feat: centralized logging with --verbose/--quiet/--log-file/--output json CLI flags by tangym · Pull Request #22 · responsibleai/ASSERT

tangym · 2026-05-06T19:48:53Z

Summary

Replaces four competing output mechanisms (unconfigured logging.*, _progress() → sys.__stderr__, click.echo(err=True), print(file=sys.stderr)) with a single centralized logging system.

Before

After

Verbose

JSON

Changes

Phase 1: Logging bootstrap

New p2m/logging_config.py with configure_logging() — RichHandler on sys.__stderr__ (preserves OTel/Phoenix bypass), optional FileHandler
CLI flags: --verbose (DEBUG), --quiet (WARNING), --log-file <path>, --output json
LiteLLM/httpx/openai loggers pinned to WARNING at all verbosity levels

Phase 2: Unify stdlib loggers

Replace bare logging.warning()/logging.info() with named log = logging.getLogger(__name__) in transcript, rollout, judge (8 sites)

Phase 3: Migrate `_progress()` → `log.info()`

Remove the custom _progress() function and replace 32 call sites
Errors upgraded to log.error() with exc_info=True

Phase 4: Migrate raw `print(file=stderr)`

runner.py config errors → log.error()
config.py validation warnings → log.warning()

Phase 5: Add logging to silent modules

log.debug() in policy, design, seeds, systematization, systematization_convert, model_client
log.info() sub-step progress in policy stage (two sequential LLM calls)

Phase 6: Structured JSON output

--output json flag for CI pipeline consumption
_JsonFormatter emits one JSON object per log record with timestamp, level, logger, message, and optional exception

Additional improvements

All messages use consistent [stage_name] prefix format with ✓/✗ symbols
Route rollout per-seed progress lines through logging (previously raw sys.__stderr__.write())
Suppress OpenAI SDK retry spam ("Retrying request to /chat/completions")
Add log_heartbeat() for long-running LLM calls (15s periodic "still working" messages)
Per-call debug summary with latency, token usage, and finish_reason
Fix Responses API usage/finish_reason extraction (input_tokens/output_tokens → prompt_tokens/completion_tokens)
--verbose/--quiet/--log-file/--output accepted on both p2m group and run subcommand

Test results

544 passed, 13 skipped (1 pre-existing Docker permission failure)

…e CLI flags Create p2m/logging_config.py with configure_logging() that sets up: - RichHandler on sys.__stderr__ (bypasses OTel/Phoenix stderr wrappers) - Optional FileHandler for --log-file - Level mapping: default=INFO, --verbose=DEBUG, --quiet=WARNING - LiteLLM/httpx loggers pinned to WARNING regardless of verbosity Wire --verbose/-v, --quiet/-q, --log-file into the top-level CLI group so all subcommands inherit the logging configuration.

Replace root-level logging.warning()/logging.info() with module-level log = logging.getLogger(__name__) in: - p2m/core/transcript.py (2 sites) - p2m/stages/rollout.py (4 sites) - p2m/stages/judge.py (2 sites) Named loggers enable per-module filtering and show the source module in log output.

Remove the custom _progress() function (sys.__stderr__ workaround) and replace all 32 call sites with log.info(). The RichHandler from logging_config.py already writes to sys.__stderr__, preserving the same OTel/Phoenix bypass. Upgrade error messages to log.error() and route tracebacks through the logging system (exc_info=True) instead of traceback.print_exc(). Remove unused traceback import.

runner.py: 2 print() calls → log.error() for config errors config.py: 2 print() calls → log.warning() for validation warnings Remove unused sys import from config.py.

New test_logging_config.py covering: - Default/verbose/quiet level mapping - Console handler targets sys.__stderr__ - File handler creation and parent dir creation - LiteLLM loggers pinned to WARNING - Repeated calls don't duplicate handlers Add CLI flag acceptance tests to test_cli.py for -v/--verbose, -q/--quiet, and --log-file.

Tests previously patched sys.stderr/sys.__stderr__ to capture output. Now that runner output goes through the logging system, switch to assertLogs('p2m.runner') which captures log records directly. Remove unused io import.

… retry noise Replace raw sys.__stderr__.write() for per-seed rollout progress lines with log.info() (success) / log.warning() (error). These lines previously appeared without any level prefix (INFO/WARNING/ERROR). Suppress openai and httpcore loggers to WARNING — the openai SDK logs every retry attempt at INFO ('Retrying request to /chat/completions in N seconds'), which floods output during rate limiting. p2m's own model_client already handles retry reporting. Remove unused sys import from rollout.py.

Convert %s/%d placeholder-style log calls to f-strings for readability. The performance difference is negligible at this call volume (~30 calls per pipeline run).

Add log = logging.getLogger(__name__) and debug-level log lines to: - stages/policy.py — model, behavior_count, web_search - stages/design.py — factor_sizes after generation - stages/seeds.py — kinds enabled, sample sizes, tool_source - stages/systematization.py — concept, model, mode, web_search - stages/systematization_convert.py — model, behavior_count_hint - core/model_client.py — model, api_mode, schema_name per LLM call All new log lines use log.debug() so they only appear with --verbose. The runner already handles INFO-level stage start/done banners.

The policy stage makes two sequential LLM calls (systematization then convert) that can each take 30-90s. Add INFO-level progress between sub-steps so the output doesn't appear frozen: [1/2] Researching risk taxonomy... [1/2] Risk taxonomy complete [2/2] Converting to structured policy...

Strip leading ' ' from all runner log messages — these were from the _progress() era when messages were written raw to stderr and needed manual indentation. RichHandler handles column alignment, so the extra spaces caused misalignment with stage-internal logs like rollout per-seed progress.

Add log_heartbeat() async context manager to async_utils.py that logs a 'still working, Ns elapsed' message every 15 seconds while an LLM call is in progress. Prevents the CLI from appearing frozen during the policy stage's two sequential calls (systematization + convert), which can each take 30-90 seconds.

# Conflicts: # p2m/core/model_client.py

…mand Users naturally type 'p2m run --verbose' but Click only accepts group options before the subcommand. Fix by adding the same flags to the run subcommand and re-configuring logging if they're passed there. Both forms now work: p2m --verbose run --config ... p2m run --verbose --config ...

Replace pre-call debug logs with a compact post-call summary: generate: model=azure/gpt-5.4, 1.2s, 850+2400 tokens, finish=stop Each generate/generate_structured/generate_with_tools call now logs model name, wall-clock latency (including retries), prompt+completion token counts, and finish_reason. All at DEBUG level.

The Responses API uses input_tokens/output_tokens (not prompt_tokens/completion_tokens) and status (not finish_reason). _normalize_usage now falls back to the Responses API field names, and finish_reason falls back to the status field. Previously showed '? tokens, finish=?' for all Responses API calls.

Apply consistent formatting to all user-facing log messages: - Add [stage_name] prefix to all stage messages - Use ✓ for success and ✗ for failure consistently - Capitalize first word after prefix - Pipeline-level messages use 'Pipeline' (capitalized, no prefix) - Standardize skip/error/done patterns Before: INFO Generating behavior taxonomy... INFO ✓ Generated 5 behaviors (41.2s) INFO rollout [4/5] ✔ [prompt] Multi-turn... INFO pipeline completed (162.3s) After: INFO [policy] Generating behavior taxonomy... INFO [policy] ✓ Generated 5 behaviors (41.2s) INFO [rollout] [4/5] ✓ [prompt] Multi-turn... INFO Pipeline completed (162.3s)

Add JSON log format for CI pipeline consumption via --output json. Each log record emits one JSON object with timestamp, level, logger, message, and optional exception fields. - New _JsonFormatter class in logging_config.py - --output flag (text|json) on the top-level CLI group - Tests for JSON format validity and exception inclusion

Apply consistent formatting to all user-facing log messages: - Add [stage_name] prefix to all stage messages - Use ✓ for success and ✗ for failure consistently - Capitalize first word after prefix - Pipeline-level messages use 'Pipeline' (capitalized, no prefix) - Standardize skip/error/done patterns Before: INFO Generating behavior taxonomy... INFO ✓ Generated 5 behaviors (41.2s) INFO rollout [4/5] ✔ [prompt] Multi-turn... INFO pipeline completed (162.3s) After: INFO [policy] Generating behavior taxonomy... INFO [policy] ✓ Generated 5 behaviors (41.2s) INFO [rollout] [4/5] ✓ [prompt] Multi-turn... INFO Pipeline completed (162.3s)

Resolves conflicts in p2m/runner.py and p2m/stages/rollout.py introduced by main's centralized-logging refactor (PR #22) and policy/rollout error-handling improvements (PR #24). p2m/runner.py: kept the artifact-cache stage path (prepare_artifact_plan / activate_artifact_plan / override_cacheable_output_paths) and combined it with main's logging style. The two stage-skip messages now use log.info with the new '[stage] Skipped' prefix instead of the deleted _progress() helper, and _progress() itself was removed since main's logging configuration writes to sys.__stderr__ via RichHandler, neutralizing the original Phoenix sys.stderr-wrapping concern. p2m/stages/rollout.py: dropped the sys.__stderr__ rollout progress writer in favor of main's log.info / log.warning calls (same rationale). Kept the 're' import that the cache code uses for _VERSIONED_ARTIFACT_RE; dropped the now-unused 'sys' import. Verified: pytest passes for tests/test_artifact_cache.py, tests/test_runner_artifact_cache.py, tests/test_runner_progress.py, and tests/test_viewer_server_artifacts.py (56 passed, 2 skipped). Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

…user save_dir override Addresses Jake's review on the artifact-cache PR. 1. Disk-leak on stage failure (Jake's #2): when a cacheable suite stage failed after prepare_artifact_plan allocated vNNNN/ but before finalize_artifact_plan wrote the sidecar, the partial directory stayed on disk forever. _next_version kept incrementing past abandoned slots and the stage_root accumulated dead version directories on every failed run. Adds discard_artifact_plan(ctx, plan) to remove the version directory and pop the orphaned ctx['artifact_versions'] entry. Wired into runner.py's stage exception path. No-ops for reused plans so a downstream failure cannot blow away a healthy upstream cache hit. latest.json is left untouched (finalize is its only writer for non-reused plans, so a discarded plan never touched it). 2. Silent override of user save_dir (Jake's non-blocking #2): override_cacheable_output_paths now log.warnings whenever it replaces a user-supplied save_dir/save_path with the versioned cache location. Customers who set save_dir in YAML get a clear actionable message instead of seeing their value silently ignored. No warning when no user value was set. 3. Concept-hash transitive flow (Jake's non-blocking #1): added a comment in _stage_descriptor explaining that concept_hash is computed only for policy and propagates through the dependency chain to design and seeds via _dependency_descriptor. Notes the safety invariant (every cacheable stage must depend on its upstream) so a future stage that breaks the chain triggers a code-review flag rather than silent stale-cache reads after a concept edit. Note: Copilot findings on _metadata_outputs_exist / _metadata_output_paths primary-key KeyError and basename validation (Jake's #1) were already addressed in de9a32a (round 4) — _metadata_output_paths now overlays metadata onto the canonical _output_paths default keyset and _metadata_outputs_exist verifies every expected file exists on disk via the merged path map. _is_safe_artifact_basename rejects unsafe filenames before they become Path components. Tests: +7 regression tests (4 for discard helper covering missing dir, reused plan, ctx cleanup, version-slot reuse; 2 for override warning behavior; 1 runner integration test that fails design mid-stage and asserts vNNNN is cleaned up plus next run reuses the freed slot). Verified: 585 passed, 14 skipped (2 pre-existing Windows logging tempdir flakes from main's PR #22 deselected). npm --prefix viewer run check: 0 errors / 0 warnings. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

…cache Align artifact_cache.py and viewer_read_model.py with the centralized logging convention from PR #22. Replace 10 print(file=sys.stderr) calls with log.warning() using structured placeholders so they respect --quiet, --log-file, and --output json. Remove unused sys imports from both modules.

…pecific error types (#34) * fix: add try-except for file I/O and YAML/JSON parse errors Wrap unguarded file reads and parse calls with specific exception handlers that include the file path in the error message: - config.py: load_config catches FileNotFoundError, PermissionError, yaml.YAMLError and re-raises as ConfigError - otel.py: _parse_otlp_json catches FileNotFoundError, JSONDecodeError - tools.py: load_toolset_file catches FileNotFoundError, yaml.YAMLError - systematization_convert.py: run_systematization_to_policy catches FileNotFoundError, JSONDecodeError * fix: add error handling for callable and HTTP session imports - CallableSession.open: catch ModuleNotFoundError, AttributeError with actionable messages including the callable ref string - OTelTracedSession.open: same pattern for OTel-traced callable imports - HTTPEndpointSession.run_turn: catch aiohttp.ClientResponseError and ClientError with endpoint URL and HTTP status in the message * fix: improve judge and policy stage error handling - judge.py: catch JSONDecodeError when loading policy file, add LLM error types and json/ValueError before the catch-all in the worker, log traceback via log.debug for all caught worker exceptions - policy.py: include model name and raw response text in the error message when policy generation returns non-JSON output * fix: improve rollout and systematization error handling - rollout.py worker: add LLM error types re-raise before catch-all, add ValueError/KeyError handler, log.debug with traceback and seed_id for all caught worker exceptions - rollout.py auditor: log.debug with traceback on each retry failure - systematization.py: catch JSONDecodeError on json.loads fallback with raw text snippet in the error message * fix: add specific error handling for Phoenix collector Catch ConnectionError separately from generic Exception in PhoenixCollector.get_spans with a targeted message about Phoenix connectivity. Validate 'context.trace_id' column existence before filtering. * test: add 21 test cases for exception handling paths Cover all new error handling across config, otel, tools, session, otel_session, judge, systematization_convert, rollout, and collector. Verify exception types, error messages include context (file paths, module names, endpoint URLs), and debug logging fires with tracebacks in worker failures. * fix: skip Phoenix collector tests when pandas is not installed CI environment doesn't include pandas (optional dependency). The PhoenixCollector.get_spans method does 'import pandas' at the top, which fails before our exception handling runs. * style: replace print(file=sys.stderr) with log.warning() in artifact cache Align artifact_cache.py and viewer_read_model.py with the centralized logging convention from PR #22. Replace 10 print(file=sys.stderr) calls with log.warning() using structured placeholders so they respect --quiet, --log-file, and --output json. Remove unused sys imports from both modules. * fix: address review feedback — hoist aiohttp import, use high port in test - Move 'import aiohttp' from run_turn to open(), store as self._aiohttp to avoid re-importing on every turn call - Use port 59123 instead of port 1 in HTTPEndpointSession test for better portability across CI/container environments

Lifts the four product-code changes from PR #44 into our branch and removes the corresponding benchmark monkey-patches that were doing the same job out-of-tree. Changes: * CallableSession.run_turn now reclassifies user-callable provider errors via `_classify_llm_error` so litellm BadRequestError / ContentPolicyViolationError emerge as typed `LLMInputError` (with `__cause__` preserved) instead of bypassing the typed-exception layer the rollout/judge stages key off. Unclassified exceptions propagate untouched. (PR #44 commit 0184d8d) * `_run_prompt_seed` records a `[TARGET INPUT REFUSED: ...]` event in the transcript and sets `stop_reason='target_input_refused'` when the target's run_turn raises `LLMInputError`. Other classified errors still propagate. (PR #44 commit 82cf339) * `_run_auditor_target_loop` splits the combined classified-error catch into a typed `LLMInputError` branch that records `[TARGET INPUT REFUSED: ...]` / `[AUDITOR INPUT REFUSED: ...]` events and sets `stop_reason='target_input_refused'` / `'auditor_input_refused'` respectively. Auth / rate-limit / 5xx errors still propagate. (PR #44 commits 82cf339 + f265154) * judge `_worker` splits the combined catch so `LLMInputError` becomes a synthesized `score_row` with `judge_status='filter_skipped'` and `judge_error='judge_input_refused: ...'`. Rate-limit / provider 5xx still go to the per-row error path. Auth still fails fast. (PR #44 commit dcaa91f) * New `P2M_ROLLOUT_ERROR_FAIL_RATIO` env var (default 0.10) caps the rollout stage's tolerance for *untyped* worker errors. Below the threshold: warn and continue (existing soft-fail). Above: raise the first error. Typed refusals (target_input_refused, auditor_input_refused, target_error) are NOT counted toward the ratio — they're recorded data, not real failures. Inspired by PR #44 commit 15332c8 but scoped to untyped errors only instead of #44's blanket runtime_error catch-all (which would have hidden real bugs behind the same threshold). * scripts/benchmark.py: drops `_install_content_filter_tolerance`, `_install_judge_skip_blocked`, and `_is_content_policy_violation` helpers (~210 lines). The product-code typed handlers above replace them. The benchmark now scans `transcripts.jsonl` and `scores.jsonl` after the run completes and counts typed stop_reasons / judge_status values. CSV columns updated to typed names: `target_input_refused_count`, `auditor_input_refused_count`, `target_error_count`, `judge_filter_skipped_count` (replaces the legacy `content_filter_blocked` / `target_error_tolerated` aggregates). `--no-tolerate-content-filter` flag preserved as a no-op for back-compat. Tests: * test_run_turn_reclassifies_litellm_bad_request_as_input_error * test_run_turn_passes_through_unclassified_exceptions * test_run_prompt_seed_records_target_input_refusal * test_run_rollout_isolates_target_input_refusal_to_one_seed * test_run_rollout_isolates_auditor_input_refusal_to_one_seed * test_run_rollout_still_fails_fast_on_provider_5xx * test_run_rollout_fails_when_untyped_error_ratio_exceeds_threshold (new — covers the threshold introduced in this commit) * test_run_judge_isolates_input_refusal_to_one_seed * test_run_rollout_keeps_partial_successful_transcripts_when_later_worker_fails (updated — sets P2M_ROLLOUT_ERROR_FAIL_RATIO=0.6 since the original 2-seed/1-failure setup now exceeds the production 10% default) 654 passed, 14 skipped, 13 subtests, 2 deselected (pre-existing Windows tempdir flakes in test_logging_config.py from main's PR #22). 0 new test failures. viewer check: 0 errors / 0 warnings. Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

All other modules use 'log = logging.getLogger(__name__)' per PR #22.

…ndling (#42) * fix: add try-except for file I/O and YAML/JSON parse errors Wrap unguarded file reads and parse calls with specific exception handlers that include the file path in the error message: - config.py: load_config catches FileNotFoundError, PermissionError, yaml.YAMLError and re-raises as ConfigError - otel.py: _parse_otlp_json catches FileNotFoundError, JSONDecodeError - tools.py: load_toolset_file catches FileNotFoundError, yaml.YAMLError - systematization_convert.py: run_systematization_to_policy catches FileNotFoundError, JSONDecodeError * fix: add error handling for callable and HTTP session imports - CallableSession.open: catch ModuleNotFoundError, AttributeError with actionable messages including the callable ref string - OTelTracedSession.open: same pattern for OTel-traced callable imports - HTTPEndpointSession.run_turn: catch aiohttp.ClientResponseError and ClientError with endpoint URL and HTTP status in the message * fix: improve judge and policy stage error handling - judge.py: catch JSONDecodeError when loading policy file, add LLM error types and json/ValueError before the catch-all in the worker, log traceback via log.debug for all caught worker exceptions - policy.py: include model name and raw response text in the error message when policy generation returns non-JSON output * fix: improve rollout and systematization error handling - rollout.py worker: add LLM error types re-raise before catch-all, add ValueError/KeyError handler, log.debug with traceback and seed_id for all caught worker exceptions - rollout.py auditor: log.debug with traceback on each retry failure - systematization.py: catch JSONDecodeError on json.loads fallback with raw text snippet in the error message * fix: add specific error handling for Phoenix collector Catch ConnectionError separately from generic Exception in PhoenixCollector.get_spans with a targeted message about Phoenix connectivity. Validate 'context.trace_id' column existence before filtering. * test: add 21 test cases for exception handling paths Cover all new error handling across config, otel, tools, session, otel_session, judge, systematization_convert, rollout, and collector. Verify exception types, error messages include context (file paths, module names, endpoint URLs), and debug logging fires with tracebacks in worker failures. * fix: skip Phoenix collector tests when pandas is not installed CI environment doesn't include pandas (optional dependency). The PhoenixCollector.get_spans method does 'import pandas' at the top, which fails before our exception handling runs. * style: replace print(file=sys.stderr) with log.warning() in artifact cache Align artifact_cache.py and viewer_read_model.py with the centralized logging convention from PR #22. Replace 10 print(file=sys.stderr) calls with log.warning() using structured placeholders so they respect --quiet, --log-file, and --output json. Remove unused sys imports from both modules. * fix: address review feedback — hoist aiohttp import, use high port in test - Move 'import aiohttp' from run_turn to open(), store as self._aiohttp to avoid re-importing on every turn call - Use port 59123 instead of port 1 in HTTPEndpointSession test for better portability across CI/container environments * Security hardening: validate dynamic imports, URLs, and credential handling Add p2m/core/security.py with centralized security validation: - validate_callable_ref(): block dangerous path segments in callable target references (FINDING-001, CWE-94) - validate_sys_path_addition(): scope sys.path additions to workspace boundaries (FINDING-002, CWE-427) - validate_endpoint_url(): block SSRF via private IPs and cloud metadata endpoints (FINDING-003, CWE-918) - sanitize_payload(): redact credentials from artifact files (FINDING-004, CWE-532) - validate_module_ref(): block dangerous segments in connector/tool module references (FINDING-006, CWE-94) - _validate_module_file_path(): enforce workspace boundaries on direct .py file loading (FINDING-010, CWE-94) - validate_dotenv_location(): warn on suspicious .env locations (FINDING-009, CWE-426) - P2M_NO_AUTO_ENV opt-out for auto_envvar_prefix (FINDING-005, CWE-15) - _require_within() on resolved artifact paths (FINDING-007, CWE-22) Opt-out environment variables for backward compatibility: - P2M_TRUST_CALLABLE=1: suppress callable import warnings - P2M_ALLOW_PRIVATE_ENDPOINTS=1: allow localhost/private endpoints - P2M_NO_DOTENV=1: disable automatic .env loading - P2M_NO_AUTO_ENV=1: disable P2M_* env var overrides Update test_exception_handling.py to use P2M_ALLOW_PRIVATE_ENDPOINTS for localhost endpoint test. * Fix SSRF DNS rebinding bypass, sanitization depth leak, and response credential leak - validate_endpoint_url(): resolve hostnames via socket.getaddrinfo() and validate resulting IPs against blocked ranges (FINDING-002, CWE-918) - sanitize_payload(): redact at max_depth instead of passing through unsanitized (FINDING-003, CWE-200) - HTTPEndpointSession: add _sanitize_response_text() to redact credential patterns before writing to transcripts (FINDING-004, CWE-532) * fix: add Azure Wireserver IP (168.63.129.16) to SSRF blocklist * refactor: remove vestigial dotenv/env-var security machinery Remove P2M_NO_DOTENV, P2M_NO_AUTO_ENV, P2M_TRUST_CALLABLE and associated code (should_load_dotenv, validate_dotenv_location, is_callable_trusted, P2MSecurityWarning). These addressed scenarios from an incorrect threat model — load_dotenv(override=False) already safe, CI owns its env vars, and callable import warnings block the primary workflow. Keep: format validation, blocklist, SSRF prevention, credential sanitization. * style: use 'log' instead of '_log' in session.py for consistency All other modules use 'log = logging.getLogger(__name__)' per PR #22. * test: add unit tests for security module 50 tests covering validate_callable_ref, validate_module_ref, validate_sys_path_addition, validate_endpoint_url (SSRF/DNS rebinding), sanitize_payload, and _sanitize_response_text.

tangym added 15 commits May 6, 2026 18:09

refactor: replace raw print(file=stderr) with structured logging

84714b1

runner.py: 2 print() calls → log.error() for config errors config.py: 2 print() calls → log.warning() for validation warnings Remove unused sys import from config.py.

fix: update runner progress tests for logging migration

5e3af3e

Tests previously patched sys.stderr/sys.__stderr__ to capture output. Now that runner output goes through the logging system, switch to assertLogs('p2m.runner') which captures log records directly. Remove unused io import.

style: use f-strings consistently in all log calls

4cc17d5

Convert %s/%d placeholder-style log calls to f-strings for readability. The performance difference is negligible at this call volume (~30 calls per pipeline run).

Merge remote-tracking branch 'origin/main' into yemingtang/logging

9af090d

# Conflicts: # p2m/core/model_client.py

tangym force-pushed the yemingtang/logging branch from 6430d2e to ba29c84 Compare May 6, 2026 23:10

tangym added 4 commits May 6, 2026 23:16

tangym force-pushed the yemingtang/logging branch from 1c64431 to ebb0a5d Compare May 6, 2026 23:53

Merge branch 'yemingtang/logging-json-output' into yemingtang/logging

1d75a1d

tangym changed the title ~~feat: centralized logging with --verbose/--quiet/--log-file CLI flags~~ feat: centralized logging with --verbose/--quiet/--log-file/--output json CLI flags May 7, 2026

AaronAspinwall123 approved these changes May 8, 2026

View reviewed changes

tangym merged commit ef6d45a into main May 8, 2026
3 checks passed

changliu2 mentioned this pull request May 8, 2026

Hashing and rerun logic #32

Merged

tangym deleted the yemingtang/logging branch May 9, 2026 00:48

changliu2 mentioned this pull request May 11, 2026

fix(viewer): collapse tool calls into target turn; surface sub-agent routing #38

Merged

tangym added a commit that referenced this pull request May 14, 2026

style: use 'log' instead of '_log' in session.py for consistency

e7dc871

All other modules use 'log = logging.getLogger(__name__)' per PR #22.

tangym mentioned this pull request May 14, 2026

Security hardening: validate dynamic imports, URLs, and credential handling #42

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

feat: centralized logging with --verbose/--quiet/--log-file/--output json CLI flags - #22

feat: centralized logging with --verbose/--quiet/--log-file/--output json CLI flags#22
tangym merged 20 commits into
mainfrom
yemingtang/logging

tangym commented May 6, 2026 •

edited

Loading

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

tangym commented May 6, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Changes

Phase 1: Logging bootstrap

Phase 2: Unify stdlib loggers

Phase 3: Migrate _progress() → log.info()

Phase 4: Migrate raw print(file=stderr)

Phase 5: Add logging to silent modules

Phase 6: Structured JSON output

Additional improvements

Test results

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

tangym commented May 6, 2026 •

edited

Loading

Phase 3: Migrate `_progress()` → `log.info()`

Phase 4: Migrate raw `print(file=stderr)`