| name | runtime-log-repair |
| description | Use when AgentCanon runtime dashboard evidence should be turned into owner-routed repair work, including dashboard next actions, repair failing hook evidence, hook entries status=fail, missing actual wave rows, workflow attribution gaps, consulted source URLs, reference missing URLs, AGENT_RUNTIME_DASHBOARD_WAVE_MISSING_ACTUAL, AGENT_RUNTIME_DASHBOARD_HOOK_WORKFLOW_MISSING, or AGENT_RUNTIME_DASHBOARD_REFERENCE_MISSING_URLS. |
Runtime Log Repair
Tool Commands
Use the command packet before applying this skill's workflow:
python3 tools/agent_tools/skill_tool_commands.py show --skill runtime-log-repair --format text
Execute the required and task-matching conditional commands that the packet prints.
- Read
agents/skills/runtime-log-repair.md.
- Start from
$agent-log-analysis dashboard artifacts:
reports/agent-runtime-dashboard/agent-log-analysis-api.json and
reports/agent-runtime-dashboard/agent-log-analysis-compact.md. If they are
missing or stale for the request, run $agent-log-analysis first.
- Do not read raw JSONL broadly for normal repair routing; raw event drilldown
stays with
$agent-log-analysis tool development, schema debugging, or an
API-named drilldown path.
- Build a Runtime Log Repair Packet with
repair_class,
dashboard_evidence, owner_surface, repair_route, required_input,
non_goals, and closeout_gate before editing or launching repair work.
- Route repairs to owners: eval gaps to
$agent-eval-accumulation, durable
artifacts to $result-artifact-writeout, issue candidates to
$issue-finding-report, wave mechanics to $subagent-bootstrap,
prompt/config or selection repair to $task-routing plus the affected owner,
and recurrence learning to $agent-learning.
- Keep boundaries explicit: this skill does not own raw log analysis,
dashboard schema, eval producer loops, artifact placement, durable issue
writing, subagent launch mechanics, prompt/config review, hook
implementation details, or reference extraction internals.
- Verify closeout with the owner-selected gate. Rerun a full dashboard only
when the owner gate needs accumulated post-change evidence.
- If the owner-selected gate fails, add
failing_contract,
observation_level, cause_classification, intent_preservation, and
evidence to the Runtime Log Repair Packet before changing repair
intent, simplifying to pass, reverting, deleting intended behavior/tests,
weakening an oracle, or downscoping validation. Preserve owner intent for
implementation bugs and route oracle/spec, fixture/environment/stale
artifact, unrelated, and approved-design/user-request conflicts to owner
repair, residual, or escalation.