Give the takedown researcher graphify, ast-grep, and lsp-probe on cases 736 and 803 #971

New issue

Open

opened 2026-10-09 16:58:42 +00:00 by jared · 0 comments

jared commented

2026-10-09 16:58:42 +00:00

Owner

Problem

Cases 736 and 803 stay at workaround (rank 1) in every graded run. The researcher used only Grep and Read. It never used graphify, ast-grep, or lsp-probe. The designer stops at the caller the ticket names.

Observed

Runs 736-20261008-112622 and 803-20261008-112907 scored workaround (1). No ADR text reached the designer in these runs, so ADRs do not explain the result.

Reproduce

n/a: design ticket. See docs/handoffs/2026-10-08-takedown-autoresearch.md, "Next" item 3.

Approach

Let the researcher use graphify, ast-grep, and lsp-probe on the scratch tree. Rerun 736 and 803. If both stay at 1, rerun with --designers 2. Record the mean rank per case.

Origin

Trigger: the 2026-10-08 ticket-takedown autoresearch session; freeze-and-merge bookkeeping on 2026-10-09.
Improvised this session: none.
Chain: workaround design <- designer stops at the ticket's named caller <- researcher saw only Grep/Read output (IMPROVISATION: none; observed in run replies)
Root candidate: none.
Where: plugins/os-sdlc/eval/ticket-takedown (researcher card, tool list)
Session: 58499175-8e67-4b3a-8cfe-ed601216e8da
Transcript: /home/jared/.claude/projects/-home-jared-dev-cc-os--claude-worktrees-prism-architect/58499175-8e67-4b3a-8cfe-ed601216e8da.jsonl

VERDICT: CREATE

## Problem Cases 736 and 803 stay at workaround (rank 1) in every graded run. The researcher used only Grep and Read. It never used graphify, ast-grep, or lsp-probe. The designer stops at the caller the ticket names. ## Observed Runs 736-20261008-112622 and 803-20261008-112907 scored workaround (1). No ADR text reached the designer in these runs, so ADRs do not explain the result. ## Reproduce n/a: design ticket. See docs/handoffs/2026-10-08-takedown-autoresearch.md, "Next" item 3. ## Approach Let the researcher use graphify, ast-grep, and lsp-probe on the scratch tree. Rerun 736 and 803. If both stay at 1, rerun with `--designers 2`. Record the mean rank per case. ## Origin - Trigger: the 2026-10-08 ticket-takedown autoresearch session; freeze-and-merge bookkeeping on 2026-10-09. - Improvised this session: none. - Chain: workaround design <- designer stops at the ticket's named caller <- researcher saw only Grep/Read output (IMPROVISATION: none; observed in run replies) - Root candidate: none. - Where: plugins/os-sdlc/eval/ticket-takedown (researcher card, tool list) - Session: 58499175-8e67-4b3a-8cfe-ed601216e8da - Transcript: /home/jared/.claude/projects/-home-jared-dev-cc-os--claude-worktrees-prism-architect/58499175-8e67-4b3a-8cfe-ed601216e8da.jsonl VERDICT: CREATE

jared added the

plugin/os-sdlc

filed-by/agent

category/eval-run

labels

2026-10-09 16:58:42 +00:00