Repository navigation
Conversation
Each feature gets its own file under hooks/features/, called by name from features/index.ts in a fixed order: its own hooks, the session and turn events, its band line and its tool-call phases. One tool.call hook (tool-chain.ts) runs Bash and file handling in fixed phases: bashRewrite, bashGate, the call, bashAfter, editAfter, readOrEditAfter. The status line is built from parts (status-line.ts). runTool moves to altimate-cli.ts with a `write` option that goes through the CLI with --allow-write. dbt-command.ts finds the dbt invocations in a shell command line. New userConfig fields: maxScanBytes, healthBlocksPush, healthConfigPath, commentsPollSeconds, commentsInContext. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Lists the model's CTEs through altimate-code's dbt_cte_profile, then times them one call at a time so the pane fills live and Stop works between queries. Marginal times and hot/warm/cool tiers are worked out in the mod. Runs only on request, and the finished profile reaches Claude with its next tool result. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Subshells after keywords, case patterns, process substitution, arrays, arithmetic and function definitions parse again; quotes in a heredoc inside a substitution are text; a function or case pattern named dbt is not an invocation. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…laude a bounded, quoted summary Stop pressed while the list loads ends the run before any query; a run ends with its session (a /clear or reload) and hands nothing to the next one; a second command while one runs starts nothing; a refusal ends the run at the first CTE; a failed CTE leaves the next one with its total but no own part; and the note for Claude is quoted data under a size cap, delivered once with the main conversation's next tool result (a read too, never a subagent's). Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
…he profile to Claude only once Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
…ight before each query, and ignores a subagent's reads A call with no usable answer ends the run instead of starting the next prefix query while the last may still be running. Finishing is bound to the run's session, a failed CTE costs only the next row its own part, and the Tell Claude button waits for the run to stop. Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
|
Review of this PR by two independent reviewers (Claude and Codex), then a fix gate (also both) on the fixes. Every finding and what happened to it (fixes are in c31bf14): Fixed
Left as is
|
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
|
Final-gate round. Claude: the two "Ended early" message writes could land after the session ended or a new run started, overwriting the clean-up state. Fixed in 21dd352: each write now checks the live run inside the state update. Its |
|
review-loop: APPROVED at 21dd352 |
Code Review SummaryStatus: No Issues Found | Recommendation: Merge Verification of previous findings
Files Reviewed (2 files)
Notes: incremental review of Previous Review Summary (commit 21dd352)Current summary above is authoritative. Previous snapshots are kept for context only. Previous review (commit 21dd352)Status: 4 Issues Found | Recommendation: Address before merge Overview
Issue Details (click to expand)WARNING
SUGGESTION
Files Reviewed (21 files)
Notes: the current diff still contains the Reviewed by deepseek-v4.1-flash-1017 · Input: 49.7K · Output: 9.3K · Cached: 311.6K |
…ailed CTEs in a long note, and drops the note only after it is sent Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
What this does
/profile-ctes [model]lists a dbt model's CTEs with altimate-code'sdbt_cte_profiletool, then times them one tool call at a time, so the pane fills live. Each CTE is one warehouse query that holds the CTEs before it (a warehouse may skip those the CTE does not use); nothing runs on edit.altimate-cte-profile: rows slowest first, each with its own added time (marginal), total query time, row count and a hot/warm/cool tier. Hot is half the largest marginal time or more, warm a fifth, as in dbt Power User./profile-ctesprofiles the.sqlmodel last read or edited by the main conversation (not a subagent's read, not a file undertarget/ordbt_packages/).--allow-write(it starts the dbt adapter), so it never uses the read-only MCP server.dbt_cost_estimateis still a placeholder on this base.types/index.d.ts:CteProfileStatefilled in (cteProfilewas already declared by the foundation).This needs the altimate-code PR for
dbt_cte_profile(AltimateAI/altimate-code#1731) merged and released first. The branch contains the foundation commit, which drops out of the diff when the foundation merges.Checks
claude plugin test ./plugins/altimate-code: 233 pass (new: tiers, marginal times, summary, live progress, Stop, a timed-out query, missing tool, no model, a subagent's read).bash tests/*.shpass.claude plugin validate --stricton the marketplace passes; on the plugin it reports warnings fromhooks/lineage.test.tsx(not touched here) and none fromcte-profile.Review: two independent reviewers (Claude and Codex). Fixed after review: a timed-out query kept the run going, Stop or a
/cleararriving just before a query still let it start, a finishing run could write its result back after a/clear, a subagent's.sqlread changed the bare-command target, one failed CTE hid the own times of all later ones, Tell Claude shown while stopping. Left as is: the result reaches Claude with the next main-conversation tool result (also offered by the Tell Claude button); a window of one clock read insiderunTool, a shared file, between the last Stop check and the process start; a project that sits inside a folder namedtargetis not remembered for the bare command.Changelog
/profile-ctescommand: times each CTE of a dbt model and lists them slowest first.Demo video (hosted)
P-03.mp4 — watch on Notion (Altimate workspace; recorded from the approved head of this PR).
🤖 Generated with Claude Code