Repository navigation
Labels
agent-readyFully specified and briefed; eligible for autonomous pickup from the frontier.Fully specified and briefed; eligible for autonomous pickup from the frontier.
Description
Parent
Refs #15703
Source: PLAN Phase 12, topic ecosystem-specifics
Depends on
Outcome
/harness-config:audit-automation-gaps prints, read-only, how a repository's enforcement is spread across the ladder's tiers: a count per tier and the rules that sit lower than they could, both rolled up from one generated rule-to-enforcer inventory.
Demo: run the audit against this repository; it prints five tier counts and a "sits lower than it could" list.
What changes:
- A generator produces one inventory row per rule: rule, enforcer, rung, tier.
- The audit's "Present Results" phase prints the per-tier count and the lower-than-it-could list, read-only, with no target.
- When this item's pull request opens, post one comment on Count corrections that became checks and generate the rule-to-enforcer inventory #15650 narrowing that item to the corrections counter; the inventory is built here, once.
Method:
- Tests first for the generator: a fixture repository with one rule per tier yields one row each and a per-tier count; a rule held by guidance that a static check could decide appears under "sits lower than it could".
- The skill-text change ships on a two-version compare.
- Run budget: 100 runs (about 3 automation-gaps audit cases, counted first, plus 2 new, on the
opusandsonnetaliases, three runs per side, after two measuring runs). Select cases with--case, use--runs 3and a--max-cost-usdceiling, and start no batch after a usage-limit error. - Eval batches run in the one local eval-runner session in the main checkout. Open the pull request with the
do-not-mergelabel; the compare record releases it.
Decisions
- Design Q22: the rule-to-enforcer inventory and the per-tier count are one artifact; the count and "sits lower than it could" are rollups of it.
- D5: Count corrections that became checks and generate the rule-to-enforcer inventory #15650 is narrowed by one comment to the corrections-to-checks counter, which stays a separate measure (approved by the owner with the plan).
- Her team's numbers are not targets: the audit prints no target.
Key interfaces
- "
harness-config(audit-automation-gaps): owns one generated rule-to-enforcer inventory (rule, enforcer, rung, tier). The per-tier count and 'sits lower than it could' are rollups of it." - Inventory row: rule, enforcer, rung, tier.
- Two-way door: the generated inventory.
Done when
- AC9: the automation-gaps audit reports, read-only, a count per ladder tier and the rules that sit lower than they could.
- The generator's test file passes.
- A run of the audit against this repository prints five tier counts (dogfood), named in the pull request body.
- The Count corrections that became checks and generate the rule-to-enforcer inventory #15650 comment's URL is in the pull request body.
- The two-version compare is non-inferior and its record is in the pull request body.
- AC7: nothing is cut here, so no control result is owed.
- AC10: no new enforcement-placing skill is added here.
Out of scope
- The corrections-to-checks counter (Count corrections that became checks and generate the rule-to-enforcer inventory #15650).
- Selection rules rows in the ladder (a later item).
Metadata
Metadata
Assignees
Labels
agent-readyFully specified and briefed; eligible for autonomous pickup from the frontier.Fully specified and briefed; eligible for autonomous pickup from the frontier.
Type
Fields
Work status
confirmed
Priority
medium
Activity
github-actions commentedon Nov 13, 2026
The triage sweep dispatched intake triage for this issue (dispatch 1; 2 runs that reach intake, or 6 dispatches, hand it to a maintainer).
github-actions commentedon Nov 13, 2026
Size: Medium
Complexity: Medium
Uncertainty: Known
Flags: none
Blocked by: #15708
I understood this as: add a generator that produces a rule-to-enforcer inventory (rule, enforcer, rung, tier). Then make
/harness-config:audit-automation-gapsprint, read-only, a count per tier and a "sits lower than it could" list from that inventory. The issue also asks for a narrowing comment on #15650, which is related work (the inventory moves here). The issue states an exact outcome and done-when, so I have no open questions. It depends on #15708 (the ladder release), which is still open. I found no duplicate and no merged fix.Automated triage. A maintainer reviews it.
melodic-automation-lanes commentedon Nov 14, 2026
Not admitted to the issue lane yet: blocker #15708 is not closed by a listed human or a merge.
melodic-automation-lanes commentedon Nov 15, 2026
{"v":1,"actor_id":337595936,"title_sha256":"2f07f90a4ee2544ecd1a06068dd3a1dafccd9b5310616dfc4b6ba9b2e108de6f","body_sha256":"8ba52f207993a1ad9445df17011f19aa4004a06781a440e8de740a4cd353567b"}
github-actions commentedon Nov 15, 2026
melodic-automation-lanes commentedon Nov 15, 2026
The work-resolve-issue lane stopped on this issue (
host-only): the run reports the work needs a host run the lane cannot make.It added
host-runand set Work status to confirmed: an agent run on a host builds it, and it waits on no decision from a person.In the run's words:
What the run says it did not do:
Run: https://github.com/melodic-software/claude-code-plugins/actions/runs/43751910317
To hand it back to the lane, remove
host-runand set Work status to agent-ready again.feat(harness-config): name the sources the rule inventory does not count