Herdr team coordination policy for AI coding agents
93%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
Low
Low-risk findings worth noting
For adopted legacy source, read the original delivered report in full and verify its relevant claims against the supplied source and current criteria. Source adoption supplies no passing test or implementation approval.
Independently test the current full implementation against the admitted criteria and affected behavior. Remain read-only. Inspect actual outcomes, including relevant failure paths; report the commands/checks and evidence. Do not substitute a developer or reviewer summary for your own test result.
Own the full required gate once at the reviewed head. Inspect and bind an independently executed CI artifact's actual command, output and result to that head when available; run missing gates and task-specific acceptance checks. Another worker's summary or a different head cannot substitute. Report the artifact locator and exact head with the evidence. Distinguish contention or baseline failures through a focused reproduction; an unexplained failure remains blocking.
Return the actual full head, one criterion entry per admitted criterion,
checks for the tests responsibility, explicit correction facts and unique
finding identities. Advisories create no repair or follow-up obligation.
Unavailable or uncertain evidence cannot become passing coverage.
Return exactly one VERDICT and one complete WORKFLOW claim that agree. Use
verified only when every criterion is met and actual checks passed. A
blocking outcome names its unresolved blocking findings. Keep disputes,
changed scope/plans and non-convergence explicit for the task's decision owner.
.tessl-plugin
rules
skills
herdr-foreman
classify
foreman
references
specialists
templates
herdr-observer
herdr-optimizer
herdr-standup