-
Notifications
You must be signed in to change notification settings - Fork 11
All issues
Issue creation is restricted in this repository
Issues
is:issue state:open
is:issue state:open
Search results
An exception from an adapter's live() aborts the whole run — evaluate_candidate is unprotected while the optimizer call beside it is
bugSomething isn't workingSomething isn't workinghonestyHonesty/consistency of reported results (brand-critical)Honesty/consistency of reported results (brand-critical)priority-p1High impactHigh impactStatus: Open.#286 In skillberry-ai/cap-evolve;The optimizer can edit the harness that scores it, and the gain is recorded as a capability improvement
bugSomething isn't workingSomething isn't workingenhancementNew feature or requestNew feature or requesthonestyHonesty/consistency of reported results (brand-critical)Honesty/consistency of reported results (brand-critical)priority-p0Flagship / brand-critical — do firstFlagship / brand-critical — do firstStatus: Open.#276 In skillberry-ai/cap-evolve;Dashboard backend has no SPA fallback — a hard load or refresh of any deep route 404s
bugSomething isn't workingSomething isn't workingdashboardDashboard backend/frontendDashboard backend/frontenddxDeveloper/onboarding experienceDeveloper/onboarding experiencepriority-p2NormalNormalreview-2026-07Competitive review batch (cap-evolve vs Arbor), Jul 2026Competitive review batch (cap-evolve vs Arbor), Jul 2026Status: Open.#254 In skillberry-ai/cap-evolve;A relative optimizer_instructions_file resolves differently in check and run — run silently falls back to a generic template
algorithmOptimization algorithms: GEPA / SkillOpt / hill-climbOptimization algorithms: GEPA / SkillOpt / hill-climbbugSomething isn't workingSomething isn't workingdxDeveloper/onboarding experienceDeveloper/onboarding experiencehonestyHonesty/consistency of reported results (brand-critical)Honesty/consistency of reported results (brand-critical)observabilityLive run visibility, logging, tracingLive run visibility, logging, tracingpriority-p0Flagship / brand-critical — do firstFlagship / brand-critical — do firstpriority-p1High impactHigh impactreview-2026-07Competitive review batch (cap-evolve vs Arbor), Jul 2026Competitive review batch (cap-evolve vs Arbor), Jul 2026Status: Open.#252 In skillberry-ai/cap-evolve;Optimizer stderr is discarded on success by three separate capture_output layers — agent CLI diagnostics never reach the user
bugSomething isn't workingSomething isn't workingdxDeveloper/onboarding experienceDeveloper/onboarding experiencehonestyHonesty/consistency of reported results (brand-critical)Honesty/consistency of reported results (brand-critical)observabilityLive run visibility, logging, tracingLive run visibility, logging, tracingpriority-p1High impactHigh impactreview-2026-07Competitive review batch (cap-evolve vs Arbor), Jul 2026Competitive review batch (cap-evolve vs Arbor), Jul 2026tech-debtDead code, duplication, refactorsDead code, duplication, refactorsStatus: Open.#251 In skillberry-ai/cap-evolve;test_core_passes_on_a_correctly_activated_venv asserts a macOS-only symlink property and fails on ubuntu-latest
bugSomething isn't workingSomething isn't workingdxDeveloper/onboarding experienceDeveloper/onboarding experiencepriority-p1High impactHigh impactreview-2026-07Competitive review batch (cap-evolve vs Arbor), Jul 2026Competitive review batch (cap-evolve vs Arbor), Jul 2026tech-debtDead code, duplication, refactorsDead code, duplication, refactorsStatus: Open.#247 In skillberry-ai/cap-evolve;run_step swallows TamperError: a bare except Exception around the optimizer call downgrades honesty aborts to transient errors
algorithmOptimization algorithms: GEPA / SkillOpt / hill-climbOptimization algorithms: GEPA / SkillOpt / hill-climbbugSomething isn't workingSomething isn't workinghonestyHonesty/consistency of reported results (brand-critical)Honesty/consistency of reported results (brand-critical)priority-p0Flagship / brand-critical — do firstFlagship / brand-critical — do firstpriority-p1High impactHigh impactreview-2026-07Competitive review batch (cap-evolve vs Arbor), Jul 2026Competitive review batch (cap-evolve vs Arbor), Jul 2026Status: Open.#239 In skillberry-ai/cap-evolve;Iterations have no stable identity: 4 bugs from consumers re-deriving iteration-ness from event-kind strings
algorithmOptimization algorithms: GEPA / SkillOpt / hill-climbOptimization algorithms: GEPA / SkillOpt / hill-climbblocked-by-dependencyWaits on another issue's PRWaits on another issue's PRbugSomething isn't workingSomething isn't workingobservabilityLive run visibility, logging, tracingLive run visibility, logging, tracingpriority-p1High impactHigh impactreview-2026-07Competitive review batch (cap-evolve vs Arbor), Jul 2026Competitive review batch (cap-evolve vs Arbor), Jul 2026tech-debtDead code, duplication, refactorsDead code, duplication, refactorsStatus: Open.#224 In skillberry-ai/cap-evolve;cap-evolve run prints two JSON objects to stdout — the machine-readable contract has never been parseable
bugSomething isn't workingSomething isn't workingdxDeveloper/onboarding experienceDeveloper/onboarding experienceobservabilityLive run visibility, logging, tracingLive run visibility, logging, tracingpriority-p1High impactHigh impactreview-2026-07Competitive review batch (cap-evolve vs Arbor), Jul 2026Competitive review batch (cap-evolve vs Arbor), Jul 2026tech-debtDead code, duplication, refactorsDead code, duplication, refactorswave-1Epic 127 execution wave 1Epic 127 execution wave 1Status: Open.#217 In skillberry-ai/cap-evolve;GEPA runs never write a run-level JOURNAL.md — _reconcile_journal is only called from run_step, which GEPA bypasses
algorithmOptimization algorithms: GEPA / SkillOpt / hill-climbOptimization algorithms: GEPA / SkillOpt / hill-climbbugSomething isn't workingSomething isn't workingdashboardDashboard backend/frontendDashboard backend/frontendhonestyHonesty/consistency of reported results (brand-critical)Honesty/consistency of reported results (brand-critical)observabilityLive run visibility, logging, tracingLive run visibility, logging, tracingpriority-p1High impactHigh impactpriority-p2NormalNormalreview-2026-07Competitive review batch (cap-evolve vs Arbor), Jul 2026Competitive review batch (cap-evolve vs Arbor), Jul 2026tech-debtDead code, duplication, refactorsDead code, duplication, refactorsStatus: Open.#216 In skillberry-ai/cap-evolve;CI never runs ./install.sh — the documented host-agnostic install path is untested (hid a total optimizer failure)
bugSomething isn't workingSomething isn't workingdocumentationImprovements or additions to documentationImprovements or additions to documentationdxDeveloper/onboarding experienceDeveloper/onboarding experienceenhancementNew feature or requestNew feature or requesthonestyHonesty/consistency of reported results (brand-critical)Honesty/consistency of reported results (brand-critical)priority-p1High impactHigh impactreview-2026-07Competitive review batch (cap-evolve vs Arbor), Jul 2026Competitive review batch (cap-evolve vs Arbor), Jul 2026tech-debtDead code, duplication, refactorsDead code, duplication, refactorsStatus: Open.#208 In skillberry-ai/cap-evolve;CI never runs the 13 frontend test files — vitest is configured but not invoked
bugSomething isn't workingSomething isn't workingdashboardDashboard backend/frontendDashboard backend/frontenddxDeveloper/onboarding experienceDeveloper/onboarding experiencejavascriptPull requests that update javascript codePull requests that update javascript codepriority-p1High impactHigh impactreview-2026-07Competitive review batch (cap-evolve vs Arbor), Jul 2026Competitive review batch (cap-evolve vs Arbor), Jul 2026tech-debtDead code, duplication, refactorsDead code, duplication, refactorsStatus: Open.#207 In skillberry-ai/cap-evolve;