agent-usage-tracker 0.3.0__tar.gz → 0.4.0__tar.gz
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/PKG-INFO +3 -1
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/docs/development.md +28 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/docs/evidence.md +100 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/pyproject.toml +2 -1
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/__init__.py +1 -1
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/backup.py +2 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/access.py +7 -2
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/access_cli.py +86 -3
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/access_mcp.py +27 -1
- agent_usage_tracker-0.4.0/src/agent_usage_tracker/improve/investigation.py +256 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/runtime.py +12 -6
- agent_usage_tracker-0.4.0/src/agent_usage_tracker/improve/session_observability.py +55 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/turn_revisions.py +11 -3
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/collectors/test_cursor_api.py +5 -3
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/collectors/test_source_registry.py +3 -1
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_evidence_access.py +9 -4
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_evidence_mcp.py +50 -0
- agent_usage_tracker-0.4.0/tests/improve/test_investigation.py +147 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_native_evidence.py +8 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_runtime.py +26 -0
- agent_usage_tracker-0.4.0/tests/improve/test_session_observability.py +43 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_turn_revisions.py +44 -5
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/test_observation_server.py +14 -3
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/.gitignore +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/CONTRACT.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/DEPLOY.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/LICENSE +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/README.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/config.example.toml +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/docs/assets/aut-local-setup.png +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/docs/assets/aut-workflow.png +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/docs/collectors-to-add.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/docs/first-run.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/docs/harness-coverage.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/docs/human-html/2026-11-08-understanding-implicit-state-machines.html +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/docs/human-html/2026-11-09-status-state-reliability-fixes.html +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/docs/human-html/docs-index.json +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/docs/human-html/index.html +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/docs/improvement-studies.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/docs/improvement.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/docs/local-labels.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/docs/new-laptop.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/docs/observations.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/docs/quota.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/docs/reference.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/docs/research/codexbar-learnings.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/docs/research/otel-lane.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/docs/research/sdk-usage-capture.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/docs/superpowers/plans/2026-07-22-aut-showcase-metrics.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/docs/superpowers/specs/2026-07-22-aut-showcase-metrics-design.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/packaging/README.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/packaging/claude/commands/outcome.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/packaging/launchd/aut-scan.sh +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/packaging/launchd/dev.aut.backup.plist +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/packaging/launchd/dev.aut.improve-audit.plist +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/packaging/launchd/dev.aut.improve-draft.plist +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/packaging/launchd/dev.aut.retry.plist +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/packaging/launchd/dev.aut.scan.plist +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/packaging/observations/Dockerfile +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/packaging/observations/Dockerfile.dockerignore +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/packaging/observations/README.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/packaging/observations/aut-observations.service +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/packaging/sql/turn-snapshot-grant.sql +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/packaging/systemd/agent-usage-tracker.service +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/packaging/systemd/agent-usage-tracker.timer +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/packaging/windows/register-task.ps1 +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/cli.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/collectors/__init__.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/collectors/_util.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/collectors/activity_common.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/collectors/activitywatch.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/collectors/claude_code.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/collectors/claude_code_activity.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/collectors/codex.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/collectors/commits.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/collectors/cursor.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/collectors/cursor_api.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/collectors/cursor_cli.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/collectors/cursor_sdk.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/collectors/generic_sdk.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/collectors/pi.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/collectors/session_digest.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/collectors/t3.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/config.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/coverage.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/coverage_catalog.json +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/cursor_run.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/demo.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/demo_cli.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/diagnostics.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/episode_store.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/event.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/git_day_cache.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/git_days.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/git_safe.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/harness_inventory.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/history_steers.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/__init__.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/case_view.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/cli.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/cursor_evidence.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/decide.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/desktop.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/draft.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/draft_adapters.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/draft_gguf.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/evidence.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/file_reads.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/groups.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/inbox.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/judge.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/outputs.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/proposals.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/qualify.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/ranked_search.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/report_card.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/rubric.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/rubrics/avoidable_recovery.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/rubrics/code_quality.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/rubrics/completion_supported.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/rubrics/redundant_tests.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/rubrics/verbosity.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/schedule.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/scores.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/search_check.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/setup_git.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/skill_proposals.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/store.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/studies.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/study_candidates.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/study_cli.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/study_decision.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/study_followup.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/study_independence.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/study_model_calls.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/study_observe.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/study_reviews.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/t3.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/trial.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/validation.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve_publish.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/inspector.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/label/__init__.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/label/backends.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/label/bakeoff.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/label/baseline.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/label/clef.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/label/context.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/label/core.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/label/covered_run.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/label/evalset.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/label/gold.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/label/local.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/label/local_parallel.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/label/merge.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/label/outcome_coverage.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/label/replay.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/label/run.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/label/tasks.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/label/teach.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/local_page.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/manual_time.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/net.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/observation_cli.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/observation_report.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/observation_server.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/observations.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/operational_status.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/outcome_git.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/outcomes.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/overview.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/overview_cli.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/pricing.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/pricing_cli.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/push.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/quota.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/quota_cli.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/redact.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/repo_keys.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/review.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/scan.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/scheduled.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/sdk.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/setup_cli.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/setup_schedule.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/sink.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/sink_stack/README.md +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/sink_stack/apply.sh +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/sink_stack/compose.yaml +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/sink_stack/env.example +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/sink_stack/grafana/dashboards/agents-effort.json +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/sink_stack/grafana/dashboards/agents-usage.json +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/sink_stack/grafana/provisioning/dashboards/aut.yaml +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/sink_stack/grafana/provisioning/datasources/agents.yaml +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/sink_stack/sql/01_schema.sql +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/sink_stack/sql/02_price.sql +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/sink_stack/sql/03_effort.sql +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/sink_stack/sql/04_grants.sql +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/sink_stack/sql/refresh.sql +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/sources.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/state.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/summary.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/summary_cli.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/telemetry.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/usage_price.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/collectors/codex_pi_helpers.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/collectors/test_activity.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/collectors/test_activitywatch.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/collectors/test_activitywatch_macos.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/collectors/test_claude_code.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/collectors/test_claude_profiles.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/collectors/test_codex.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/collectors/test_codex_pi.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/collectors/test_collectors.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/collectors/test_commits.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/collectors/test_cursor.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/collectors/test_cursor_cli.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/collectors/test_cursor_cli_cache.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/collectors/test_cursor_sdk.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/collectors/test_generic_sdk.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/collectors/test_links.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/collectors/test_links_claude.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/collectors/test_pi.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/collectors/test_read_horizon.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/collectors/test_session_digest.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/collectors/test_t3.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/conftest.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/fixtures/claude-statusline-quota.json +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/__init__.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/conftest.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_case_context.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_case_view.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_cli.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_count_evidence.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_decide.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_desktop.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_draft.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_draft_adapters.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_evidence.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_evidence_citations.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_file_reads.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_groups.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_inbox.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_judge.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_judge_choices.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_judge_deadline.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_judge_native_deadline.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_judge_queue.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_judge_sampling.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_judge_stall.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_outputs.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_proposals.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_qualify.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_ranked_evidence.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_report_card.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_rubric.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_schedule.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_scores.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_search_check.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_skill_proposals.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_store.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_studies.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_studies_decision.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_study_candidates.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_t3.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_trial.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_turn_provenance.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/improve/test_validation.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/label/label_helpers.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/label/test_context_paths.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/label/test_covered_run.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/label/test_cursor_origin.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/label/test_history_steers.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/label/test_label.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/label/test_label_bakeoff.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/label/test_label_baseline.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/label/test_label_context.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/label/test_label_core.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/label/test_label_evalset.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/label/test_label_gap.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/label/test_label_gold.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/label/test_label_local.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/label/test_label_models.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/label/test_label_parallel.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/label/test_label_reliability.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/label/test_label_replay.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/label/test_label_tasks.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/label/test_label_teach.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/label/test_label_writing.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/label/test_outcome_coverage.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/outcomes/conftest.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/outcomes/test_outcome_git.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/outcomes/test_outcome_reliability.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/outcomes/test_outcomes.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/fake_psycopg.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/price_fixture.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_backup.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_cli.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_coverage.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_coverage_diagnostics.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_coverage_scan.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_cursor_run.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_delivery_cli.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_delivery_notices.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_demo.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_first_run.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_git_day_cache.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_git_days.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_harness_inventory.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_heartbeat.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_inspect.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_ledger_zone.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_local_page.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_net.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_overview.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_overview_quota.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_pricing.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_publish_improve.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_push.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_push_outbox.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_push_snapshots.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_quota.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_redact.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_repo_keys.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_review.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_scan_wrapper.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_sessions.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_sink.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_sink_sql.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_state.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_summary.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/pipeline/test_summary_native_counts.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/test_observation_cli.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/test_observation_report.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/test_observations.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/test_observe_entrypoint.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/test_public_projection.py +0 -0
- {agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/tests/test_telemetry.py +0 -0
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
Metadata-Version: 2.5
|
|
2
2
|
Name: agent-usage-tracker
|
|
3
|
-
Version: 0.
|
|
3
|
+
Version: 0.4.0
|
|
4
4
|
Summary: Track coding-agent usage and outcomes from local logs, find recurring problems, and test proposed fixes before adopting them.
|
|
5
5
|
Project-URL: Homepage, https://github.com/rhnfzl/agent-usage-tracker
|
|
6
6
|
Project-URL: Issues, https://github.com/rhnfzl/agent-usage-tracker/issues
|
|
@@ -14,6 +14,8 @@ Classifier: Programming Language :: Python :: 3
|
|
|
14
14
|
Classifier: Topic :: Software Development
|
|
15
15
|
Requires-Python: >=3.11
|
|
16
16
|
Requires-Dist: tzdata; sys_platform == 'win32'
|
|
17
|
+
Provides-Extra: code
|
|
18
|
+
Requires-Dist: pydantic-monty<1.7,>=1.6; extra == 'code'
|
|
17
19
|
Provides-Extra: label
|
|
18
20
|
Requires-Dist: huggingface-hub>=1.3; (sys_platform == 'darwin' and platform_machine == 'arm64') and extra == 'label'
|
|
19
21
|
Requires-Dist: mlx-lm<0.33,>=0.32.2; (sys_platform == 'darwin' and platform_machine == 'arm64') and extra == 'label'
|
|
@@ -36,6 +36,34 @@ In GitHub's branch protection for `main`, require a pull request and the `CI req
|
|
|
36
36
|
Require branches to be up to date before merging, and apply the rule to administrators.
|
|
37
37
|
CI also runs on `main` to verify the landed commit.
|
|
38
38
|
SQL integration tests require a scratch database on a TimescaleDB server and `psql`, through `AUT_TEST_PG_DSN` with a literal IP address and port; otherwise those tests skip.
|
|
39
|
+
|
|
40
|
+
## Local preflight and packages
|
|
41
|
+
|
|
42
|
+
Run these commands on macOS or Linux with `uv`, Docker Compose, `psql`, and Perl with `shasum` available.
|
|
43
|
+
|
|
44
|
+
```bash
|
|
45
|
+
./scripts/local-ci.sh quick
|
|
46
|
+
./scripts/local-ci.sh check
|
|
47
|
+
./scripts/local-ci.sh package HEAD
|
|
48
|
+
```
|
|
49
|
+
|
|
50
|
+
`quick` runs Ruff. `check` runs Ruff, mypy, the two suite time zones, the MCP suite, and the SQL and Compose checks.
|
|
51
|
+
These checks run concurrently. The preflight reserves cores for other checks and uses up to six pytest workers per time zone.
|
|
52
|
+
Set `AUT_CI_WORKERS` to lower the worker count per time zone if the machine is busy.
|
|
53
|
+
It uses a disposable TimescaleDB container on a random loopback port.
|
|
54
|
+
The preflight defaults to Python 3.11. Set `AUT_CI_PYTHON` to test another supported version.
|
|
55
|
+
For the Mac batch publisher, pass `./scripts/local-ci.sh quick` as `--light` and `./scripts/local-ci.sh check` as `--full`.
|
|
56
|
+
The full local preflight fails above 60 seconds, including dependency setup and cleanup.
|
|
57
|
+
It lets every check finish and reports the results before rejecting an over-budget run.
|
|
58
|
+
GitHub still runs the required platform matrix for the pull request. That run has its own runner queue and timing.
|
|
59
|
+
|
|
60
|
+
`package` builds a wheel from the source distribution of the named commit.
|
|
61
|
+
It checks both distributions, installs the wheel in isolation, and saves the artifacts under `dist/local-<short-commit>/`.
|
|
62
|
+
The directory contains `SOURCE_COMMIT` and `SHA256SUMS` for deployment verification.
|
|
63
|
+
Build from the merged commit when preparing a deployment candidate.
|
|
64
|
+
Check the bundle with `shasum -a 256 -c SHA256SUMS` from its output directory.
|
|
65
|
+
Use the verified wheel with the [standalone runtime installer](improvement.md#install-a-stable-runtime) to stage an upgrade before switching a timer.
|
|
66
|
+
|
|
39
67
|
To release, raise `__version__` in `src/agent_usage_tracker/__init__.py`, merge, and push the tag `v<version>`.
|
|
40
68
|
The [release workflow](../.github/workflows/release.yml) builds it and publishes it to PyPI through trusted publishing.
|
|
41
69
|
Use synthetic fixtures in public contributions. Keep session logs, personal labels, and private benchmarks out of commits.
|
|
@@ -179,3 +179,103 @@ hash. To trial the addition, explicitly list the file with a null value in `file
|
|
|
179
179
|
SHA-256 hash in `files.candidate`. This binds the trial to the absence of the skill before the change.
|
|
180
180
|
A successful tool result is evidence for the cited step. Review whether the proposed instructions
|
|
181
181
|
apply to the whole workflow before you adopt the skill.
|
|
182
|
+
|
|
183
|
+
## Run a programmable investigation
|
|
184
|
+
|
|
185
|
+
A recipe can use evidence from one session to decide what to search for in another.
|
|
186
|
+
Give it explicit audits or bundles, then return citations, a timeline, a hypothesis,
|
|
187
|
+
or a statistic with its counting rule.
|
|
188
|
+
|
|
189
|
+
```sh
|
|
190
|
+
pip install 'agent-usage-tracker[code]'
|
|
191
|
+
aut evidence code-catalog
|
|
192
|
+
aut evidence --audit FIRST --content redacted run-code follow.py --include-audit SECOND --save
|
|
193
|
+
```
|
|
194
|
+
|
|
195
|
+
For example, put this program in `follow.py`. It takes a lead from a match in the
|
|
196
|
+
first audit and searches the second. The last expression becomes the JSON result.
|
|
197
|
+
|
|
198
|
+
```python
|
|
199
|
+
seed = read("search", query="setup failed", limit=1)
|
|
200
|
+
found = []
|
|
201
|
+
if seed["data"]["items"]:
|
|
202
|
+
query = seed["data"]["items"][0]["preview"].split()[-1]
|
|
203
|
+
cursor = ""
|
|
204
|
+
while cursor is not None:
|
|
205
|
+
page = read("search", scope="2", lead=seed["receipt"], query=query, cursor=cursor)
|
|
206
|
+
for hit in page["data"]["items"]:
|
|
207
|
+
found.append({"session_id": hit["session_id"], "event_id": hit["event_id"]})
|
|
208
|
+
cursor = page["data"]["next_cursor"]
|
|
209
|
+
found
|
|
210
|
+
```
|
|
211
|
+
|
|
212
|
+
Use `-` in place of the path to read source from stdin. Pass a JSON object with
|
|
213
|
+
`--params` to populate the program's `params` variable. The `scopes` variable lists
|
|
214
|
+
the actual coverage and content grant per source. Scope `1` is the main audit or
|
|
215
|
+
bundle. Additional audits come next, followed by additional bundles. Each type
|
|
216
|
+
retains command-line order. Bundles can narrow the content grant. Code mode
|
|
217
|
+
defaults to redacted access. Existing direct commands keep their own defaults.
|
|
218
|
+
|
|
219
|
+
The catalog lists the read operations, their parameters, and defaults. Set `lead`
|
|
220
|
+
to an earlier read's receipt to record which observation motivated a later read.
|
|
221
|
+
Each successful read records available event references, a result digest, and
|
|
222
|
+
whether a further page was offered. Continue an empty page if it has a cursor.
|
|
223
|
+
Check scope and session coverage before drawing conclusions. A returned citation
|
|
224
|
+
identifies an observed record. A person or a reviewed rule must check what it
|
|
225
|
+
supports. Snapshot IDs may repeat across audits or change after a recapture.
|
|
226
|
+
When comparing distinct source sessions, use an available source identity and
|
|
227
|
+
report cases where it is unknown. The readers expose recorded Git outcomes and
|
|
228
|
+
study evidence. They require independently verified links for joins with usage
|
|
229
|
+
or records outside the selected evidence.
|
|
230
|
+
|
|
231
|
+
Recipes run in fresh local Monty sessions. The runner limits source size, host
|
|
232
|
+
calls, evidence bytes, output, and wall time. Failed execution reports an error.
|
|
233
|
+
Recipes use the Python subset supported by Monty. They get no general host
|
|
234
|
+
filesystem, network, or shell operations. Plan a stopping rule for recursive
|
|
235
|
+
searches, and report which leads you left open if you stop early.
|
|
236
|
+
|
|
237
|
+
Save a run with `--save` to keep its program, parameters, scopes, and read receipts
|
|
238
|
+
in the private AUT store. Saving can retain sensitive text in a program or its
|
|
239
|
+
parameters. The default AUT backup exclusions cover this directory under the
|
|
240
|
+
standard store and `AUT_IMPROVE_ROOT`. Add a backup exclusion when using a different
|
|
241
|
+
`--root` inside a backed-up directory.
|
|
242
|
+
|
|
243
|
+
```sh
|
|
244
|
+
aut evidence --audit FIRST replay-code RUN_ID --include-audit SECOND
|
|
245
|
+
aut evidence --audit NEW_FIRST run-code --recipe RUN_ID --include-audit NEW_SECOND
|
|
246
|
+
```
|
|
247
|
+
|
|
248
|
+
Replay checks the original scope and compares the result and read record with the
|
|
249
|
+
saved digest. Deleted inputs can prevent replay. Changes in AUT or Monty versions
|
|
250
|
+
can change the digest. Use `--recipe` to run the same program on new inputs and
|
|
251
|
+
optionally pass new parameters.
|
|
252
|
+
|
|
253
|
+
Install `agent-usage-tracker[mcp,code]` to use a recipe through MCP. If you use
|
|
254
|
+
the pinned-wheel runtime installer, pass `--code` to include both optional dependencies. Add `--code`
|
|
255
|
+
to the server arguments after `mcp` and include any additional scopes there.
|
|
256
|
+
The server adds `code_catalog` and `run_code(source, params)`. The client receives
|
|
257
|
+
the result under the connection's grant. Restart the MCP server to select a
|
|
258
|
+
changed bundle. A client connected to a hosted model can
|
|
259
|
+
forward that result. Session observability settings govern a separate delivery
|
|
260
|
+
path.
|
|
261
|
+
|
|
262
|
+
## Send session evidence to a configured receiver
|
|
263
|
+
|
|
264
|
+
Set `AUT_SESSION_OBSERVABILITY_URL` to an HTTPS endpoint that accepts AUT evidence
|
|
265
|
+
bundle JSON by POST. Loopback HTTP is available for local development. Set
|
|
266
|
+
`AUT_SESSION_OBSERVABILITY_TOKEN` if your receiver requires bearer authentication.
|
|
267
|
+
AUT uses a dedicated endpoint for session evidence. The existing usage sink
|
|
268
|
+
continues to receive usage data.
|
|
269
|
+
|
|
270
|
+
With a destination configured, `aut evidence capture` sends the new audit with
|
|
271
|
+
redacted content by default. Set `AUT_SESSION_OBSERVABILITY=metadata` for event
|
|
272
|
+
references and coverage, `full` for available raw session text, or `off` to stop
|
|
273
|
+
delivery. Use `capture --no-observe` to keep a single capture local. A capture or
|
|
274
|
+
bundle can further restrict what is sent. Redaction masks recognized patterns.
|
|
275
|
+
Review the content policy and the receiver's retention before sharing private work.
|
|
276
|
+
|
|
277
|
+
A delivery failure appears in the capture's JSON as `observability.status=failed`.
|
|
278
|
+
The private audit remains available. Retry with
|
|
279
|
+
`aut evidence --audit AUDIT_ID observe`, or select a bundle to deliver a smaller
|
|
280
|
+
set of sessions. The request carries an `Idempotency-Key` that matches the bundle
|
|
281
|
+
digest. A receiver can use it to deduplicate identical deliveries.
|
|
@@ -36,6 +36,7 @@ merge = ["torch>=2.14", "peft>=0.21", "transformers>=5.17", "huggingface_hub>=1.
|
|
|
36
36
|
|
|
37
37
|
# The evidence CLI runs without the optional MCP server.
|
|
38
38
|
mcp = ["mcp>=1.30,<2"]
|
|
39
|
+
code = ["pydantic-monty>=1.6,<1.7"]
|
|
39
40
|
|
|
40
41
|
# Run the suite: pip install -e '.[test]' && pytest -q
|
|
41
42
|
test = ["pytest", "pytest-xdist==3.8.0"]
|
|
@@ -89,5 +90,5 @@ disallow_any_generics = true
|
|
|
89
90
|
# Optional extras: the base install, and CI's lint job, have none of them.
|
|
90
91
|
[[tool.mypy.overrides]]
|
|
91
92
|
module = ["psycopg.*", "mlx.*", "mlx_lm.*", "huggingface_hub.*", "transformers.*", "torch.*", "peft.*",
|
|
92
|
-
"mcp.*", "anyio.*", "uvicorn.*", "starlette.*"]
|
|
93
|
+
"mcp.*", "anyio.*", "uvicorn.*", "starlette.*", "pydantic_monty.*"]
|
|
93
94
|
ignore_missing_imports = true
|
|
@@ -71,6 +71,8 @@ def default_excludes() -> list[str]:
|
|
|
71
71
|
data_dir() / "label" / "models",
|
|
72
72
|
data_dir() / "improve" / "session_content",
|
|
73
73
|
root.resolve() / "session_content",
|
|
74
|
+
data_dir() / "improve" / "investigation_runs",
|
|
75
|
+
root.resolve() / "investigation_runs",
|
|
74
76
|
))))
|
|
75
77
|
|
|
76
78
|
|
{agent_usage_tracker-0.3.0 → agent_usage_tracker-0.4.0}/src/agent_usage_tracker/improve/access.py
RENAMED
|
@@ -460,7 +460,7 @@ class Evidence:
|
|
|
460
460
|
agent, before, include_excluded, cursor, limit, self._label,
|
|
461
461
|
supplemental=self.supplemental)
|
|
462
462
|
|
|
463
|
-
def
|
|
463
|
+
def bundle_bytes(self, ids: list[str]) -> tuple[str, bytes]:
|
|
464
464
|
if not ids or len(set(ids)) != len(ids):
|
|
465
465
|
raise AutError("evidence: select distinct session IDs for export")
|
|
466
466
|
sessions = {}
|
|
@@ -481,9 +481,14 @@ class Evidence:
|
|
|
481
481
|
sessions[key] = projected
|
|
482
482
|
payload = {"version": 1, "audit_id": self.audit_id, "content": self.content,
|
|
483
483
|
"coverage": _coverage(self.coverage), "sessions": sessions}
|
|
484
|
-
|
|
484
|
+
identity = digest(payload)
|
|
485
|
+
data = (canonical({"sha256": identity, "evidence": payload}) + "\n").encode("utf-8")
|
|
485
486
|
if len(data) > BUNDLE_LIMIT:
|
|
486
487
|
raise AutError("evidence: select fewer sessions for the 64 MiB bundle limit")
|
|
488
|
+
return identity, data
|
|
489
|
+
|
|
490
|
+
def export(self, path: Path, ids: list[str]) -> dict[str, Any]:
|
|
491
|
+
_, data = self.bundle_bytes(ids)
|
|
487
492
|
_create_bundle(path, data)
|
|
488
493
|
return {"path": str(path), "sessions": len(ids), "content": self.content}
|
|
489
494
|
|
|
@@ -25,7 +25,7 @@ def register(commands: Any) -> None:
|
|
|
25
25
|
scope = parser.add_mutually_exclusive_group()
|
|
26
26
|
scope.add_argument("--audit", help="freeze access to the selected audit ID")
|
|
27
27
|
scope.add_argument("--bundle", type=Path, help="open a previously exported evidence bundle")
|
|
28
|
-
parser.add_argument("--content", choices=MODES,
|
|
28
|
+
parser.add_argument("--content", choices=MODES,
|
|
29
29
|
help="content available through this command or MCP connection")
|
|
30
30
|
actions = parser.add_subparsers(dest="evidence_cmd", required=True)
|
|
31
31
|
|
|
@@ -34,6 +34,8 @@ def register(commands: Any) -> None:
|
|
|
34
34
|
capture.add_argument("--limit", type=int, default=20)
|
|
35
35
|
capture.add_argument("--idle-s", type=int, default=120)
|
|
36
36
|
capture.add_argument("--project", type=Path)
|
|
37
|
+
capture.add_argument("--no-observe", action="store_true", help="keep this capture local")
|
|
38
|
+
actions.add_parser("observe", help="send the selected audit or bundle to session observability")
|
|
37
39
|
capture.add_argument("--max-mib", type=int, default=BYTE_LIMIT // _MIB,
|
|
38
40
|
help="maximum MiB read per source (1..256, default: 16)")
|
|
39
41
|
record = actions.add_parser("record-codex-turn", help="record Git context from a Codex prompt hook")
|
|
@@ -94,12 +96,31 @@ def register(commands: Any) -> None:
|
|
|
94
96
|
export.add_argument("ids", nargs="+")
|
|
95
97
|
mcp = actions.add_parser("mcp", help="serve the selected evidence through MCP")
|
|
96
98
|
mcp.add_argument("--transport", choices=("stdio", "http"), default="stdio")
|
|
99
|
+
code = actions.add_parser("run-code", help="run a local evidence recipe")
|
|
100
|
+
code.add_argument("source", nargs="?", help="Python recipe path or - for stdin")
|
|
101
|
+
code.add_argument("--recipe", help="reuse the source from a saved run")
|
|
102
|
+
code.add_argument("--params", help="JSON object passed to the program")
|
|
103
|
+
code.add_argument("--save", action="store_true", help="save private recipe and replay record")
|
|
104
|
+
replay = actions.add_parser("replay-code", help="replay a saved investigation")
|
|
105
|
+
replay.add_argument("run_id")
|
|
106
|
+
actions.add_parser("code-catalog", help="list operations available to recipes")
|
|
107
|
+
mcp.add_argument("--code", action="store_true", help="enable programmable investigations")
|
|
108
|
+
for action in (code, replay, mcp):
|
|
109
|
+
action.add_argument("--include-audit", action="append", default=[])
|
|
110
|
+
action.add_argument("--include-bundle", action="append", type=Path, default=[])
|
|
97
111
|
mcp.add_argument("--port", type=int, default=8765, help="loopback port for HTTP")
|
|
98
112
|
|
|
99
113
|
|
|
100
114
|
def run(args: argparse.Namespace) -> None:
|
|
101
115
|
root = args.root or Path(os.environ.get("AUT_IMPROVE_ROOT", str(data_dir() / "improve")))
|
|
102
116
|
action = args.evidence_cmd
|
|
117
|
+
with_code = action in {"run-code", "replay-code", "code-catalog"} or (
|
|
118
|
+
action == "mcp" and args.code)
|
|
119
|
+
args.content = args.content or ("redacted" if with_code else "full")
|
|
120
|
+
if action == "code-catalog" or action in {"run-code", "replay-code"}:
|
|
121
|
+
value = _code(args, root)
|
|
122
|
+
print(json.dumps(value, sort_keys=True))
|
|
123
|
+
return
|
|
103
124
|
if action == "record-codex-turn":
|
|
104
125
|
if args.audit or args.bundle:
|
|
105
126
|
raise AutError("evidence: the Codex turn hook uses a store, not an audit or bundle")
|
|
@@ -112,13 +133,24 @@ def run(args: argparse.Namespace) -> None:
|
|
|
112
133
|
raise AutError("evidence: capture takes no scope and requires full or redacted content")
|
|
113
134
|
if not 1 <= args.max_mib <= _MAX_CAPTURE_MIB:
|
|
114
135
|
raise AutError("evidence: use a per-source capture budget of 1..256 MiB")
|
|
115
|
-
|
|
136
|
+
writable = Store(root)
|
|
137
|
+
value = audit(writable, Path.home(), days=args.days, limit=args.limit,
|
|
116
138
|
idle_s=args.idle_s, project=args.project, capture_full=args.content == "full",
|
|
117
139
|
max_bytes=args.max_mib * _MIB)
|
|
140
|
+
if not args.no_observe and os.environ.get("AUT_SESSION_OBSERVABILITY_URL"):
|
|
141
|
+
from .session_observability import publish
|
|
142
|
+
try:
|
|
143
|
+
value["observability"] = publish(args.content, writable, value["audit_id"], None)
|
|
144
|
+
except AutError as exc:
|
|
145
|
+
value["observability"] = {"status": "failed", "error": str(exc)}
|
|
118
146
|
else:
|
|
119
147
|
if not (args.audit or args.bundle):
|
|
120
148
|
raise AutError("evidence: select --audit ID or --bundle PATH")
|
|
121
149
|
store = Store(root, read_only=True) if args.audit else None
|
|
150
|
+
if action == "observe":
|
|
151
|
+
from .session_observability import publish
|
|
152
|
+
print(json.dumps(publish(args.content, store, args.audit, args.bundle), sort_keys=True))
|
|
153
|
+
return
|
|
122
154
|
evidence = Evidence(args.content, store, args.audit, args.bundle)
|
|
123
155
|
if action == "mcp":
|
|
124
156
|
try:
|
|
@@ -127,7 +159,14 @@ def run(args: argparse.Namespace) -> None:
|
|
|
127
159
|
if exc.name not in {"mcp", "anyio", "uvicorn", "starlette"}:
|
|
128
160
|
raise
|
|
129
161
|
raise AutError("evidence: install agent-usage-tracker[mcp] for MCP access") from exc
|
|
130
|
-
|
|
162
|
+
scopes = None
|
|
163
|
+
if args.code:
|
|
164
|
+
from .investigation import scope_specs
|
|
165
|
+
scopes = scope_specs(root, args.content, args.audit, args.bundle,
|
|
166
|
+
args.include_audit, args.include_bundle)
|
|
167
|
+
elif args.include_audit or args.include_bundle:
|
|
168
|
+
raise AutError("investigation: additional scopes require mcp --code")
|
|
169
|
+
serve(evidence, args.transport, args.port, scopes)
|
|
131
170
|
return
|
|
132
171
|
if action == "propose-skill":
|
|
133
172
|
if store is None:
|
|
@@ -178,3 +217,47 @@ def run(args: argparse.Namespace) -> None:
|
|
|
178
217
|
else:
|
|
179
218
|
value = evidence.describe()
|
|
180
219
|
print(json.dumps(value, sort_keys=True))
|
|
220
|
+
|
|
221
|
+
|
|
222
|
+
def _code(args: argparse.Namespace, root: Path) -> dict[str, Any]:
|
|
223
|
+
from .investigation import CONTRACT, MAX_SOURCE, catalog, run, scope_specs
|
|
224
|
+
from .store import digest
|
|
225
|
+
|
|
226
|
+
if args.evidence_cmd == "code-catalog":
|
|
227
|
+
return catalog()
|
|
228
|
+
if not (args.audit or args.bundle):
|
|
229
|
+
raise AutError("investigation: select --audit ID or --bundle PATH")
|
|
230
|
+
specs = scope_specs(root, args.content, args.audit, args.bundle,
|
|
231
|
+
args.include_audit, args.include_bundle)
|
|
232
|
+
if args.evidence_cmd == "replay-code":
|
|
233
|
+
record = Store(root, read_only=True).get("investigation_runs", args.run_id)
|
|
234
|
+
if record["contract"] != CONTRACT:
|
|
235
|
+
raise AutError("investigation: saved recipe uses an unsupported operation contract")
|
|
236
|
+
value = run(record["source"], record["params"], specs, record["scopes"])
|
|
237
|
+
return {**value, "replayed": args.run_id, "matches": digest(value) == record["result_sha256"]}
|
|
238
|
+
if bool(args.source) == bool(args.recipe):
|
|
239
|
+
raise AutError("investigation: select a source path, stdin, or --recipe ID")
|
|
240
|
+
try:
|
|
241
|
+
params = json.loads(args.params) if args.params is not None else {}
|
|
242
|
+
if args.recipe:
|
|
243
|
+
recipe = Store(root, read_only=True).get("investigation_runs", args.recipe)
|
|
244
|
+
if recipe["contract"] != CONTRACT:
|
|
245
|
+
raise AutError("investigation: saved recipe uses an unsupported operation contract")
|
|
246
|
+
source = recipe["source"]
|
|
247
|
+
if args.params is None:
|
|
248
|
+
params = recipe["params"]
|
|
249
|
+
elif args.source == "-":
|
|
250
|
+
source = sys.stdin.buffer.read(MAX_SOURCE + 1).decode("utf-8")
|
|
251
|
+
else:
|
|
252
|
+
with Path(args.source).open("rb") as stream:
|
|
253
|
+
source = stream.read(MAX_SOURCE + 1).decode("utf-8")
|
|
254
|
+
except (ValueError, OSError, RecursionError) as exc:
|
|
255
|
+
raise AutError("investigation: provide UTF-8 source and valid JSON parameters") from exc
|
|
256
|
+
value = run(source, params, specs)
|
|
257
|
+
if args.save:
|
|
258
|
+
record = {"contract": CONTRACT, "source": source, "params": params,
|
|
259
|
+
"scopes": value["scopes"], "result_sha256": digest(value),
|
|
260
|
+
"reads": value["reads"], "aut_version": value["aut_version"],
|
|
261
|
+
"runtime_version": value["runtime_version"]}
|
|
262
|
+
value["run_id"] = Store(root).put("investigation_runs", record)
|
|
263
|
+
return value
|
|
@@ -7,6 +7,7 @@ import os
|
|
|
7
7
|
import re
|
|
8
8
|
from collections.abc import Callable
|
|
9
9
|
from functools import partial, wraps
|
|
10
|
+
from threading import BoundedSemaphore
|
|
10
11
|
from typing import Any
|
|
11
12
|
|
|
12
13
|
import anyio
|
|
@@ -51,7 +52,8 @@ def _tool(method: Callable[..., dict[str, Any]]) -> Callable[..., Any]:
|
|
|
51
52
|
return call
|
|
52
53
|
|
|
53
54
|
|
|
54
|
-
def serve(evidence: Evidence, transport: str, port: int
|
|
55
|
+
def serve(evidence: Evidence, transport: str, port: int,
|
|
56
|
+
code_scopes: list[dict[str, Any]] | None = None) -> None:
|
|
55
57
|
token = os.environ.get("AUT_EVIDENCE_TOKEN", "")
|
|
56
58
|
if transport == "http" and not re.fullmatch(r"[A-Za-z0-9_-]{32,128}", token):
|
|
57
59
|
raise AutError("evidence: HTTP requires a URL-safe AUT_EVIDENCE_TOKEN of 32..128 characters")
|
|
@@ -68,6 +70,30 @@ def serve(evidence: Evidence, transport: str, port: int) -> None:
|
|
|
68
70
|
for operation in operations:
|
|
69
71
|
server.add_tool(_tool(operation), annotations=ToolAnnotations(
|
|
70
72
|
readOnlyHint=True, destructiveHint=False, openWorldHint=False))
|
|
73
|
+
if code_scopes is not None:
|
|
74
|
+
from .investigation import catalog, pin_scopes, require_runtime, run
|
|
75
|
+
|
|
76
|
+
require_runtime()
|
|
77
|
+
slots = BoundedSemaphore(2)
|
|
78
|
+
scopes = code_scopes
|
|
79
|
+
pinned = pin_scopes(scopes, evidence)
|
|
80
|
+
|
|
81
|
+
def code_catalog() -> dict[str, Any]:
|
|
82
|
+
"""List the evidence operations and arguments accepted by run_code."""
|
|
83
|
+
return {**catalog(), "scopes": pinned}
|
|
84
|
+
|
|
85
|
+
def run_code(source: str, params: dict[str, Any] | None = None) -> dict[str, Any]:
|
|
86
|
+
"""Run scoped Python using read(operation, scope='1', lead=None, **arguments)."""
|
|
87
|
+
if not slots.acquire(blocking=False):
|
|
88
|
+
raise AutError("investigation: both execution slots are busy")
|
|
89
|
+
try:
|
|
90
|
+
return run(source, params if params is not None else {}, scopes, pinned)
|
|
91
|
+
finally:
|
|
92
|
+
slots.release()
|
|
93
|
+
|
|
94
|
+
for code_tool in (code_catalog, run_code):
|
|
95
|
+
server.add_tool(_tool(code_tool), annotations=ToolAnnotations(
|
|
96
|
+
readOnlyHint=True, destructiveHint=False, openWorldHint=False))
|
|
71
97
|
if transport == "stdio":
|
|
72
98
|
server.run()
|
|
73
99
|
else:
|
|
@@ -0,0 +1,256 @@
|
|
|
1
|
+
"""Run evidence recipes against an explicitly selected set of scopes."""
|
|
2
|
+
|
|
3
|
+
from __future__ import annotations
|
|
4
|
+
|
|
5
|
+
import inspect
|
|
6
|
+
import json
|
|
7
|
+
import os
|
|
8
|
+
import subprocess
|
|
9
|
+
import sys
|
|
10
|
+
from datetime import datetime, timezone
|
|
11
|
+
from importlib.metadata import version
|
|
12
|
+
from importlib.util import find_spec
|
|
13
|
+
from pathlib import Path
|
|
14
|
+
from typing import Any, Literal, get_args, get_origin, get_type_hints
|
|
15
|
+
|
|
16
|
+
from .. import __version__
|
|
17
|
+
from ..config import AutError
|
|
18
|
+
from .access import Evidence
|
|
19
|
+
from .store import Store, canonical, digest
|
|
20
|
+
|
|
21
|
+
CONTRACT = 1
|
|
22
|
+
MAX_SOURCE = 16_384
|
|
23
|
+
MAX_PARAMS = 8_192
|
|
24
|
+
MAX_SCOPES = 8
|
|
25
|
+
MAX_CALLS = 48
|
|
26
|
+
MAX_DATA = 2 * 1024 * 1024
|
|
27
|
+
MAX_OUTPUT = 65_536
|
|
28
|
+
TIMEOUT = 12
|
|
29
|
+
OPERATIONS = ("describe", "search", "session", "event", "compare", "commits", "case",
|
|
30
|
+
"ordinary_work")
|
|
31
|
+
WORKER = "agent_usage_tracker.improve.investigation"
|
|
32
|
+
|
|
33
|
+
|
|
34
|
+
def require_runtime() -> None:
|
|
35
|
+
if find_spec("pydantic_monty") is None:
|
|
36
|
+
raise AutError("investigation: install agent-usage-tracker[code] to run recipes")
|
|
37
|
+
|
|
38
|
+
|
|
39
|
+
def _matches(value: Any, annotation: Any) -> bool:
|
|
40
|
+
if get_origin(annotation) is Literal:
|
|
41
|
+
return any(type(value) is type(choice) and value == choice
|
|
42
|
+
for choice in get_args(annotation))
|
|
43
|
+
if options := get_args(annotation):
|
|
44
|
+
return any(_matches(value, option) for option in options)
|
|
45
|
+
return type(value) is annotation
|
|
46
|
+
|
|
47
|
+
|
|
48
|
+
def catalog() -> dict[str, Any]:
|
|
49
|
+
operations = {}
|
|
50
|
+
for name in OPERATIONS:
|
|
51
|
+
method = getattr(Evidence, name)
|
|
52
|
+
hints = get_type_hints(method)
|
|
53
|
+
fields = {}
|
|
54
|
+
for param in list(inspect.signature(method).parameters.values())[1:]:
|
|
55
|
+
field = {"type": inspect.formatannotation(hints[param.name])}
|
|
56
|
+
if param.default is not inspect.Parameter.empty:
|
|
57
|
+
field["default"] = param.default
|
|
58
|
+
fields[param.name] = field
|
|
59
|
+
operations[name] = {"description": method.__doc__, "arguments": fields}
|
|
60
|
+
return {"contract": CONTRACT, "operations": operations,
|
|
61
|
+
"interface": "read(operation, scope='1', lead=None, **arguments) returns receipt and data",
|
|
62
|
+
"limits": {"source_bytes": MAX_SOURCE, "parameter_bytes": MAX_PARAMS,
|
|
63
|
+
"calls": MAX_CALLS, "seconds": TIMEOUT, "output_bytes": MAX_OUTPUT}}
|
|
64
|
+
|
|
65
|
+
|
|
66
|
+
def scope_specs(root: Path, content: str, audit: str | None, bundle: Path | None,
|
|
67
|
+
audits: list[str], bundles: list[Path]) -> list[dict[str, Any]]:
|
|
68
|
+
sources = [(audit, bundle), *((key, None) for key in audits),
|
|
69
|
+
*((None, path) for path in bundles)]
|
|
70
|
+
if len(sources) > MAX_SCOPES:
|
|
71
|
+
raise AutError(f"investigation: select at most {MAX_SCOPES} evidence scopes")
|
|
72
|
+
specs = [{"audit": key, "bundle": str(path.expanduser().resolve()) if path else None,
|
|
73
|
+
"root": str(root.expanduser().resolve()) if key else None, "content": content}
|
|
74
|
+
for key, path in sources]
|
|
75
|
+
if len({(row["audit"], row["bundle"]) for row in specs}) != len(specs):
|
|
76
|
+
raise AutError("investigation: select distinct evidence scopes")
|
|
77
|
+
return specs
|
|
78
|
+
|
|
79
|
+
|
|
80
|
+
def _bounded(value: Any, limit: int, label: str) -> bytes:
|
|
81
|
+
try:
|
|
82
|
+
data = canonical(value).encode("utf-8")
|
|
83
|
+
except (TypeError, ValueError, RecursionError) as exc:
|
|
84
|
+
raise AutError(f"investigation: {label} must be finite JSON") from exc
|
|
85
|
+
if len(data) > limit:
|
|
86
|
+
raise AutError(f"investigation: {label} exceeds {limit} bytes")
|
|
87
|
+
return data
|
|
88
|
+
|
|
89
|
+
|
|
90
|
+
def run(source: str, params: dict[str, Any], specs: list[dict[str, Any]],
|
|
91
|
+
expected_scopes: dict[str, Any] | None = None) -> dict[str, Any]:
|
|
92
|
+
require_runtime()
|
|
93
|
+
try:
|
|
94
|
+
source_size = len(source.encode("utf-8"))
|
|
95
|
+
except UnicodeError as exc:
|
|
96
|
+
raise AutError("investigation: source must be UTF-8") from exc
|
|
97
|
+
if not source.strip() or source_size > MAX_SOURCE:
|
|
98
|
+
raise AutError(f"investigation: provide 1..{MAX_SOURCE} bytes of Python source")
|
|
99
|
+
if not isinstance(params, dict):
|
|
100
|
+
raise AutError("investigation: parameters must be a JSON object")
|
|
101
|
+
_bounded(params, MAX_PARAMS, "parameters")
|
|
102
|
+
payload = canonical({"source": source, "params": params, "specs": specs,
|
|
103
|
+
"expected_scopes": expected_scopes}).encode("utf-8")
|
|
104
|
+
env = {name: os.environ[name] for name in ("PATH", "SYSTEMROOT", "WINDIR", "TMP", "TEMP", "TMPDIR")
|
|
105
|
+
if name in os.environ}
|
|
106
|
+
try:
|
|
107
|
+
process = subprocess.run([sys.executable, "-I", "-m", WORKER], input=payload,
|
|
108
|
+
capture_output=True, timeout=TIMEOUT, env=env, check=False)
|
|
109
|
+
except subprocess.TimeoutExpired as exc:
|
|
110
|
+
raise AutError("investigation: execution exceeded the wall-clock limit") from exc
|
|
111
|
+
except OSError as exc:
|
|
112
|
+
raise AutError("investigation: could not start the local runtime") from exc
|
|
113
|
+
if process.returncode:
|
|
114
|
+
raise AutError("investigation: local runtime failed")
|
|
115
|
+
try:
|
|
116
|
+
response = json.loads(process.stdout)
|
|
117
|
+
except (ValueError, RecursionError) as exc:
|
|
118
|
+
raise AutError("investigation: local runtime returned an invalid response") from exc
|
|
119
|
+
if "error" in response:
|
|
120
|
+
raise AutError(response["error"])
|
|
121
|
+
return dict(response)
|
|
122
|
+
|
|
123
|
+
|
|
124
|
+
def _references(operation: str, value: dict[str, Any]) -> list[dict[str, str]]:
|
|
125
|
+
references: dict[tuple[str, str], None] = {}
|
|
126
|
+
|
|
127
|
+
def include(session: Any, *events: Any) -> None:
|
|
128
|
+
if isinstance(session, str):
|
|
129
|
+
for event in events:
|
|
130
|
+
if isinstance(event, str):
|
|
131
|
+
references[session, event] = None
|
|
132
|
+
|
|
133
|
+
if operation == "compare":
|
|
134
|
+
difference = value["first_tool_difference"] or {}
|
|
135
|
+
for side in ("baseline", "candidate"):
|
|
136
|
+
row = value[side]
|
|
137
|
+
tool = difference.get(side) or {}
|
|
138
|
+
include(row["id"], *row["error_event_ids"], tool.get("call_event_id"),
|
|
139
|
+
tool.get("result_event_id"))
|
|
140
|
+
elif operation == "event":
|
|
141
|
+
include(value["session_id"], value["id"])
|
|
142
|
+
elif operation == "case":
|
|
143
|
+
include(value["session_id"], *value["event_ids"])
|
|
144
|
+
else:
|
|
145
|
+
session = value.get("session", {}).get("id")
|
|
146
|
+
for item in value.get("items", []):
|
|
147
|
+
key = item.get("session_id") or item.get("session_snapshot") or session
|
|
148
|
+
include(key, item.get("event_id") or item.get("id"), item.get("call_event_id"),
|
|
149
|
+
item.get("result_event_id"))
|
|
150
|
+
return [{"session_id": session, "event_id": event} for session, event in references]
|
|
151
|
+
|
|
152
|
+
|
|
153
|
+
def _open_scope(spec: dict[str, Any]) -> Evidence:
|
|
154
|
+
store = Store(Path(spec["root"]), read_only=True) if spec["root"] else None
|
|
155
|
+
bundle = Path(spec["bundle"]) if spec["bundle"] else None
|
|
156
|
+
return Evidence(spec["content"], store, spec["audit"], bundle)
|
|
157
|
+
|
|
158
|
+
|
|
159
|
+
def _describe_scope(evidence: Evidence) -> dict[str, Any]:
|
|
160
|
+
return {**evidence.describe(), "bundle": digest(evidence.bundled) if evidence.bundled else None}
|
|
161
|
+
|
|
162
|
+
|
|
163
|
+
def pin_scopes(specs: list[dict[str, Any]], primary: Evidence) -> dict[str, Any]:
|
|
164
|
+
scopes = {"1": _describe_scope(primary)}
|
|
165
|
+
for i, spec in enumerate(specs[1:], 2):
|
|
166
|
+
scopes[str(i)] = _describe_scope(_open_scope(spec))
|
|
167
|
+
return scopes
|
|
168
|
+
|
|
169
|
+
|
|
170
|
+
def _execute(source: str, params: dict[str, Any], specs: list[dict[str, Any]],
|
|
171
|
+
expected_scopes: dict[str, Any] | None) -> dict[str, Any]:
|
|
172
|
+
from pydantic_monty import Monty, MontyError, MontyRuntimeError, MontySyntaxError
|
|
173
|
+
|
|
174
|
+
sources = {str(i): _open_scope(spec) for i, spec in enumerate(specs, 1)}
|
|
175
|
+
scopes = {alias: _describe_scope(evidence) for alias, evidence in sources.items()}
|
|
176
|
+
if expected_scopes is not None and scopes != expected_scopes:
|
|
177
|
+
raise AutError("investigation: evidence scope or content grant changed")
|
|
178
|
+
|
|
179
|
+
reads: list[dict[str, Any]] = []
|
|
180
|
+
printed: list[str] = []
|
|
181
|
+
calls = data_bytes = print_bytes = 0
|
|
182
|
+
|
|
183
|
+
def capture(stream: str, message: str) -> None:
|
|
184
|
+
nonlocal print_bytes
|
|
185
|
+
print_bytes += len(message.encode())
|
|
186
|
+
if print_bytes > MAX_OUTPUT:
|
|
187
|
+
raise AutError("investigation: printed output exceeds the output budget")
|
|
188
|
+
printed.append(message)
|
|
189
|
+
|
|
190
|
+
def read(operation: str, *, scope: str = "1", lead: int | None = None,
|
|
191
|
+
**arguments: Any) -> dict[str, Any]:
|
|
192
|
+
nonlocal calls, data_bytes
|
|
193
|
+
calls += 1
|
|
194
|
+
if calls > MAX_CALLS:
|
|
195
|
+
raise AutError("investigation: evidence call budget exhausted")
|
|
196
|
+
if not isinstance(operation, str) or operation not in OPERATIONS:
|
|
197
|
+
raise AutError("investigation: unknown evidence operation")
|
|
198
|
+
if not isinstance(scope, str) or scope not in sources:
|
|
199
|
+
raise AutError("investigation: unknown evidence scope")
|
|
200
|
+
if lead is not None and (type(lead) is not int or not 1 <= lead <= len(reads)):
|
|
201
|
+
raise AutError("investigation: lead must reference an earlier receipt")
|
|
202
|
+
_bounded(arguments, MAX_PARAMS, "operation arguments")
|
|
203
|
+
method = getattr(sources[scope], operation)
|
|
204
|
+
hints = get_type_hints(method)
|
|
205
|
+
try:
|
|
206
|
+
inspect.signature(method).bind(**arguments)
|
|
207
|
+
except TypeError as exc:
|
|
208
|
+
raise AutError("investigation: invalid argument names for evidence operation") from exc
|
|
209
|
+
if any(not _matches(value, hints[name]) for name, value in arguments.items()):
|
|
210
|
+
raise AutError("investigation: invalid argument types for evidence operation")
|
|
211
|
+
data = method(**arguments)
|
|
212
|
+
encoded = _bounded(data, MAX_DATA - data_bytes, "evidence data")
|
|
213
|
+
data_bytes += len(encoded)
|
|
214
|
+
receipt = len(reads) + 1
|
|
215
|
+
reads.append({"receipt": receipt, "lead": lead, "scope": scope, "operation": operation,
|
|
216
|
+
"arguments_sha256": digest(arguments), "data_sha256": digest(data),
|
|
217
|
+
"has_more": any(data.get(key) is not None for key in ("next_cursor", "next_offset")),
|
|
218
|
+
"references": _references(operation, data)})
|
|
219
|
+
return {"receipt": receipt, "data": data}
|
|
220
|
+
|
|
221
|
+
try:
|
|
222
|
+
with Monty(max_processes=1, request_timeout=TIMEOUT) as pool, pool.checkout(
|
|
223
|
+
limits={"max_feed_duration_secs": 3, "max_memory": 32 * 1024 * 1024,
|
|
224
|
+
"max_recursion_depth": 128, "max_suspensions": 256, "max_total_sleep_secs": 0},
|
|
225
|
+
os_policy={"datetime": datetime(2000, 1, 1, tzinfo=timezone.utc),
|
|
226
|
+
"random_start": {"seed": 0}},
|
|
227
|
+
) as session:
|
|
228
|
+
result = session.feed_run(source, inputs={"params": params, "scopes": scopes},
|
|
229
|
+
external_lookup={"read": read}, print_callback=capture)
|
|
230
|
+
except MontySyntaxError as exc:
|
|
231
|
+
raise AutError(f"investigation: invalid Python syntax: {exc.display('type-msg')[:300]}") from exc
|
|
232
|
+
except MontyRuntimeError as exc:
|
|
233
|
+
error_type = type(exc.exception()).__name__
|
|
234
|
+
raise AutError(f"investigation: program stopped with {error_type}") from exc
|
|
235
|
+
except MontyError as exc:
|
|
236
|
+
raise AutError("investigation: sandbox could not complete the program") from exc
|
|
237
|
+
output = {"result": result, "stdout": "".join(printed)}
|
|
238
|
+
_bounded(output, MAX_OUTPUT, "program output")
|
|
239
|
+
_bounded(reads, MAX_DATA, "read receipts")
|
|
240
|
+
return {**output, "contract": CONTRACT, "aut_version": __version__,
|
|
241
|
+
"runtime_version": version("pydantic-monty"), "scopes": scopes, "reads": reads,
|
|
242
|
+
"calls": calls, "read_errors": calls - len(reads), "evidence_bytes": data_bytes,
|
|
243
|
+
"source_sha256": digest(source), "parameters_sha256": digest(params)}
|
|
244
|
+
|
|
245
|
+
|
|
246
|
+
def worker() -> None:
|
|
247
|
+
try:
|
|
248
|
+
request = json.load(sys.stdin.buffer)
|
|
249
|
+
value = _execute(**request)
|
|
250
|
+
except AutError as exc:
|
|
251
|
+
value = {"error": str(exc)}
|
|
252
|
+
sys.stdout.buffer.write(canonical(value).encode("utf-8") + b"\n")
|
|
253
|
+
|
|
254
|
+
|
|
255
|
+
if __name__ == "__main__":
|
|
256
|
+
worker()
|