v1.0.1 — post-review patch

Patch release responding to external technical review (2026-05-14). Four critical fixes, no rule semantics changed except as noted.

Fixes

1. Name standardized: tokensave — the README body and quick-start git clone command referred to the project as tokensave while the repo had been renamed to plz-save-token. The clone command was an actual broken command. Renamed back to the original tokensave; all 150 in-repo references aligned.

2. R4/R9 rules judge structure, not author identity — the previous version exempted the meta-skill from R4/R9 via meta_skill: true frontmatter (a rule-author-becomes-first-exception pattern). The rules are now reformulated to check actionable-content ratio: a SKILL.md > 8K passes if it has sufficient Phase/MODE definitions and decision-tree references. The skill's own SKILL.md passes its own audit on the structure check; no special-case exception is applied.

3. "External validation" wording weakened — the previous README overclaimed: R5 firing across three external frameworks is a trivial text-mention check, and R2/R3/R8 by manual inspection is anecdote. README now calls it a "cross-framework spot check, not yet a validation claim." Real external validation lands with v1.1 adapters and community-submitted baselines.

4. Core 5 vs Extended 4 marked — the 9-rule audit table presented all rules at equal weight. Pareto reality: R1, R2, R3, R5, R8 explain ~80% of cost impact. Those five are now marked ★ Core; R4/R6/R7/R9 are explicitly Extended. First-time audit triage focuses on the core five.

Evolution tags

• v0.1 — initial v1.0 release (pre-review state)
• v0.2 — Korean README + skill deploy guide + catalog surfacing
• v1.0 — same as v0.1 (initial public release)
• v1.0.1 — current head (post-review patch, this release)

Self-audit status

Without any meta-skill exception, the tool now passes its own audit cleanly: FAIL 0 · WARN 0 · PASS 4 · N/A 5. The rule and its author follow the same rule.

Full README: https://github.com/epoko77-ai/tokensave

v1.0 — Cost-optimization lint for multi-agent Claude Code harnesses

plz-save-token helps you build well-formed multi-agent Claude Code harnesses. Catches misallocated team compositions and burning cost patterns — not the multi-agent design itself. Companion to harness-diagnostic.

What ships in v1.0

• 10 categories × 31 sub-patterns, registered only when N≥2 independent external sources confirm
• 4 modes: Pre-flight · Authoring · Audit · Hook
• 3 headline artifacts:
  • references/task_to_model_matrix.md — 24-row matrix (Python / Haiku / Sonnet / Opus with citations)
  • scripts/model_selector.py — LLM-0 decision tree CLI
  • references/optimal_team_composition.md — 8 workload patterns × 5 cost-optimization patterns
• scripts/audit.py — 9 static rules across your ~/.claude (standard library only)
• scripts/estimate_cost.py — single-source pricing module
• scripts/hook_check.py — runtime guard (PreToolUse + UserPromptSubmit, never blocks)
• Community channels: issue templates for baseline submission, pattern proposal, and quality regression reports

Self-consistency

Tool passes its own audit — R1-R9 → FAIL 0, WARN 0, PASS 4, N/A 5. meta_skill: true frontmatter exempts the meta-skill from R4/R9 (context-bloat rules).

Quick start

git clone https://github.com/epoko77-ai/plz-save-token
cd plz-save-token
python3 scripts/audit.py                                      # catalog-wide
python3 scripts/model_selector.py --task "your task description"

Standard library only. Python 3.10+. No pip install.

Caveats (honest)

• Single-operator baseline (one 27-harness catalog). Inbound diagnostics welcome.
• Claude Code-specific. LangChain/CrewAI/AutoGen need adapters (planned).
• Pricing is point-in-time (2026-05-14). Manual update when Anthropic prices change.
• Quality regression risk: downgrade recommendations need golden-task validation. See README.

Full README: https://github.com/epoko77-ai/plz-save-token