tq-forge-improvelisted

# /tq-forge-improve — targeted rewrite of low-scoring sections ## When to use A forged skill or agent scored <7 and you want to fix it without rewriting the whole artifact. Give it a slug (e.g. `notion-daily-summary`) or an absolute path; the skill identifies which dimensions are pulling the score down, rewrites only those sections, then verifies. ## Procedure ```bash export TQ_FORGE_HOME="${TQ_FORGE_HOME:-$HOME/.tq-forge}" S="${CLAUDE_PLUGIN_ROOT:-${TQ_FORGE_HOME:-$HOME/.tq-forge}/install}/scripts" source "$S/common.sh" && tq_ensure_home ``` 1. **Assess the artifact.** ```bash bash "$S/improve-loop.sh" assess "<slug>" ``` This prints per-dimension scores, highlights weak dims (<7), and gives a coaching hint for each. Capture which sections need work. 2. **Read the current content.** Read the SKILL.md (or AGENT.md + system-prompt.md) at the path `assess` printed. Understand what each weak section currently says. 3. **Rewrite only the weak sections:** | Dimension | Section to rewrite | |-----------|--------------------| | `clarity` | `## When to use` — <=3 sentences, concrete trigger | | `actionability` | `## Procedure` — add copy-pasteable commands | | `completeness` | Add missing section(s): Pitfalls / Verification / Tags | | `specificity` | Replace vague language with real paths / tool names | | `consistency` | Define the output-format schema once, up top (agents) | | `handoff_clarity` | `hand_off_to` in `tools.json` — ad