Metrics
Plan
| Indicator | Definition | Value/unit | n | Note |
|---|---|---|---|---|
| conversation_to_intent | First intent addition minus recorded conversation start | 1.8508 hours | 3 | |
| intent_survival | Accepted / decided intents, excluding open | 66.6667 percent | 3 | |
| intent_changes_after_spec | Distinct intent touches after first spec addition | 1 commits/change | 2 |
Design
| Indicator | Definition | Value/unit | n | Note |
|---|---|---|---|---|
| intent_to_spec | First spec addition minus first intent addition | 0.2211 hours | 2 | |
| spec_rework | Distinct spec touches after first plan addition | 0.5 commits/change | 2 |
Build
| Indicator | Definition | Value/unit | n | Note |
|---|---|---|---|---|
| first_pass_merge | Merged changes with one attempt / merged changes with attempts | 0.0 percent | 1 | |
| plan_to_final_merge | Final linked merge minus first plan addition (approval proxy) | 12.8892 hours | 1 | |
| rework_cycles | Attempts minus one plus distinct fix jobs | 19 cycles/change | 3 | May double count review passes and inflate planned increments; source diagnostics: 13 |
| plan_match | True explicit plan_match / recorded merged verdicts | 100.0 percent | 1 |
CLAUDE.md
| Indicator | Definition | Value/unit | n | Note |
|---|---|---|---|---|
| repeated_mistakes | Sum of distinct occurrences beyond the first in each recorded class | 4 repeat occurrences | 5 | Coverage is recorded mistakes only |
Parallel
| Indicator | Definition | Value/unit | n | Note |
|---|---|---|---|---|
| peak_jobs | Peak simultaneous active job intervals | 4 jobs | 110 | ; source diagnostics: 13 |
| mean_jobs | Time-weighted jobs over union of active intervals | 1.1157 jobs | 110 | ; source diagnostics: 13 |
| review_quality_held | Review quality during overlap | unavailable boolean | 0 | not shown: structured quality evidence is not shown |
| weekly_merges | Distinct linked merged PRs per Pacific/Auckland ISO week | 18 PRs/week | 18 | |
| weekly_completed_changes | Completed changes grouped by final merge week | 1 changes/week | 1 | ; source diagnostics: 13 |
| weekly_rework | Completed changes grouped by final merge week | 2 cycles/change | 1 | ; source diagnostics: 13 |
Test
| Indicator | Definition | Value/unit | n | Note |
|---|---|---|---|---|
| first_pass_ci | PRs whose earliest check in every app/name group succeeded | 94.4444 percent | 18 | Final stored PR head proxy; original implementation head is not shown; source diagnostics: 13 |
| review_time | First submitted non-pending review minus PR creation | 0.1417 minutes | 4 | Submitted-review latency proxy; supervisor identity and effort are not recorded |
| change_failure_rate | Completed changes reverted or fixed strictly after final merge | 0.0 percent | 1 | Conservative change-level proxy; pre-completion fixes excluded; source diagnostics: 13 |
Evals
| Indicator | Definition | Value/unit | n | Note |
|---|---|---|---|---|
| eval_pass_rate | True eval results / boolean results | unavailable percent | 0 | not shown: missing optional evidence |
| incident_to_eval | First result timestamp per eval minus incident_at | unavailable hours | 0 | not shown: missing optional evidence |
| regressions_ci | Structured regression detection | unavailable events | 0 | not shown: no structured source yet |
| regressions_after_merge | Structured regression detection | unavailable events | 0 | not shown: no structured source yet |
| regressions_ci_share | Structured regression detection | unavailable percent | 0 | not shown: no structured source yet |
Deploy
| Indicator | Definition | Value/unit | n | Note |
|---|---|---|---|---|
| first_review | Earliest submitted review or root comment minus PR creation | 0.1417 minutes | 4 | Same-supervisor-run creation and review can yield minutes by construction; not independent response speed |
| human_free_resolution | Comments resolved without human branch edits | unavailable percent | 0 | not shown: REST comments do not establish resolution or branch-editor identity |
| agent_resolution_proxy | Merged commented PRs / PRs with root review comments | no data (n=0) percent | 0 | Assumes merged comments were agent-resolved; human-free editing is not verified; no data (n=0) |
| defects_caught | Unique defects caught before merge | unavailable defects | 0 | not shown: review findings lack structured severity and identity |
| defects_escaped | Unique defects escaped after merge | unavailable defects | 0 | not shown: fix jobs do not establish unique defects |
| pre_merge_comments | Root review comments submitted before merge (activity proxy) | 0 comments | 18 | |
| post_merge_fixes | Attributed fix jobs starting after final change merge (activity proxy) | 0 jobs | 1 | ; source diagnostics: 13 |
Gates
| Indicator | Definition | Value/unit | n | Note |
|---|---|---|---|---|
| closed_gate_wait | Closed needs-byron issue lifetime | no data (n=0) hours | 0 | Issue lifetime proxies label-addition and gate wait; no data (n=0) |
| open_gate_age | Open needs-byron issue age censored at HEAD time | no data (n=0) hours | 0 | HEAD is a deterministic censor, not an API observation cutoff; no data (n=0) |
CI/CD
| Indicator | Definition | Value/unit | n | Note |
|---|---|---|---|---|
| unpaged_failure_proxy | Terminal failed CI runs without a run URL in a gate / failed CI runs | 100.0 percent | 10 | Absence of a linked gate does not establish triage |
| human_free_triage | Pipeline failures triaged without human intervention | unavailable percent | 0 | not shown: absence of paging does not establish triage |
| deployment_frequency | Successful Pages deploy jobs / observed calendar days, including zero days | 35 deployments/day | 4 | Successful Pages runs imply a successful deploy job under pages.yml's build -> deploy dependency; run updated_at proxies deploy completed_at |
| deployment_lead_time | First successful deploy of exact merge SHA minus merge | 0.0191 hours | 15 | Successful Pages runs imply a successful deploy job under pages.yml's build -> deploy dependency; run updated_at proxies deploy completed_at |
| dora_change_failure_rate | Change-level failure proxy, same cohort as Test change failure rate | 0.0 percent | 1 | ; source diagnostics: 13 |
| restore_time | First failed deploy to next successful deploy; consecutive failures form one episode | 0.3675 hours | 1 | Successful Pages runs imply a successful deploy job under pages.yml's build -> deploy dependency; run updated_at proxies deploy completed_at; pipeline recovery, not service restoration; open episodes excluded: 0 |
Maintain
| Indicator | Definition | Value/unit | n | Note |
|---|---|---|---|---|
| breach_to_intent | Band breach to first intent addition | unavailable hours | 0 | not shown: no structured source yet |
| findings_to_fixes | Merged finding changes / decided finding changes | no data (n=0) percent | 0 | Pending findings excluded: 0; no data (n=0) |
| repeat_incidents | Later finding changes sharing explicit incident identifiers | unavailable repeat incidents | 0 | not shown: no structured source yet |
Cost
| Indicator | Definition | Value/unit | n | Note |
|---|---|---|---|---|
| input_tokens | Reported ledger usage; resumed-thread context may overlap | 283788280.0 tokens | 110 | Total tokens = input + output; cache and reasoning are components; source diagnostics: 13 |
| cached_input_tokens | Reported ledger usage; resumed-thread context may overlap | 271442560.0 tokens | 110 | Total tokens = input + output; cache and reasoning are components; source diagnostics: 13 |
| cache_write_input_tokens | Reported ledger usage; resumed-thread context may overlap | 0.0 tokens | 110 | Total tokens = input + output; cache and reasoning are components; source diagnostics: 13 |
| output_tokens | Reported ledger usage; resumed-thread context may overlap | 2108552.0 tokens | 110 | Total tokens = input + output; cache and reasoning are components; source diagnostics: 13 |
| reasoning_output_tokens | Reported ledger usage; resumed-thread context may overlap | 428338.0 tokens | 110 | Total tokens = input + output; cache and reasoning are components; source diagnostics: 13 |
| total_tokens | Reported ledger usage; resumed-thread context may overlap | 285896832.0 tokens | 110 | Total tokens = input + output; cache and reasoning are components; source diagnostics: 13 |
| weekly_window | Latest observed 10080-minute window per NZ day | 70.0 percent | 102 | Snapshot, not consumption; reset drops are not negative usage; source diagnostics: 13 |
| supervisor_cost_usd | Mean per reported run, including failed and zero-valued runs | 2.339138 USD | 79 | |
| supervisor_cost_usd_runs | Reported supervisor values keyed by timestamp and source line | 184.791887 USD | 79 | |
| supervisor_turns | Mean per reported run, including failed and zero-valued runs | 15.7342 turns | 79 | |
| supervisor_turns_runs | Reported supervisor values keyed by timestamp and source line | 1243.0 turns | 79 | |
| wall_time | Final linked merge minus change started, not summed job time | 23.425 hours/change | 1 | |
| wall_time_by_change | Elapsed wall time per completed change | 23.425 hours/change | 1 | |
| job_time | Sum of valid attributed job durations, distinct from elapsed wall time | 27170.0 job-seconds | 100 | ; source diagnostics: 13 |