head.txt"
HEAD_SIG="$L5_DIR/telemetry-head.sig"
AUDIT_PRIV="${CASAN_AUDIT_PRIV:-$CASAN_GOVERNANCE_ROOT/audit-private.pem}"
AUDIT_PUB="${CASAN_AUDIT_PUB:-$CASAN_GOVERNANCE_ROOT/audit-public.pem}"

# Telemetry files to bind. Missing files hash to the literal "MISSING" so the
# manifest is stable and a deletion is itself a detectable change.
TELEMETRY_FILES=(
  "$L5_DIR/provider-usage.jsonl"
  "$COST_DIR/metrics.jsonl"
)

compute_head() {
  # Prints: manifest-json on line 1, head-hash on line 2.
  python - "$@" <<'PY'
import hashlib, json, os, sys
files = sys.argv[1:]
manifest = {}
for path in files:
    name = os.path.basename(path)
    if os.path.isfile(path):
        with open(path, "rb") as f:
            manifest[name] = hashlib.sha256(f.read()).hexdigest()
    else:
        manifest[name] = "MISSING"
canonical = json.dumps(manifest, sort_keys=True, separators=(",", ":"))
head = hashlib.sha256(canonical.encode()).hexdigest()
print(canonical)
print(head)
PY
}

case "$CMD" in
  sign)
    OUT="$(compute_head "${TELEMETRY_FILES[@]}")"
    CANON="$(printf '%s' "$OUT" | sed -n '1p')"
    HEAD="$(printf '%s' "$OUT" | sed -n '2p')"
    printf '%s' "$CANON" > "$MANIFEST"
    printf '%s' "$HEAD" > "$HEAD_FILE"
    if [[ -f "$AUDIT_PRIV" ]] && command -v openssl >/dev/null 2>&1; then
      openssl dgst -sha256 -sign "$AUDIT_PRIV" -out "$HEAD_SIG" "$HEAD_FILE"
      echo "TELEMETRY_INTEGRITY_SIGNED head=$HEAD anchor=signed files=${#TELEMETRY_FILES[@]}"
    else
      rm -f "$HEAD_SIG"
      echo "TELEMETRY_INTEGRITY_SIGNED head=$HEAD anchor=unsigned files=${#TELEMETRY_FILES[@]} (no private key)"
    fi
    ;;
  verify)
    if [[ ! -f "$HEAD_FILE" ]]; then
      echo "TELEMETRY_INTEGRITY_MISSING no telemetry-head.txt (run: telemetry-integrity.sh sign)" >&2
      exit 1
    fi
    OUT="$(compute_head "${TELEMETRY_FILES[@]}")"
    HEAD_NOW="$(printf '%s' "$OUT" | sed -n '2p')"
    HEAD_STORED="$(cat "$HEAD_FILE")"
    if [[ "$HEAD_NOW" != "$HEAD_STORED" ]]; then
      echo "TELEMETRY_INTEGRITY_MISMATCH computed=$HEAD_NOW stored=$HEAD_STORED" >&2
      exit 1
    fi
    if [[ -f "$HEAD_SIG" && -f "$AUDIT_PUB" ]] && command -v openssl >/dev/null 2>&1; then
      if ! openssl dgst -sha256 -verify "$AUDIT_PUB" -signature "$HEAD_SIG" "$HEAD_FILE" >/dev/null 2>&1; then
        echo "TELEMETRY_INTEGRITY_SIGNATURE_INVALID head=$HEAD_STORED" >&2
        exit 1
      fi
      echo "TELEMETRY_INTEGRITY_VALID anchor=signed head=$HEAD_STORED"
    else
      # SEC-01 (H-01): enforced mode treats a missing/unverifiable signature as FAIL,
      # otherwise deleting telemetry-head.sig after rewriting token counts would pass.
      if [[ "${CASAN_PROFILE:-}" == "prod" || "${CASAN_VERIFY_STRICT:-}" == "1" ]]; then
        MISSING=""
        [[ -f "$HEAD_SIG" ]] || MISSING="$MISSING head-sig"
        [[ -f "$AUDIT_PUB" ]] || MISSING="$MISSING pubkey"
        command -v openssl >/dev/null 2>&1 || MISSING="$MISSING openssl"
        echo "TELEMETRY_INTEGRITY_UNSIGNED_STRICT_FAIL head=$HEAD_STORED missing=${MISSING# }" >&2
        exit 1
      fi
      echo "TELEMETRY_INTEGRITY_VALID anchor=unsigned head=$HEAD_STORED"
    fi
    ;;
  *)
    echo "Usage: telemetry-integrity.sh {sign|verify}" >&2
    exit 64
    ;;
esac