Harden benchmark evidence and execution safety

[rica-v3]

Summary

• add deterministic v2 prompt-compaction evidence targets and schema validation
• move raw telemetry to parent-owned storage with journal reconciliation
• pin provider/Git executables, sanitize PATH, and enforce finite deadlines
• quarantine unsafe workspace entries and harden process cleanup
• replace executable quality inspection with a bounded AST oracle
• add benchmark documentation and security/integration coverage

Verification

• 862 passed, 1 skipped, 143 subtests passed
• credential-free live preflight records reserved_count=0 and telemetry_record_count=0
• no provider credentials were available, so repeated live Before/After measurements remain pending

Notes

A detached-descendant process limitation remains documented as a host-level safety boundary for follow-up hardening.

[Copilot]

Pull request overview

Adds secure, measurable full-sprint benchmarking with prompt compaction, telemetry, bounded model execution, and deterministic quality validation.

Changes:

• Adds prompt-history compaction and telemetry across model runtimes.
• Introduces hardened live A/B benchmark execution and reporting.
• Improves sprint requirement review batching and supporting documentation/tests.

Reviewed changes

Copilot reviewed 41 out of 42 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
workflows/sprints/lifecycle.py	Extracts confirmation and review-checkpoint helpers.
workflows/roles/research.py	Applies prompt-history projection to research decisions.
workflows/repository_ops.py	Hardens benchmark Git execution.
workflows/orchestration/team_service.py	Propagates execution, telemetry, and research policies.
tests/test_sprint_lifecycle.py	Tests requirement checkpoint behavior.
tests/test_prompt_context.py	Tests projection and configuration behavior.
tests/test_orchestration_sprint_execution.py	Tests batched manual-sprint reviews.
tests/test_execution_policy.py	Covers benchmark execution safety.
templates/scaffold/team_runtime.yaml	Adds prompt-context and telemetry defaults.
shared/prompt_context.py	Implements deterministic event projection.
shared/paths.py	Adds telemetry storage paths.
shared/models.py	Adds telemetry and prompt configuration models.
shared/config.py	Parses and validates new configuration.
shared/init.py	Exports telemetry models.
runtime/research_runtime.py	Adds telemetry and external-research policy enforcement.
runtime/internal/intent_parser.py	Instruments parser model calls.
runtime/internal/goal_sourcing.py	Instruments goal-sourcing calls.
runtime/execution_policy.py	Defines benchmark budgets and safety controls.
runtime/benchmark_launcher.py	Gates provider launch on durable registration.
runtime/base_runtime.py	Integrates projection, telemetry, and execution policy.
models.py	Re-exports new public models.
docs/telemetry.md	Documents telemetry operation and schema.
docs/specification.md	Specifies prompt projection and telemetry configuration.
docs/README.md	Links telemetry documentation.
docs/performance_benchmarking.md	Documents the live benchmark workflow.
docs/operations_guide.md	Updates sprint review and prompt-context guidance.
docs/implementation.md	Records prompt-context ownership.
docs/configuration_guide.md	Documents compaction configuration and algorithm.
cli.py	Adds metrics and benchmark commands.
benchmarking/scenario.py	Defines fixtures and bounded AST validation.
benchmarking/runner.py	Runs benchmark arms and retains artifacts.
benchmarking/reporting.py	Produces comparison reports.
benchmarking/models.py	Defines benchmark contracts and validation.
benchmarking/init.py	Exposes benchmark APIs.
adapters/cli/commands.py	Adds CLI parsing and dispatch.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.