mirror of
https://github.com/colbymchenry/codegraph.git
synced 2026-11-15 01:29:02 +08:00
de0a4eb362bbe4deb412f79aedcce11ee036dfbc
6
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 Colby MchenryandClaude Opus 5.5
|
cb7fb61e76 |
fix(docs): agent and command lists match the code; worktree warning drops init -i (#3067)
The warning for a git worktree answered from another checkout's index
still said to run "codegraph init -i", a flag that has done nothing
since init started indexing by default. Its one-line form is prepended
to every read-tool result and codegraph_status embeds the long form, so
both told the agent to build the index, which the server instructions
rule out ("Indexing is the user's decision ... don't run it yourself").
The CLI's codegraph status now says: Run "codegraph init" in this
worktree. What an agent reads (the read-tool notice, which also reaches
it through the prompt hook and the explore/node CLI, and
codegraph_status) says the user can run "codegraph init" there.
Docs that no longer matched the code:
- AGENTS.md listed 4 of the 11 installer targets, called a new one "a
5th agent", said the contract suite covers "all 4" targets, and listed
11 of the 24 subcommands. It now lists every target and subcommand,
and names the hidden ones (serve --mcp, prompt-hook, and ui/web unless
CODEGRAPH_UI=1).
- README's Supported Agents section said no instructions file is
written, but Claude Code, Codex CLI, Gemini CLI and opencode get the
short CodeGraph section. Its Codex entry now says the Codex IDE
extension reads the same config: $CODEX_HOME/config.toml (default
~/.codex/config.toml) for a global install, and .codex/config.toml,
used only in a trusted project, for a local one.
- README's WSL2 "Transport closed" entry named only the in-process
fallback. The daemon also moves its socket to the temp directory when
the project's filesystem can't host one.
- init -i and -iv leftovers in internal docs, scripts, the add-lang
skill and the repo's own Cursor rule, which now also leaves indexing
to the user. Two stay on purpose: scripts/agent-eval/audit.sh can
audit a published version older than 0.9.8, where init built the
index only with -i (now commented there), and
docs/benchmarks/codegraph-ab-matrix.md records a 0.9.4 run.
Validation:
- worktree-detection.test.ts: on main 61776927's worktree.ts and
tools.ts, 4 of its 20 tests fail (the CLI wording, the agent wording,
the read-tool notice and codegraph_status). With only the worktree.ts
fix, the codegraph_status test still fails. All 20 pass here, and so
do mcp-nested-repo-projectpath.test.ts's 13.
- Full suite on this patch over
|
||
|
Colby MchenryandClaude Opus 5.5
|
45a29b4685 |
fix(worktree): a linked worktree is recognized however Windows spells its path (#2937)
On Windows one directory has several spellings. VS Code and Cursor pass a
lowercase drive letter, NTFS ignores case, a subst drive is a second name
for a folder, and long names have 8.3 aliases. git answers with the
spelling on disk, but src/sync/worktree.ts compared its paths with the
caller's as strings, after a JS realpath that keeps the caller's spelling.
The main checkout's git common dir comes back relative (".git"), so it
took the caller's spelling while a linked worktree's came back absolute
in git's. A worktree inside the indexed checkout, spelled unlike git,
failed all four comparisons:
- detectWorktreeIndexMismatch found no mismatch, so codegraph status
(text and --json) dropped the borrowed-index warning;
- nestedRepositoryBelow took the worktree for a different repository, so
the MCP tools refused it as unindexed and told the agent not to call
codegraph for it again.
Fix: the four comparisons now ask whether two paths name one directory,
with the identity canonicalProjectRoot already uses (the native realpath,
lowercased on Windows), falling back to the JS realpath where the native
call fails. Returned paths keep the caller's spelling for display. A
nested repository below an index root outside git is still reported, as
before.
Validation (Windows 11, Node 22):
- New Windows-gated tests with a lowercase drive letter, at the helper
level and through ToolHandler (search and status, with and without a
default project): the 3 behaviour tests fail on main's worktree.ts and
pass with the fix. Both files: 32 passed.
- Fixture: a checkout with a worktree under .claude/worktrees and an
embedded clone, spelled four ways (on-disk, lowercase drive, other
capitals, subst drive). Helper verdicts wrong: 6 on main, 0 with the
fix. codegraph status warned in 2 of 8 runs on main, 8 of 8 with the
fix. The real MCP server refused the lowercase, capitals and subst
spellings on main; with the fix it serves all four with the
borrowed-index notice, codegraph_status warns, and the embedded clone
is still refused. An 8.3 short name (the contributor's case) went from
2 wrong verdicts to 0.
Reported in PR #2517 by @nk33-dev; this reimplements the fix
independently.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
||
|
Colby MchenryandClaude Opus 4.8
|
f73227d2f5 |
fix(mcp): don't warn "different git working tree" for submodules covered by the parent index (#1031, #1033) (#1039)
Indexing a super-repo now descends into its submodules and gitlinked clones, so a query run from inside one resolves up to the parent's unified index — whose graph DOES contain that nested repo's files. But the git-worktree-mismatch warning still fired, telling the agent the results were from "a different working tree" and to run `codegraph init -i` — which would split the submodule back into its own index and undo the unified view. A false positive carrying harmful advice. Distinguish a genuine borrowed worktree (the SAME repository on a different branch — shares a git common dir with the index root) from a submodule/embedded clone (a DIFFERENT repository — its own common dir), and suppress the warning only for the latter. Add gitCommonDir() for the check. The issue-#155 linked-worktree case is unchanged. Verified end-to-end: the warning no longer fires for a submodule-rooted MCP session and still fires for a real linked worktree. Edit-sync (manual sync + the live watcher) keeps the nested repo's files current on both macOS and Linux (active-submodule and bare-gitlink shapes), so suppressing the warning is safe. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Colby MchenryandClaude Opus 4.8
|
62d5cdde2c |
fix(mcp): re-resolve a path's index every call so worktree state isn't pinned (#926) (#939)
A long-lived MCP server (the shared daemon) cached both the project handle (projectCache) and the worktree-mismatch verdict (worktreeMismatchCache) for its whole lifetime — cleared only on shutdown — keyed off the input path and never re-checked. So a worktree path first resolved BEFORE it had its own .codegraph/ — when the walk-up reached the main checkout — stayed pinned to the main checkout: every query kept hitting the wrong index and every result carried a false "this index belongs to a different git working tree" warning, until the server restarted. The CLI was correct (fresh process per run); re-indexing didn't help. - getCodeGraph: re-resolve findNearestCodeGraphRoot on every call (cheap stat walk, no git, no reopen) and cache the open DB by RESOLVED ROOT only. Drops the input-path short-circuit that pinned the first resolution, and the double-keying (which also double-closed each instance in closeAll()). - worktreeMismatchFor: key the verdict on (startPath, indexRoot) so a changed index root recomputes instead of serving the stale "borrowed the parent's index" verdict. Adds a regression test that fails pre-fix (the stale warning survives the index-root flip); the existing "no further git spawn" caching test still passes. The query-staleness half (a second project opened through the handler) isn't unit-testable under vitest, so it was validated with a dist A/B. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com> |
||
|
Colby Mchenry
|
68eaf0dbd8 |
feat(mcp): codegraph_explore as the sole primary tool + store coverage + overload disambiguation (#647)
## Summary
Completes the explore-overhaul arc: `codegraph_explore` becomes the single primary tool an agent reaches for, and its coverage + output shape are tuned so flow/architecture questions resolve with near-zero Read/Grep.
### What changed
- **explore is the sole primary tool** — removed `codegraph_context` (the fuzzy-input Read-trigger) and `codegraph_trace` (under-picked by agents); explore already surfaces the call flow among the symbols you name. A plain natural-language question now works as the query.
- **Store/handler coverage** — functions defined inside object literals (Zustand `create((set, get) => ({ … }))`, Redux/Pinia/MobX, exported handler/route maps) are indexed as real symbols, including calls through `useStore.getState().fn()` and destructured `const { fn } = useStore.getState()`. A general AST rule, not a per-lib hack.
- **Overload disambiguation** — explore leads with the *right* definition when a method name is overloaded across types (a PascalCase type token in the query biases to that type's own def); `codegraph_node` returns *every* overload's body in one call, with an optional `file`/`line` selector to pin one.
- **Method-atomic render** — explore never returns half a method; at the size budget it drops whole methods/files (and lists what it dropped) instead of truncating a body mid-method.
- **Native-read-shaped output** — per-call output is capped to ~24K with a 25K hard ceiling and concentrated into ~150–250-line flow windows, mirroring how the agent natively reads; repo size scales the *call* budget, not the per-call size (a larger response just gets externalized to a file the host Reads back).
- **Blast radius** folded into explore (dependents + covering tests, locations only).
### Benchmark (refreshed on this build)
Re-validated the 7-repo A/B on 2026-06-02 (Opus 4.8, effort=high, median of 4). WITH arm re-measured on this build, WITHOUT reused:
**~16% cheaper · 47% fewer tokens · 22% faster · 58% fewer tool calls** — 0 file reads on 6 of 7 repos (Gin ~1).
The arc trades larger, cache-heavy explore responses for guaranteed near-zero reads, so cost/token margins soften vs the prior build (Excalidraw and Tokio land at cost break-even) while time and tool-calls stay clear wins everywhere — consistent with the project's stated optimization target (latency + tool-calls, not token cost).
### Validation
- Full suite green: **1112 passed, 2 skipped**.
- 28/28 plain WITH runs across the 7 README repos completed clean; reads median 0 on 6/7.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
|
||
|
신주안andClaude Opus 4.7
|
4a4a37d135 |
feat(mcp): detect borrowed git worktree index and surface on read tools (#312)
When a worktree is nested inside the main checkout (e.g. agent tools that place worktrees under .claude/worktrees/<name>/), the nearest-.codegraph walk resolves UP to the main checkout's index and queries silently return that tree's code — usually a different branch. Symbols changed only in the worktree are invisible, and nothing tells the user (#155). Two layers: - **Detection** (src/sync/worktree.ts): detectWorktreeIndexMismatch() compares the caller's git working-tree root vs the resolved index root via 'git rev-parse --show-toplevel'. Best-effort; no git / not a repo / monorepo subdir / plain-ancestor index → no warning. - **Surface**: codegraph status (CLI + MCP) embeds a verbose multi-line warning; every MCP read tool (search/context/trace/callers/callees/impact/explore/node/ files) prefixes a compact one-line notice naming the borrowed index and the fix (codegraph init -i in the worktree). Detection is cached per session per start path, so it costs at most a single pair of 'git rev-parse' spawns per project no matter how many tool calls — respects the wall-clock-latency invariant. Real-git tests (no mocking) cover both layers. Validated on macOS / Linux (Docker) / Windows (Parallels VM); 11/11 worktree tests green on all three. Closes #155 Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> |