h-dv/code-index

Watch

Fork 0

Before/after view of a change: include_hunks on changed_symbols/review_diff, read_code rev= (zero-fallback) #558

New issue

Closed

opened 2026-11-09 12:29:11 +01:00 by buildagent · 2 comments

buildagent commented

2026-11-09 12:29:11 +01:00

Member

Part of the zero-fallback epic (#555). Cause 2: reviewing a change needs git show.

Observed

review_diff/changed_symbols say WHICH symbols changed (risk, test reach, freshness), but a reviewer needs WHAT changed. read_code serves only the working tree, so the lead reviewed every worker diff with git show (2026-11-09; operator: "do you use code-index???").

Proposal

changed_symbols/review_diff take include_hunks: true: the unified-diff hunk clipped to each changed symbol's span (base and head side), under the existing token budget, with hunks_truncated disclosed. A deleted symbol gets its base-side body.
read_code takes rev: "<commit-ish>": serves a path, range or symbol id as of that revision (base-side bytes from git; symbol spans from parsing the base-side file, which changed_symbols already does for classification).
A whole review then stays in code-index: review_diff(include_hunks) → find_callers on a finding → read_code(rev=base) for the old caller.

Acceptance

e2e: rename + body change + deletion in one diff → hunks per symbol; read_code(rev=base) returns the deleted symbol.
Benchmark task "review a diff" (sub-issue 1) needs zero shell calls.
Mutation: hunks not clipped to spans → the per-symbol test goes red.

Part of the zero-fallback epic (#555). Cause 2: **reviewing a change needs `git show`.** ## Observed `review_diff`/`changed_symbols` say WHICH symbols changed (risk, test reach, freshness), but a reviewer needs WHAT changed. `read_code` serves only the working tree, so the lead reviewed every worker diff with `git show` (2026-11-09; operator: "do you use code-index???"). ## Proposal - `changed_symbols`/`review_diff` take `include_hunks: true`: the unified-diff hunk clipped to each changed symbol's span (base and head side), under the existing token budget, with `hunks_truncated` disclosed. A deleted symbol gets its base-side body. - `read_code` takes `rev: "<commit-ish>"`: serves a path, range or symbol id as of that revision (base-side bytes from git; symbol spans from parsing the base-side file, which changed_symbols already does for classification). - A whole review then stays in code-index: `review_diff(include_hunks)` → `find_callers` on a finding → `read_code(rev=base)` for the old caller. ## Acceptance - e2e: rename + body change + deletion in one diff → hunks per symbol; `read_code(rev=base)` returns the deleted symbol. - Benchmark task "review a diff" (sub-issue 1) needs zero shell calls. - Mutation: hunks not clipped to spans → the per-symbol test goes red.

buildagent added the

Priority

High

Kind/Feature

labels

2026-11-09 12:29:11 +01:00

buildagent added a new dependency

2026-11-09 12:29:11 +01:00

#555 Agents still need shell fallbacks to finish code investigation and review

buildagent referenced this issue

2026-11-09 12:29:12 +01:00

Agents still need shell fallbacks to finish code investigation and review #555

buildagent referenced this issue

2026-11-09 13:03:23 +01:00

Fallback-rate metric: measure shell-on-indexed-code per agent session + agent task benchmark (zero-fallback, do first) #556

buildagent referenced this issue

2026-11-09 13:03:23 +01:00

Agents still need shell fallbacks to finish code investigation and review #555

buildagent commented

2026-11-09 14:34:13 +01:00

Author

Member

Done on branch codex-558 (de7bfc2a, 805492ef), merged into integration-1109 (ecbe801a).

read_code rev=<commit-ish>: path and range reads come from git through gitdiff::base_blob, under the same budget and total_lines contract. Refusals are typed: revision_symbol_unsupported, invalid_base_ref, revision_file_unavailable.
changed_symbols/review_diff include_hunks: each hunk is owned by the innermost changed symbol. Containers keep only their own lines and get hunk_in_children. The budget is spent leaves-first under the 4000-token cap, and omissions are disclosed. The default reply is byte-identical to before.
Wire: SourceSymbol.end_line defaults for old daemons. No new RPC was needed, and both legs are green. Six mutations were run.
The worker reported three "would have used #558 here" moments; after the rebuild, the new tool answered its own review.

Done on branch `codex-558` (`de7bfc2a`, `805492ef`), merged into `integration-1109` (`ecbe801a`). - `read_code rev=<commit-ish>`: path and range reads come from git through `gitdiff::base_blob`, under the same budget and `total_lines` contract. Refusals are typed: `revision_symbol_unsupported`, `invalid_base_ref`, `revision_file_unavailable`. - `changed_symbols`/`review_diff include_hunks`: each hunk is owned by the innermost changed symbol. Containers keep only their own lines and get `hunk_in_children`. The budget is spent leaves-first under the 4000-token cap, and omissions are disclosed. The default reply is byte-identical to before. - Wire: `SourceSymbol.end_line` defaults for old daemons. No new RPC was needed, and both legs are green. Six mutations were run. - The worker reported three "would have used #558 here" moments; after the rebuild, the new tool answered its own review.

buildagent referenced this issue

2026-11-10 08:48:18 +01:00

Tool discovery and replies cost extra calls, while review and navigation still have gaps #521

buildagent referenced this issue from a commit

2026-11-10 10:34:57 +01:00

feat(mcp): expose revision reads and symbol diff hunks (#558)

buildagent referenced this issue from a commit

2026-11-10 10:34:57 +01:00

fix(mcp): keep diff hunks with changed leaves (#558)

buildagent referenced this issue from a commit

2026-11-10 10:34:57 +01:00

Merge codex-558: before/after view of a change (#558)

buildagent referenced this issue

2026-11-11 07:25:31 +01:00

read_code with rev rejects a valid git path when its parent directory no longer exists in the working tree #568

buildagent referenced this issue

2026-11-11 16:46:41 +01:00

Tool discovery and replies cost extra calls, while review and navigation still have gaps #521

buildagent referenced this issue

2026-11-12 02:34:37 +01:00

Tool discovery and replies cost extra calls, while review and navigation still have gaps #521

buildagent referenced this issue

2026-11-12 22:22:54 +01:00

Tool discovery and replies cost extra calls, while review and navigation still have gaps #521

buildagent commented

2026-11-12 22:23:09 +01:00

Author

Member

Merged into #521.
mcp_smoke.rs::review_bytes_hunks_are_clipped_and_defaults_stay_unchanged grades hunks; revision_read_survives_removed_parent grades history.
Remaining: Review classification and hunks, including revision-by-symbol and zero-shell benchmark acceptance.

Merged into #521. mcp_smoke.rs::review_bytes_hunks_are_clipped_and_defaults_stay_unchanged grades hunks; revision_read_survives_removed_parent grades history. Remaining: Review classification and hunks, including revision-by-symbol and zero-shell benchmark acceptance.

buildagent closed this issue

2026-11-12 22:23:10 +01:00

No milestone

No project

No assignees

1 participant

Notifications

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Blocks

#555 Agents still need shell fallbacks to finish code investigation and review

h-dv/code-index

Reference

h-dv/code-index#558

No description provided.

Rows
Columns