Add cargo-publish job that runs in parallel with binary builds.
The release job now depends on [build-macos, build-linux, cargo-publish]
so the crate is always published to crates.io before the GitHub release
and binaries are attached.
Requires CARGO_REGISTRY_TOKEN secret in repo settings.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Add ini_safe() helper that strips CR/LF from user-supplied string params
before they are written into terminal.ini and backtest_config.ini.
Applies to: expert path, symbol, timeframe, set_file — any value that
could carry an embedded newline and inject extra INI directives.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
database.rs:
- search_by_tags: replace string-interpolated LIKE with parameterized ?N bindings
- search_by_notes: replace format!() SQL with prepared statement + ?1/?2 params
- search_by_set_file: same — use ?1 for pattern and ?2 for limit
backtest.rs (extract_and_store / monitor_backtest_completion):
- extract_and_store now returns bool — callers only delete report file if true
- guard both success paths with conditional delete + warning on failure
- replace .parent().unwrap() panic with explicit match + error log + graceful return
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- mcp_server.rs: #[allow(dead_code)] on NotificationCallback type alias and get_notification_sender method
- handlers/mod.rs: #[allow(dead_code)] on past_complete_month helper
- Cargo.toml: move maintenance key into [package.metadata] (was invalid at package level)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Deals-in-DB architecture: per-deal data now stored in SQLite instead of
deals.csv/deals.json files.
- storage: add `deals` table with report_id FK; insert_deals/get_deals/get_by_report_dir methods
- extract: stop writing deals.csv and deals.json; only metrics.json written to disk
- pipeline: call db.insert_deals() after extraction; HTML report still deleted after parse
- analytics: all 19 tools now load deals from DB (report_id > report_dir > latest)
- analytics: resolve_report() helper replaces load_report_data/read_deals_from_csv
- analytics: calling any analytics tool with no args uses the latest report automatically
- tools: add export_deals_csv for on-demand CSV generation from DB
- tools: add get_by_report_dir for backward-compat lookup by filesystem path
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Added list_symbols to Core workflow
- Added list_jobs to Optimization section
- Added promote_to_baseline to Reports section
- Added read_set_file and write_set_file to .set files section
- Added diff_set_files to .set analysis section
- Added check_update and update to Debugging section
- Added get_backtest_history to History & Comparison section
Brings documented tools from 87 to 89 (matching actual count)
- Correct path: /mcp-server/bin/mt5-quant (not project root)
- Verified by downloading and extracting actual release tarball
- Updated both Full Example and Environment Variables sections
- Changed /target/release/mt5-quant to /mt5-quant for prebuilt binary path
- Added path notes explaining prebuilt vs dev build locations
- Fixed both Full Example and Environment Variables sections
Others can run optimization via terminal but lack:
- Background execution with polling
- Results parsing into structured data
- Automatic .set file generation
- Optimization history and search
- Update 'Others' column to acknowledge Windows MT5 Python package
- Clarify MQL5 compilation: others via GUI/terminal, we do headless
- Add 'Report organization' as unique MT5-Quant feature
- Emphasize focus on backtest analytics and workflow management
On the first tool call of each session, a background task fires once
and fetches the latest release tag from the GitHub API (5 s timeout).
The result is cached in a static OnceLock for the lifetime of the
process — no repeated network calls.
check_update: returns cached result instantly, or fetches on demand
if the background task hasn't completed yet.
update: downloads the latest tarball for the current platform, extracts
the binary, and atomically replaces the running executable via a
temp-file rename. Requires MCP reconnect to activate new version.
Platform support: macos-aarch64, macos-x86_64, linux-x86_64.
Unsupported platforms return a build-from-source hint.
Tool count: 87 → 89. README and MCP_TOOLS.md updated.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
release.sh:
- Add --yes/-y flag (and CI=true) for non-interactive execution
- Auto-resolve server.json conflict during rebase-before-push so CI's
SHA256 commit no longer blocks the push
- Fix major version bump (was double-assigned, now single expression)
build-release.sh:
- Remove cp of /CLAUDE.md (gitignored personal file — caused build failure)
- Bundle all IDE setup docs (QUICKSTART, CLAUDE, CURSOR, VSCODE, WINDSURF)
- Use set -euo pipefail; chmod +x binary; clean up dist/ subdir after tar
build-rust.sh:
- Replace Windsurf-specific install hint with generic cargo install --path . --force
- Use set -euo pipefail
setup.sh:
- Replace "Antigravity" with "Claude Desktop" in comments and output
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
v1.6.0 dropped the version number from asset filenames
(mcp-publisher_linux_amd64.tar.gz, not mcp-publisher_1.6.0_linux_amd64.tar.gz).
Look up the linux/amd64 .tar.gz URL directly from the GitHub releases API
instead of constructing it, so future renames don't break the workflow.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Devid