uditgoenka/autoresearch v1.9.12

v1.9.12 — /autoresearch skill reliably triggers on all subcommands

on GitHub

What's fixed

/autoresearch and its 9 subcommands (plan, debug, fix, security, ship, scenario, predict, learn, reason) now dispatch reliably on the first invocation. Previously the skill failed to trigger ~80% of the time, requiring repeated attempts (often 10+) before Claude Code would activate it.

Root cause

Claude Code routes slash commands through a description-based fuzzy matcher. None of the skill or command descriptions contained the literal /autoresearch... token the user would type — they only described behavior in prose ("Autonomous Goal-directed Iteration..."). Without any overlap between the user's input and the description, matching was unreliable and non-deterministic, especially when other skills with similar themes (ck-autoresearch, ralph-loop, etc.) sat in the same candidate set.

Fix

Every skill and slash-command description now begins with an explicit trigger sentence:

Use when user types /autoresearch: _{or ...}

This pattern applies uniformly across the Claude Code source (.claude/...), the plugin distribution mirror (claude-plugin/...), the OpenCode mirror (.opencode/..., uses /autoresearch_<sub> naming), and the Codex mirror (.agents/..., uses $autoresearch <sub> naming).

Changes

• 37 files updated (source + 3 mirror trees + marketplace + plugin.json + README badge).
• Version bump: 1.9.0 → 1.9.12 across plugin.json, marketplace.json, all 4 SKILL.md files, and guide/README.md badge.
• Zero behavior changes — this is a pure metadata patch.

Install / update

Plugin users: reinstall or update via your Claude Code plugin manager to pick up v1.9.12.

Source users: git pull on master and re-run your install script (bash scripts/install.sh or the OpenCode / Codex equivalents).

PR

Merged via #72.