* perf(mcp): steer agents to answer directly instead of delegating to subagents
CodeGraph beats native grep/read on cost only when the agent queries it
directly. When the agent delegates to file-reading sub-agents, those
sub-agents read files regardless of the index, so CodeGraph becomes net
overhead on top of the reads. The install templates even told agents to
"spawn a subagent for explore-class questions" — the expensive path.
Changes:
- server-instructions + both install templates: add an "Answer directly —
don't delegate exploration" directive; reposition codegraph_explore as the
efficient one-call multi-symbol tool (was: "spawn a subagent for it").
- codegraph_explore: hard-cap output to its adaptive budget (it overran,
~30k vs a 28k cap) and tighten the medium tier (28k->13k).
- codegraph_node: return a member outline for container kinds instead of the
full class body.
Rigorous N>=4-per-arm warm-block benchmark (median total_cost_usd):
excalidraw (~600 files): WITH $0.54 vs native $1.02 (-47%)
vscode (~10k files): WITH $0.41 vs native $0.72 (-42%)
ky (~25 files): WITH $0.46 vs native $0.44 (wash)
Answers were equal-or-better (correct, file:line-cited) with ~6x fewer tool
calls; the directive drove the direct path on 14/14 codegraph runs.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* docs(readme): rebuild benchmark with real-world repos + cost/token/time/tool savings
Replace the "Claude Code (Python+Rust/Java)" rows — which benchmarked the
Claude Code CLI repo, not real codebases in those languages — with real
open-source projects per language: Django (Python), Tokio (Rust), OkHttp
(Java), Gin (Go), plus Alamofire (Swift) and the existing TypeScript repos
(VS Code, Excalidraw).
The table now reports all four savings the change targets — cost, tokens,
time, tool calls — as the median of 4 runs per arm (Claude Opus 4.7,
headless claude -p, with vs empty MCP config). Averages across the 7 repos:
35% cheaper, 59% fewer tokens, 49% faster, 70% fewer tool calls. Adds a
methodology note and raw WITH->WITHOUT medians.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* release: 0.7.7 (multi-agent installer — Cursor, Codex, opencode)
* fix(installer): opencode .jsonc + AGENTS.md (0.7.8)
v0.7.7 wrote ~/.config/opencode/opencode.json, but opencode reads
opencode.jsonc by default — so the codegraph MCP entry never appeared
in any opencode session. Also installs AGENTS.md so opencode's model
reaches for codegraph_* tools instead of native Grep.
- Prefer existing .jsonc, fall back to .json, default new installs
to .jsonc.
- Surgical edits via jsonc-parser preserve user comments and
formatting across install / re-install / uninstall round-trips.
- Install AGENTS.md (global ~/.config/opencode/AGENTS.md, local
./AGENTS.md) with the shared INSTRUCTIONS_TEMPLATE — same
marker-delimited approach Codex uses.
- +9 opencode-specific tests covering filename precedence, comment
preservation, AGENTS.md install + sibling-content preservation,
uninstall reverses both files.
575/575 tests pass. Hand-verified end-to-end: opencode session calls
codegraph_node + codegraph_callers for a structural query, zero Grep
calls.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* docs: overhaul CLAUDE.md and add scripts/release.sh + Cursor rules file
Replaces the old Claude-only CLAUDE.md with a comprehensive guide covering
the full project architecture, multi-agent installer, test conventions,
NodeKind/EdgeKind reference, and release workflow. Key additions:
- Documents the layered pipeline, all module paths, and the multi-target
installer (targets/, registry.ts, AgentTarget interface).
- Adds the Cursor `--path` quirk and the "update all three surfaces" rule
when changing MCP tool guidance.
- Documents `npm run eval`, `test:eval`, and the full set of build/test
commands including single-file patterns.
- `scripts/release.sh` — idempotent bash script that tags the current
commit, pushes the tag, and creates a GitHub Release whose notes are
extracted from the matching `## [X.Y.Z]` block in CHANGELOG.md. Safe
to re-run after partial failure.
- `.cursor/rules/codegraph.mdc` — Cursor-specific agent instructions
(tool decision table, rules of thumb, index-lag warning) written by
the installer and kept in sync with server-instructions.ts and
instructions-template.ts.
---------
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
