github uditgoenka/autoresearch v1.6.1
v1.6.1 — Hardened Git-as-Memory Mechanism

latest releases: v2.2.1, v2.2.0, v2.1.4...
3 months ago

What's Changed

Fixes #29 — The "git is memory" mechanism that powers inter-iteration learning in the autonomous loop wasn't working reliably. The agent would skip reading git history, use destructive git reset --hard (destroying experiment memory), and enter the loop without verifying git state.

This release comprehensively hardens the protocol with 10 targeted fixes discovered through a 20-iteration edge case audit.

Highlights

New: Phase 0 — Precondition Checks
Before entering the loop, the agent now verifies: git repo exists, working tree is clean, no stale lock files, HEAD is attached, and detects git hooks (including husky and pre-commit framework). Blocks loop entry on failures.

Mandatory Git History Reading
Phase 1 now uses MUST run: language for git log --oneline -20 and git diff HEAD~1. Phase 2 requires consulting git history before ideation. The agent can no longer skip reading its own experiment history.

Safe Rollbacks with safe_revert()
All 5 rollback sites in Phase 6 now use a safe_revert() helper that tries git revert HEAD --no-edit first (preserving experiment history), falling back to git reset --hard HEAD~1 only when revert conflicts occur.

Robust Commit Phase

  • Nothing-to-commit detection (no-op status instead of confusing errors)
  • git add -A scope warning with verification step
  • Hook failure recovery (2 attempts, never --no-verify)

All Changes

# Severity Fix
1 CRITICAL Phase 0 precondition checks before loop entry
2 CRITICAL safe_revert() with conflict fallback across all rollback sites
3 HIGH Mandatory git log + git diff reading (MUST language)
4 HIGH Hook failure recovery in loop body
5 MEDIUM Nothing-to-commit handling (no-op status)
6 MEDIUM git add -A scope warning + verification
7 MEDIUM Expanded hook detection (husky, pre-commit-config, commit-msg)
8 LOW Guard failure recovery uses safe_revert()
9 LOW New statuses: no-op, hook-blocked + valid statuses list
10 LOW Final summary includes skipped iteration count

Files Changed

  • skills/autoresearch/references/autonomous-loop-protocol.md — +123/-25 lines (core protocol)
  • skills/autoresearch/SKILL.md — version bump to 1.6.1
  • README.md — version badge + expanded Rule #7

Verification

  • 20-iteration edge case audit via /autoresearch:scenario
  • 10-iteration stability verification via /autoresearch:debug
  • 5-iteration dry-run against Issue #29

Full Changelog: v1.6.0...v1.6.1

Don't miss a new autoresearch release

NewReleases is sending notifications on new releases.