council-of-high-intelligence

Under set -euo pipefail, the missing file makes sed fail, and pipefail causes the pipeline with head to fail before ${PROJECT_VERSION:-0.0.0} runs. Append || true to the command substitution:

PROJECT_VERSION="$(sed -nE 's/^## \[([0-9]+\.[0-9]+\.[0-9]+)\].*/\1/p' "${SCRIPT_DIR}/CHANGELOG.md" 2>/dev/null | head -1 || true)"
PROJECT_VERSION="${PROJECT_VERSION:-0.0.0}"

Alternatively, check for the file before running sed and assign 0.0.0 when it is absent. This fix is tracked in PR #60.

Replace the non-portable glibc/BSD %-d directive with explicit date formatting. In scripts/gen-star-history.py, change:

updated = pts[-1][0].strftime("%b %-d, %Y")

to:

d = pts[-1][0]
updated = f"{d:%b} {d.day}, {d.year}"

%-d is not supported by Windows Python, while d.day produces the same unpadded day number portably.

Upgrade to a release containing PR #61, which gates Ollama availability on a non-empty parsed model list. If patching scripts/detect-providers.sh manually, parse ollama list first and only set ollama_available=true when at least one model row exists:

if model_list="$(run_with_timeout ollama list 2>/dev/null)"; then
  ollama_model_names="$(printf '%s\n' "$model_list" | tail -n +2 | awk 'NF {print $1}')"
  if [ -n "$ollama_model_names" ]; then
    ollama_available=true
    ollama_models="$(printf '%s\n' "$ollama_model_names" | head -5 | sed 's/.*/"&"/' | paste -sd ',' -)"
  fi
fi

As a temporary workaround, pull at least one model and verify that ollama list shows a data row, not only the header:

ollama pull llama3.2
ollama list

Upgrade to the skill version containing the feat: add enforcement call tracking to session metadata change. The fix adds two fields to Session Metadata: enforcement_calls, the total number of enforcement model calls outside normal rounds, and enforcement_breakdown, the per-trigger counts. Keep rounds_run unchanged because enforcement probes are not rounds. Example output: enforcement_calls: 2 and enforcement_breakdown: { agreement_check: 2 }. If you are pinned to commit 0f63c19, these calls are not recorded, so benchmark token savings can be overstated; update or manually count enforcement dispatches from dissent quota, novelty gate, agreement check, anti-recursion, and missing/parseable STANCE: handling.

Apply or merge PR #48 for the drift-guard and stale-documentation fixes, then ensure SKILL.opencode.md matches SKILL.md v1.2.0. Required content fixes: add E3 per-member reasoning_method injection from commit cbc2ac0; add E7 ./.council.yaml Project Overrides and the E9 confidence-weighted tally from commit 4ce07d0; and restore the Asset Resolution section. For E9, calculate vote weight as base weight × confidence factor using high=1.0, med=0.75, and low=0.5, while reporting W_total at base weights. Extend guards by adding SKILL.opencode.md existence and install.sh --dry-run --opencode checks to scripts/council-simulation-checklist.sh, adding the file to markdownlint in .github/workflows/lint.yml, and adding a fourth coordinator-file parity checkbox to the PR template. Also correct CLAUDE.md and add the missing Vote Tally field to Demo E. Verify with: grep -c reasoning_method SKILL.md SKILL.opencode.md, grep -ci council.yaml SKILL.md SKILL.opencode.md, sed -n '456,470p' SKILL.opencode.md, and grep -rn opencode scripts/council-simulation-checklist.sh .github/workflows/lint.yml .github/PULL_REQUEST_TEMPLATE.md.