feat: Merge post-v1.0.0 changes from sandbox (v1.1.0+)#67
Merged
ivyleavedtoadflax merged 75 commits intomainfrom Jan 18, 2026
Merged
feat: Merge post-v1.0.0 changes from sandbox (v1.1.0+)#67ivyleavedtoadflax merged 75 commits intomainfrom
ivyleavedtoadflax merged 75 commits intomainfrom
Conversation
- Add pydantic-settings dependency for config validation - Create RemoteConfig Pydantic model with field validators: - instance_name: alphanumeric, hyphens, underscores, dots - ssh_user: alphanumeric, hyphens, underscores - ssh_key_path: expands ~ to home directory - aws_region: validates AWS region format (e.g., us-east-1) - Support environment variable overrides (REMOTE_ prefix) - Create ConfigValidationResult for structured validation output - Update ConfigManager to use Pydantic validation internally - Enhance 'remote config validate' command with Pydantic validation - Add 25 new tests for Pydantic config features - Fix specs/readme.md: mark issue-23 as COMPLETED, update mypy path Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit 1f56cba)
Add --watch/-w and --interval/-i flags to the status command to enable continuous monitoring of instance status using Rich's Live display. This resolves garbled output issues when using the external `watch` command with Rich formatting. - Add _build_status_table helper function for reusable table generation - Add _watch_status function using Rich Live display with screen mode - Validate interval must be at least 1 second - Handle Ctrl+C gracefully to exit watch mode - Add comprehensive tests for watch mode functionality Closes #35 Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit d2ccacc)
- Remove unused `cfg: ConfigParser | None = None` parameter from get_instance_name() - Remove corresponding docstring documentation for the parameter - Remove now-unused `from configparser import ConfigParser` import The parameter was marked as "for backward compatibility" but was never used in the function body (it always used config_manager instead), and none of the 8 call sites passed any arguments. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit 24a886a)
The connect() function was using "ssh_key" as the config key when retrieving the SSH key path, but the valid config key is "ssh_key_path". This caused the SSH key configuration to fail silently. - Fix get_value() call to use "ssh_key_path" instead of "ssh_key" - Update help text to reference the correct config key name Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit d9b35f5)
Panel stretches beyond console width and has redundant messaging. (cherry picked from commit 9f3116b)
The builtins module was imported only to use builtins.list for a type annotation. In Python 3.9+, list can be used directly in type annotations without the builtins prefix. This change removes the unnecessary import and simplifies the type annotation. Co-authored-by: Claude <noreply@anthropic.com> (cherry picked from commit 1fc7119)
The utils.py module defined `app = typer.Typer()` but this app instance was never used - no commands were registered to it and no other modules imported it. The typer import itself is still needed for other uses (typer.Exit, typer.secho, typer.colors). Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit f57850d)
The get_sts_client() function was defined as a cached client factory but
was never used. The get_account_id() function created a new STS client
directly with boto3.client("sts") instead.
This change makes the code consistent with the EC2 client pattern where
get_ec2_client() is used throughout the codebase, and utilizes the
caching provided by @lru_cache.
Also updates tests to mock get_sts_client instead of boto3.client.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
(cherry picked from commit 8daac71)
- Issue 37: Fallback to us-east-1 for pricing API in other regions - Issue 38: Add command to show instance cost based on uptime (cherry picked from commit 55af72a)
- Remove width=200 from Console initialization in config.py to use terminal's actual width instead of stretching beyond it - Simplify validation output to show single status message instead of redundant "All checks passed" + "Status: Valid" - Update test assertion to match new output message Closes #36 Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit acd655a)
Add spec for feature to schedule instance stops after a duration, e.g., `remote instance stop --in 3h` (cherry picked from commit 01adf6b)
Use SSH to run `shutdown -h +N` on the instance instead of a local subprocess. This is simpler and survives local disconnects. (cherry picked from commit f474c02)
When a user's region is not in the region-to-location mapping, pricing lookups now fall back to us-east-1 instead of returning None. This provides users in less common regions with an estimated price rather than no price at all. - Add get_instance_price_with_fallback() function that returns (price, fallback_used) - Update get_instance_pricing_info() to include fallback_used indicator - Update instance list command to use fallback pricing - Add comprehensive tests for fallback behavior Closes #37 Co-authored-by: Claude <noreply@anthropic.com> (cherry picked from commit 875cc9d)
Add --in option to stop command to schedule shutdown after duration Add --stop-in option to start command to auto-stop after duration Add --cancel flag to cancel scheduled shutdowns Features: - Parse duration strings (e.g., 3h, 30m, 1h30m) - SSH to instance and run 'shutdown -h +N' - Uses ssh_user and ssh_key_path from config - Handles SSH errors gracefully Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit 32d8144)
Standardize all command output styles around `config show` format. (cherry picked from commit a8bb264)
Update ECS list commands to use Rich Tables instead of plain line-by-line output, matching the consistent style used across other modules. Changes: - list-clusters: Now displays clusters in a Rich Table with name and ARN - list-services: Now displays services in a Rich Table with name and ARN - prompt_for_cluster_name: Changed typer.echo to typer.secho for consistency - prompt_for_services_name: Changed typer.echo to typer.secho for consistency Closes #40 Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit 8adb22f)
#25) Adds a new `remote instance cost` command that displays the estimated cost of a running instance based on its uptime and hourly pricing rate. Features: - Shows instance ID, type, status, launch time, and uptime - Calculates estimated cost using hourly rate × uptime hours - Uses fallback to us-east-1 pricing for unsupported regions - Handles non-running instances and unavailable pricing gracefully Closes #38 Co-authored-by: Claude <noreply@anthropic.com> (cherry picked from commit ab6e4f9)
Issue 41: Fix instance cost (not displaying, panel too wide, integrate into ls) Issue 42: Evaluate overlap between instance ls and status commands (cherry picked from commit 7c6af04)
Emphasize need for Typer CLI tests to catch cost display issues. (cherry picked from commit 88852c7)
- Add --cost/-c flag to instance ls to show cost columns - Add Uptime, $/hr, and Est. Cost columns when --cost is used - Remove separate instance cost command (functionality now in ls) - Remove _get_instance_details helper (no longer needed) - Add _get_raw_launch_times helper for extracting launch times - Add comprehensive tests for cost flag functionality - Update plan.md and spec status to COMPLETED The cost information is now opt-in rather than shown by default, which improves performance since pricing API calls are skipped unless explicitly requested. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit 3980be0)
Clarify the distinct purposes of instance ls (summary view of all instances) vs instance status (detailed view of one instance). Changes: - Enhanced instance status to show comprehensive details including network config, security groups, key pair, tags, and health status - Updated help text for both commands to clearly indicate when to use each - instance status now uses a Rich Panel instead of Table for better formatting - Health status section only shown for running instances Co-authored-by: Claude <noreply@anthropic.com> (cherry picked from commit d3093e7)
The loop index was unused, making enumerate() unnecessary. Simplified to a plain for loop over reservations. Co-authored-by: Claude <noreply@anthropic.com> (cherry picked from commit b37effb)
The drop_nameless parameter was defined but never used - the function always skips instances without a Name tag regardless of its value. This was misleading since the default value of False implied nameless instances would be included. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit 0d14e97)
The module-level `ec2_client` attribute was deprecated and scheduled for removal in v0.5.0. After scanning the codebase, no code uses this deprecated attribute - all modules use `get_ec2_client()` directly. Remove: - Deprecation comment block - `__getattr__` function providing lazy access to `ec2_client` Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit c8885ac)
The __getattr__ function providing lazy access to ecs_client as a module-level attribute was dead code. All ECS functions use get_ecs_client() directly, and no imports of ecs_client exist in the codebase. Also removes the unused Any type import. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit c361dcb)
The ENV_PREFIX constant was defined but never used. The actual env prefix is hardcoded in RemoteConfig.model_config as env_prefix="REMOTE_". Co-authored-by: Claude <noreply@anthropic.com> (cherry picked from commit aae94bd)
The stop() function used parameter name 'in_duration' while the start() function used 'stop_in' for the same purpose (scheduling automatic shutdown). This inconsistency created cognitive overhead. Renamed the parameter to 'stop_in' to match the pattern used in start(). The CLI flag '--in' remains unchanged for backwards compatibility. Co-authored-by: Claude <noreply@anthropic.com> (cherry picked from commit 5e34587)
Recurring issue: Panels expand to full terminal width. Audit all Panel usage and set expand=False globally. (cherry picked from commit 209b796)
…-in (#34) Rename the `type()` command function to `instance_type()` with `@app.command("type")` decorator to preserve CLI command name. Rename the `type` parameter to `new_type` to avoid shadowing Python's built-in type() function. Co-authored-by: Claude <noreply@anthropic.com> (cherry picked from commit 89b9939)
The instance_type() function called get_instance_type() twice to fetch the same value - once at function start (stored in current_type) and again in the else branch (stored in current_instance_type). This made an unnecessary AWS API call since current_type was already available. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit 470c984)
…lates() (#48) The exception handler `except (ResourceNotFoundError, Exception)` was catching all exceptions silently, which could mask bugs. Changed to `except (ResourceNotFoundError, AWSServiceError)` to match the documented exceptions that get_launch_template_versions() can raise. Co-authored-by: Claude <noreply@anthropic.com> (cherry picked from commit c3074d4)
Replace `except Exception` with `except ValueError` in three locations: - ConfigValidationResult.validate_config() - ConfigManager.get_instance_name() - ConfigManager.get_value() Pydantic's ValidationError inherits from ValueError, so this catches validation errors while avoiding the antipattern of catching all exceptions indiscriminately. Updated test to use ValueError instead of generic Exception. Co-authored-by: Claude <noreply@anthropic.com> (cherry picked from commit c32ab4f)
Changed exception handling in ami.py's list_launch_templates() to display a warning message instead of silently ignoring errors when fetching launch template version details. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit bc7a4b5)
…ion (#52) The launch() function was duplicated nearly identically (~130 lines) in both ami.py and instance.py. This refactor: - Adds launch_instance_from_template() to utils.py with all common logic - Simplifies both launch() commands to thin wrappers calling shared function - Updates test mocks to patch correct modules (remote.utils, remote.config) Net change: -58 lines of code, single source of truth for launch logic. Co-authored-by: Claude <noreply@anthropic.com> (cherry picked from commit 1cff43a)
Two typer.Exit() calls lacked explicit exit codes: - prompt_for_cluster_name(): when no clusters found - prompt_for_services_name(): when no services found While typer.Exit() defaults to exit code 0, being explicit about exit codes is best practice and consistent with other exit calls in the codebase. Exit code 0 is correct here as these are informational cases, not errors. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit 4b564d6)
Fix inconsistent docstring formatting to match the style used in utils.py and other modules: - Move descriptions to same line as opening triple quotes - Add proper 4-space indentation to Args/Returns sections - Remove redundant type annotations from docstrings (types are in function signatures) - Remove type prefixes from Returns sections Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit 27cd751)
The status() command in instance.py used the Annotated[] style for parameter type hints while all other commands used the simpler inline style. This inconsistency made the codebase harder to read and created confusion about which style to follow. Changes: - Convert status() parameters from Annotated[] to inline style - Remove unused Annotated import from typing Co-authored-by: Claude <noreply@anthropic.com> (cherry picked from commit 15b75b6)
The function was dead code - only exercised by tests, never used in application code. The instance list command uses get_instance_price_with_fallback() directly rather than this higher-level wrapper. - Remove get_instance_pricing_info() from remote/pricing.py - Remove TestGetInstancePricingInfo from tests/test_pricing.py - Update specs to reflect actual implementation Co-authored-by: Claude <noreply@anthropic.com> (cherry picked from commit a50adcf)
Renamed `cfg` to `config` throughout the file for consistency. The codebase mixed both names for ConfigParser objects - now standardized on `config` which is more descriptive and matches the pattern used in ConfigManager class methods. Co-authored-by: Claude <noreply@anthropic.com> (cherry picked from commit 252f41d)
Remove dead code from five modules that contained unused main blocks: - remote/ami.py - remote/config.py - remote/instance.py - remote/snapshot.py - remote/volume.py These modules are library code imported into __main__.py, not executed directly. The blocks were never executed since users run `remote <cmd>` not `python -m remote.instance` etc. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit a618fe7)
The write_config() function returned its input ConfigParser object, but no caller ever used the return value. Changed return type to None and updated the corresponding test. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit 9800551)
The function was defined in utils.py but never called by any production code. It was only exercised by tests. Removes dead code and associated tests. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit b06bf72)
The validate_snapshot_id() function was defined in validation.py but never called anywhere in the application. Remove this dead code along with its associated tests to keep the codebase clean. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit fe92d81)
Move the "# If the instance is managed by Terraform, warn user" comment from line 963 to line 981, placing it directly above the terraform_managed assignment it describes. The comment was previously orphaned from its relevant code by 20 lines of unrelated confirmation prompts. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit 8a7eb0d)
The unset_value() CLI command was directly manipulating the config file instead of using the ConfigManager.remove_value() method like other similar commands (set_value, add). This violated encapsulation and missed proper state management (cache invalidation). Also fixed ConfigManager.remove_value() to read from the specified config_path parameter instead of the default path. Co-authored-by: Claude <noreply@anthropic.com> (cherry picked from commit 8c0f8ca)
Replace verbose if-else blocks with ternary operators in RemoteConfig.from_ini_file() and ConfigValidationResult.validate_config() for cleaner, more idiomatic code (SIM108). Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit 190b0f5)
…_MONTH constant (#65) The function and constant were defined but never used in production code: - No code called get_monthly_estimate() - HOURS_PER_MONTH was only used by get_monthly_estimate() - Similar to previous removal of get_instance_pricing_info() The application displays hourly prices directly without converting to monthly estimates. Co-authored-by: Claude <noreply@anthropic.com> (cherry picked from commit b55caaf)
…nfigManager (#66) The get_instance_name() and get_value() methods had identical exception handling blocks that displayed warnings for various config-related errors. This duplicated ~12 lines of code. Changes: - Add _handle_config_error() helper method to centralize error handling - Update both methods to use a single except clause delegating to helper - Use union type syntax (X | Y) in isinstance checks per UP038 lint rule Co-authored-by: Claude <noreply@anthropic.com> (cherry picked from commit dc66c51)
The get_instance_ids() function had inconsistent filtering behavior compared to get_instance_info(): - get_instance_info() iterates through ALL instances and filters out those without a Name tag - get_instance_ids() only took the FIRST instance from each reservation and did NOT filter by Name tag This could cause array length mismatches when used together. The code worked around this with strict=False in zip() calls, silently truncating to the shortest array. Changes: - Update get_instance_ids() to iterate all instances and filter by Name tag - Change strict=False to strict=True in zip calls in instance.py and config.py - Add test for the new filtering behavior Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit 16147a7)
…#68) The get_value() CLI command bypassed ConfigManager.get_value() and read directly from config file. This was inconsistent with other CLI commands (set_value, add, unset_value) which properly use ConfigManager. Changes: - Renamed function to get_value_cmd to avoid shadowing - Added key validation against VALID_KEYS (consistent with set_value) - For default path: delegate to config_manager.get_value() - For custom paths: continue reading directly from file Benefits: - Pydantic validation for config values - REMOTE_* environment variable override support - Key validation (rejects unknown keys) - Consistent API with other config commands Co-authored-by: Claude <noreply@anthropic.com> (cherry picked from commit 3bec8ed)
Replace magic numbers with named constants for better readability: - SECONDS_PER_MINUTE, SECONDS_PER_HOUR, MINUTES_PER_DAY - MAX_STARTUP_WAIT_SECONDS, STARTUP_POLL_INTERVAL_SECONDS - CONNECTION_RETRY_SLEEP_SECONDS, MAX_CONNECTION_ATTEMPTS Co-authored-by: Claude <noreply@anthropic.com> (cherry picked from commit f4640f9)
The _format_uptime() function used SECONDS_PER_MINUTE (60) to perform arithmetic on values measured in minutes. While mathematically correct, this was semantically misleading. Added MINUTES_PER_HOUR constant and updated MINUTES_PER_DAY to use it for consistency. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit 9144511)
Add SSH_READINESS_WAIT_SECONDS constant (10s) for the hardcoded sleep times used when waiting for SSH to become ready after instance startup. This follows the established pattern of extracting time-related magic numbers to named constants. Co-authored-by: Claude <noreply@anthropic.com> (cherry picked from commit ef46b1a)
Add TYPE_CHANGE_MAX_POLL_ATTEMPTS and TYPE_CHANGE_POLL_INTERVAL_SECONDS constants to replace hardcoded `5` values in instance_type() function. This follows the established pattern for time-related constants. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit 468117d)
(cherry picked from commit 8c0b1bf)
When connecting to a stopped instance: - --start: Automatically starts the instance without prompting - --no-start: Fails immediately with a helpful message - Non-interactive mode (no TTY): Fails with helpful message Also: - Refactored start command to extract _start_instance() internal function - Added comprehensive tests for new behavior - Fixed typo: "trying to starting" -> "trying to start" Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> (cherry picked from commit 48f9a1d)
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
This PR merges all changes from the remote.py-sandbox repository made after v1.0.0, including the v1.1.0 release and subsequent improvements. These 75 commits bring:
New Features
instance statuscommandssh_key_pathfrom configinstance shutdowncommand)instance costcommand and--costflag oninstance lslsvsstatuscommands for instances--start/--no-startflags to auto-start stopped instancesBug Fixes
ssh_key_pathin connect commandRefactoring
Testing
Other
Test plan
uv run pytestuv run ruff check .