Files
last30days-skill/docs/plans/2026-03-08-fix-skill-md-save-section-regression-plan.md
Matt Van Horn 2f16ff1ee8 feat(gemini): add Gemini CLI extension support
Add gemini-extension.json manifest with correct array-format settings,
symlink skills/last30days/SKILL.md to root SKILL.md for Gemini skill
discovery, add Gemini install paths to bash for-loop in both main and
open variant, and add Gemini CLI install instructions to README.

Incorporates the good parts of PR #53 (manifest, paths, README) while
avoiding duplicate SKILL.md, tool name scattering, and allowed-tools
pollution that would have created maintenance issues.

Closes #45

Co-Authored-By: Alex Ferrari <alex@thealexferrari.com>
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-08 12:07:55 -07:00

3.2 KiB

title, type, status, date
title type status date
fix: Remove re-introduced Save Research to Documents section from SKILL.md fix completed 2026-03-08

fix: Remove re-introduced Save Research to Documents section from SKILL.md

Overview

PR merges on March 7 regressed SKILL.md by re-introducing the "Save Research to Documents" section that v2.9.4 (6d5acb9) intentionally removed. The save logic was moved into the Python script via --save-dir flag, but the merged branches (PR #48 Xiaohongshu, upstream merge) were forked before v2.9.4 and brought the old SKILL.md content back via merge resolution.

Result: The skill now saves research AFTER the "I'm now an expert" invitation via a separate Bash tool call, which causes extra cogitation time, "(No output)" noise, and sometimes hallucinated follow-up messages - the exact UX problems v2.9.4 fixed.

Root Cause

  1. Commit 6d5acb9 (v2.9.4, March 6) removed the ~45-line "Save Research to Documents" section and added --save-dir=~/Documents/Last30Days to the bash command
  2. PR #48 (Xiaohongshu) branch contained commit 9950d01 with the OLD save-via-Bash version
  3. Merging PR #48 on March 7 re-introduced the save section
  4. Upstream merge (28dff6e) compounded it

Three changes needed

All in SKILL.md:

1. Fix agent mode line (line 142)

Current (broken):

Agent mode still saves the research briefing to `~/Documents/Last30Days/` using the same logic as interactive mode (see "Save Research to Documents" section).

Should be:

Agent mode saves raw research data to `~/Documents/Last30Days/` automatically via `--save-dir` (handled by the script, no extra tool calls).

2. Add --save-dir back to bash command (line 186)

Current (broken):

python3 "${SKILL_ROOT}/scripts/last30days.py" "$ARGUMENTS" --emit=compact --no-native-web

Should be:

python3 "${SKILL_ROOT}/scripts/last30days.py" "$ARGUMENTS" --emit=compact --no-native-web --save-dir=~/Documents/Last30Days

3. Delete the entire "Save Research to Documents" section (lines 517-563)

Remove from ## Save Research to Documents through the --- separator before ## WAIT FOR USER'S RESPONSE. This is ~47 lines.

4. Update "STOP and wait" line (line 567)

Current:

**STOP and wait** for the user to respond.

Should be (matching v2.9.4):

**STOP and wait** for the user to respond. Do NOT call any tools after displaying the invitation. The research script already saved raw data to `~/Documents/Last30Days/` via `--save-dir`.

Post-edit steps

  1. Run bash /Users/mvanhorn/last30days-skill-private/scripts/sync.sh to deploy to ~/.claude/skills/
  2. Push to origin/main and upstream/main
  3. Verify with diff that installed skill matches repo

Acceptance Criteria

  • "Save Research to Documents" section is gone from SKILL.md
  • --save-dir=~/Documents/Last30Days is in the bash command
  • Agent mode line references --save-dir, not the deleted section
  • "STOP and wait" line includes the "do NOT call any tools" reinforcement
  • sync.sh deployed successfully
  • diff between repo and ~/.claude/skills/last30days/SKILL.md shows no differences
  • Pushed to origin and upstream