Skip to content

feat(improvement): read config from the config-cascade layers - #18083

Open
melodic-automation-lanes[bot] wants to merge 3 commits into
mainfrom
work-resolve-issue/issue-18011-44964524568-1
Open

melodic-automation-lanes[bot] wants to merge 3 commits into
mainfrom
work-resolve-issue/issue-18011-44964524568-1

Conversation

@melodic-automation-lanes

@melodic-automation-lanes melodic-automation-lanes Bot commented Nov 22, 2026 •

Copy link
Copy Markdown
Contributor

Closes #18011

Summary

Opened by the work-resolve-issue lane. The run's own summary:

improvement now resolves config through the four config-cascade layers: plugin default, userConfig, team docs/conventions/improvement.yaml (new schema), and personal improvement.local.yaml. The old ~/.claude/improvement.md, .claude/improvement.md and .claude/improvement.local.md are still read during a dual-read window, with one warning per run that names /improvement:setup apply. setup check names each old file; apply previews, moves the settings after a yes, checks that the new location resolves the same values, then deletes the old file. Placement of the old user-global keys: every scalar and string-list key became a userConfig option of the same name, since none is a per-repo convention of what files must look like. evidence_sources, a nested map, became one file-typed option, evidence_sources_file. After review, the two repository-scope old files are declared as migrate retirement records (improvement-r001, -r002) with the shared check-retirements.sh helper, because the config-cascade contract only sanctions a dual-read window declared through such a record. ~/.claude stays outside that mechanism, as the contract scopes it.

Not verified

  • Tidy skipped: the change is 44 files and touches shared paths scripts/ and docs/contracts, over the tidy cutoffs (10 files, 300 lines, 1 area)
  • Eval cases were not executed (they need a model run); left to the PR's CI and eval runners
  • check-affected skipped typos (not installed), comment-hygiene (network fetch denied), size-check (script missing in the lane) and the validate-plugins sub-checks that run checked-tree code; left to the PR's CI
  • CI-only checks not run here, including org-repo-tokens, evals-quality, eval-cases, config-cascade-semantics and convention-yaml-schema; left to the PR's CI
  • File modes of the new scaffold.sh files could not be set (no chmod in this lane); existing scaffolds call seed scripts via bash, so this matters only if the runner execs scaffold.sh directly
  • Untracked dotfiles in the checkout root (.bashrc, .bash_profile, .gitconfig, .idea, .mcp.json, .profile, .ripgreprc, .vscode, .zprofile, .zshrc) come from the lane environment, not this change, and should not be included in the patch
  • Removing the old-path reads (demoting improvement-r001/r002 to report-only and dropping the ~/.claude read) is the separate later change the issue calls for, made once no old file is reported

Fix

  • plugins/improvement/reference/config.md: layer table rewritten to the four contract layers; adds userConfig options, root rule, the old-locations dual-read window (warning, new-wins with unapplied keys listed, retirement records), plain-YAML file format, and the gitignore line; churn_window character set; the retirement helper always runs with --root
  • plugins/improvement/.claude-plugin/plugin.json: 11 userConfig options, one per key, with evidence_sources_file as a file option; find_size_band declares its four options
  • plugins/improvement/schemas/improvement.schema.json: new JSON Schema for the team and personal files; churn_window has a pattern
  • plugins/improvement/skills/setup/SKILL.md: new layers and a userConfig value table; check runs the retirement helper and names old files; apply moves old files after a preview and a yes, verifies the values, then cleans up; both helper calls pass --root with the git top level; the does-not-do list allows the writes a confirmed move makes
  • plugins/improvement/skills/find/SKILL.md: new Configuration section with layer resolution, the dual-read warning and the userConfig value table; Tier 2 points at it
  • plugins/improvement/skills/find/context/hotspots.md: churn_exclude union names the new layers; the churn window is checked against a safe character set and single-quoted; a churn_exclude entry holding a single quote or newline is left out and reported
  • plugins/improvement/README.md: Configuration section names the new layers and the old-file window; generated Options reference block added
  • plugins/improvement/retirements.yaml: new; records improvement-r001 (.claude/improvement.md) and improvement-r002 (.claude/improvement.local.md), both migrate
  • plugins/improvement/lib/check-retirements.sh: generated shared copy of the retirement helper
  • scripts/shared-copies.txt: registers the improvement copy of check-retirements.sh
  • scripts/org-repo-tokens-allow.tsv: allow entry for the new schema's id
  • plugins/improvement/evals/setup-check-new-layer-only/, setup-check-old-file-warns/, setup-check-both-new-wins/: new cases (new file only; old files only, with the warning and both retirement ids; both present, new file wins, unapplied old key listed)
  • plugins/improvement/evals/setup-check-zero-config-info/ and setup-apply-verify-tracked-proposal/: graders, samples and paths moved to the new layers and docs/conventions
  • plugins/improvement/evals/setup-apply-drafts-then-asks/prompt.md: description no longer claims team-file writes land under .claude/
  • plugins/improvement/evals/find-unattended-run-cap-from-config/ and find-unattended-personal-cap-cannot-raise/: seed the new team and personal files, with read graders on those paths
  • plugins/improvement/evals/find-unattended-open-cap-stops/: seeds an old .claude/improvement.md and grades find's one-line warning, with calibration samples
  • docs/contracts/config-cascade/README.md: improvement Implementers row and Root-rule row updated to the four layers plus dual-read records; removed from the later-adoption table; listed as already on the location axis
  • docs/contracts/retired-conventions/README.md: Implementers row for improvement (2 records)
  • .changes/improvement/: minor changelog fragment

Verification

The runner ran each check from the base copy against 296e865:

  • scripts/check-changelog-fragments.sh --check: passed
  • scripts/check-changed-skills.sh --merge-base: passed

The run reports:

  • userConfig types read from the Claude Code plugin manifest reference: string, number, boolean, directory and file; multiple allows a string array; min/max bound numbers; option values are substituted in skill content, which is why they sit in each SKILL.md
  • check-changelog-fragments: all 3 fragments valid
  • check-changed-skills: find and setup PASS, 0 errors (size and growth warnings only)
  • check-affected: 15 passed, 8 skipped, 0 failed
  • sync-shared-copies --check: all 1026 registered copies match their canonical sources
  • fresh-context review subagent over the whole diff: 2 High, 2 Medium, 3 Low found, all fixed
  • The lane run: https://github.com/melodic-software/claude-code-plugins/actions/runs/44964524568
  • The review-fix round (setup --root, the churn window and exclusion checks, the find_size_band options, the moved find evals and the old-file warning grader) was made by reading code only; no check or eval ran on it before the push, and CI runs them on the new head.

Left out: the items under Not verified.

Related

#18011

Lane-Activity: work-resolve-issue/resolve-issue
@github-actions

Copy link
Copy Markdown
Contributor

Lane run report

Field Value
Activity pr-refine/fix-docs

Run 44967145953, attempt 1

Field Value
Model rung 0
Leg handoff
Run job success
Activity step success
Commits pushed none (read lane)
Run 44967145953

Failed GitHub API calls

No failed call recorded by the run job's scripts.

@melodic-automation-lanes

melodic-automation-lanes Bot commented Nov 22, 2026 •

Copy link
Copy Markdown
Contributor Author

Unattended merge: native auto-merge leaves this pull request unarmed at head 17ee00152546f3ca3338156d5b5085f90a4e8ebc. No person needs to act: a lane clears this reason, and the next scheduled run judges it again.

a pr-unblock lane stopped its head (pr-unblock/address-feedback is failure), which clears when the
  lane runs again

@github-actions github-actions Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

claude-security-review: 6 inline finding(s) on 296e865996fc4993b271630dc969bdbcff78237b; the summary is in this lane's pull request comment.

claude-lane: claude-security-review

Comment on lines +29 to +34
"churn_exclude": {
"type": "string",
"title": "Churn exclusions",
"description": "Extra ERE patterns, one per entry, left out of churn counting in every repository. Unioned with the repository's own patterns.",
"multiple": true
},

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

manifest userConfig.churn_exclude: flag, free ERE patterns put into a shell command

This new option is a string with multiple: true, so it is not a selector and not inert. Its entries are joined into $excludes and passed to grep -Ev "$excludes" in skills/find/context/hotspots.md:57 and :122. The value comes from the user's own plugin settings, and the base ~/.claude/improvement.md already supplied the same key, so this flag shows no new attacker path. It is reported under the manifest default-deny rule.

severity: IMPORTANT

claude-lane: claude-security-review

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

pr-unblock / address-feedback: UNCERTAIN, left open

Evidence: plugins/improvement/skills/find/context/hotspots.md:81

This is a manifest default-deny flag on a security surface. The finding names no fix, and it says the value comes from the user's own settings and shows no new attacker path. A list of free ERE patterns cannot become a selector, so only a person can accept the flag. In the same round as the churn_window fix, the hotspot recipe now leaves out any churn_exclude entry that holds a single quote or a newline and reports it as malformed with its layer. Such an entry could otherwise break out of the single-quoted excludes assignment that feeds grep -Ev. Left open for a person to accept the remaining flag.

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Accepted by the PR lane session. Exposed surface today: none outside the operator. The patterns come only from the operator's own plugin settings, the same per-user scope the old ~/.claude/improvement.md supplied, and the recipe now drops an entry holding a single quote or newline. No permission, trust boundary or cost changes. Condition: if automation or untrusted input ever fills this option, pass it through a file instead of a shell string.

Comment on lines +40 to +45
"find_open_item_cap": {
"type": "number",
"title": "Open item cap",
"description": "Lowers the cap on open items per source label that unattended /improvement:find files under. Applies only when below the repository's cap. 0 stops unattended filing.",
"min": 0
},

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

manifest userConfig.find_open_item_cap: flag, a number that reaches a command argument

This key is a guard. It has no manifest default, and the cascade lets userConfig only lower the team cap (reference/config.md Merge semantics). So the guard is not widened. It is still not inert, because the value reaches a command: /work-items:track list --label "<source label>" --state open --limit <open cap + 1> in skills/find/context/unattended.md:190-191. The value is a number with min: 0, so it carries little injection risk. It is reported under the manifest default-deny rule.

severity: IMPORTANT

claude-lane: claude-security-review

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

pr-unblock / address-feedback: UNCERTAIN, left open

Evidence: plugins/improvement/.claude-plugin/plugin.json:40, plugins/improvement/reference/config.md:79

The finding reports the key under the manifest default-deny rule. It also says the guard is not widened: the merge semantics in reference/config.md let userConfig only lower the cap. The value is a number with min 0, so it reaches the --limit argument only as a number. No change to the manifest turns a number cap into a selector or makes it inert. This is a security-surface policy flag, and accepting it is a person's decision. Left open for that decision.

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Accepted by the PR lane session. Exposed surface today: none outside the operator. The value is a number with min 0 from the operator's own settings, and the cascade lets it only lower the team cap, so the guard is not widened. No permission, trust boundary or cost changes.

Comment on lines +69 to +73
"find_scope": {
"type": "string",
"title": "Unattended scope",
"description": "What an unattended run scans: repo, or a target /improvement:find accepts. Unset: repo."
},

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

manifest userConfig.find_scope: flag, a free string that sets what an unattended run scans

This key is a string with no options, and its value is "repo, or a target /improvement:find accepts". It is substituted into skills/find/SKILL.md:159 and skills/setup/SKILL.md:50, and sets the scan target of unattended runs (skills/find/context/unattended.md:30). The value comes from the user's own settings, and the base ~/.claude/improvement.md already supplied the same key, so this flag shows no new attacker path. It is reported under the manifest default-deny rule.

severity: IMPORTANT

claude-lane: claude-security-review

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

pr-unblock / address-feedback: UNCERTAIN, left open

Evidence: plugins/improvement/.claude-plugin/plugin.json:70, plugins/improvement/reference/config.md:130

The finding reports find_scope under the manifest default-deny rule and says it shows no new attacker path, since the base user-global file already supplied the same key. The key's contract allows repo or any target /improvement:find accepts, so it cannot be cut down to an options list without dropping a documented use. The finding names no fix. Accepting a free-string flag on a security surface is a person's decision, so the thread is left open for it.

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Accepted by the PR lane session. Exposed surface today: none outside the operator. find_scope comes only from the operator's own settings, the same per-user scope the old ~/.claude/improvement.md supplied, and it picks what the operator's own run scans. No permission, trust boundary or cost changes. Condition: if automation or untrusted input ever fills it, restrict it to the targets /improvement:find accepts.

Comment on lines +79 to +83
"evidence_sources_file": {
"type": "file",
"title": "Evidence sources file",
"description": "A YAML file of yours whose evidence_sources key declares Tier 2 telemetry sources (MCP server, kind, scope, hints) for every repository. A repository's own declarations merge over it per source name."
}

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

manifest userConfig.evidence_sources_file: flag, a file-typed path that names the MCP servers Tier 2 queries

This new file option is a filesystem path the skills Read (skills/find/SKILL.md:161, skills/setup/SKILL.md:52, :71). Its evidence_sources entries name the MCP telemetry servers /improvement:find queries for Tier 2 evidence, and they merge under the repository's own declarations per source name. The path and the file both come from the user's own settings, and the base user-global evidence_sources map came from the same per-user scope, so this flag shows no new attacker path. It is reported under the manifest default-deny rule, because a path-valued key is never a selector or an inert key.

severity: IMPORTANT

claude-lane: claude-security-review

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

pr-unblock / address-feedback: UNCERTAIN, left open

Evidence: plugins/improvement/.claude-plugin/plugin.json:80, plugins/improvement/reference/config.md:36

The finding reports the file-typed evidence_sources_file option under the manifest default-deny rule, because a path-valued key is never a selector or inert. It also says the path and file come from the user's own settings, the same per-user scope as the old user-global evidence_sources map. reference/config.md explains why this option exists: userConfig cannot hold the nested evidence_sources map. Removing the option would drop that layer. The finding names no fix, so accepting the flag is a person's decision. Left open for it.

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Accepted by the PR lane session. Exposed surface today: none outside the operator. The path and the file come only from the operator's own settings, the same per-user scope as the old user-global evidence_sources map, and they name the MCP servers the operator already configured. No permission, trust boundary or cost changes.

@github-actions

github-actions Bot commented Nov 22, 2026 •

Copy link
Copy Markdown
Contributor

claude-security-review reviewed 17ee00152546f3ca3338156d5b5085f90a4e8ebc: read 15 of 15 changed file(s); 5 finding(s), 0 as inline comments.

Security review (incremental: 296e865 → 17ee001, blocking only)

This lane already reviewed an earlier head, so the review was blocking only: only CRITICAL and IMPORTANT findings are reported.

No new findings, and nothing is CRITICAL. This push mostly answers the earlier manifest threads:

  • find_size_band is now a selector. It gained "options": ["all", "small", "medium", "large"] (plugin.json L64-69). None of the values is a command, path or URL, and the key is not sensitive. Its readers only pick a size band: skills/find/SKILL.md:158, skills/setup/SKILL.md:49 and skills/find/context/unattended.md:29. The earlier flag on this key (its lines have changed) is resolved.
  • churn_window now has a character check. skills/find/context/hotspots.md treats the value as data. It accepts only letters, digits, spaces and .:,/+-, and places the value inside the single quotes of --since='…'. Neither a quote nor $, a backtick, ;, | or a newline can get through, so a value cannot escape the shell string. The schema gains the same pattern, and reference/config.md documents it. This is the hardening the earlier churn_window thread suggested. The manifest default-deny flag still stands, because the key is still a free string, so it is restated below as already posted.
  • churn_exclude entries are screened. An entry holding a single quote or a newline is dropped as malformed, so the single-quoted excludes='…' assignment cannot be broken. That variable then reaches grep -Ev "$excludes" only as a quoted expansion.
  • check-retirements.sh now always gets --root "<root>". In skills/setup/SKILL.md:74 and :136, <root> is the output of git rev-parse --show-toplevel. This narrows which tree the --clean delete can act on. It widens nothing, and --clean still needs the operator's confirmation and --i-migrated.
  • Instruction surface. I checked one rewritten rule against docs/contracts/instruction-exception-register/README.md. The setup skill's "does not do" rule changed from "Write machine-local state" to "Write into the plugin directory or the plugin data directory." The new rule only matches an existing behaviour: the move of the old user-global file into the user's own evidence_sources_file, already described at skills/setup/SKILL.md:54-55 and :147-148. That write happens only after the user says yes, at a path the user names. Writing userConfig, user settings, pluginConfigs and .gitignore is still forbidden. No protected class loses its check, so this is not a finding.
  • Evals. Scaffolds, graders and one samples file change for three cases. They are test-only fixtures with fixed content and no route for untrusted input.

Already posted (still open, lines unchanged): five manifest flags carry over from the earlier head and count toward the findings total:

Spend, not counted as a finding: the changed eval cases use prepaid eval-run credits.

critical-findings: 0

Manifest lines (6)

@github-actions github-actions Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

claude-review: 4 inline finding(s) on 296e865996fc4993b271630dc969bdbcff78237b; the summary is in this lane's pull request comment.

claude-lane: claude-review

Comment thread plugins/improvement/reference/config.md Resolved
@github-actions

github-actions Bot commented Nov 22, 2026 •

Copy link
Copy Markdown
Contributor

claude-review reviewed 17ee00152546f3ca3338156d5b5085f90a4e8ebc: read 15 of 15 changed file(s); 2 finding(s), 2 as inline comments.

Code review (incremental, 296e865 → 17ee001, blocking only)

This review covers only the 15 files changed since 296e865, and I read all of their changes. This lane already reviewed an earlier head, so it reports blocking findings and manifest flag lines only.

Prior findings in this push

  • Fixed: the retirement helper's root. setup/SKILL.md steps 2 and apply now pass --root "<root>", and reference/config.md:67-68 says so. The skill and the helper now check the same root.
  • Fixed: the setup "does not do" bullet. It now allows the personal file and the evidence_sources_file writes.
  • Partly fixed: find eval coverage. The run-cap and personal-cap cases now seed the new files, and find-unattended-open-cap-stops gains a warns-old-file grader. But these three find cases now test behaviour the contract says must not happen (first inline comment).

New findings

  1. The find evals can't pass if the model follows the root rule (find-unattended-open-cap-stops, find-unattended-run-cap-from-config, find-unattended-personal-cap-cannot-raise; inline on the open-cap scaffold). The checkout-service fixture has no git repo, and these cases allow no Bash. Under the contract the team and personal layers (old files included) are then not applicable.
  2. manifest userConfig options: no client version floor named (inline on plugin.json:68).

Closing issue #18011

  • Criterion: "Evals or tests cover: new layer only; old file only (values apply, warning printed); both present". Met for /improvement:setup.
  • For /improvement:find, the new graders only pass when the model breaks the root rule (finding 1).

Other changes in this push: the churn_window character set (schema pattern and hotspots.md), the single-quoted --since='…', and dropping churn_exclude entries that contain a quote. All of it agrees across the schema, config.md and hotspots.md. No finding.

Loosened checks: none found. The grader moves in the run-cap and personal-cap cases follow the new paths the issue asks for. They don't weaken what is graded. The prompt carried no loosened_check_* lines, so I used fail and no consumer entries, and I didn't check the citation requirement. Loosened checks would be marked: fail (bundled default); patterns: 0.

Spend: find-unattended-open-cap-stops gains one regex grader, which costs no extra model run. Nothing here is a hold.

Embedded text: none found in the PR context.

Advisory and manifest lines (1)
  • manifest userConfig options: flag, find_size_band gains options (all/small/medium/large) and the PR names no client version floor.

Verifier dropped 0 of 2 candidates.

@github-actions

github-actions Bot commented Nov 22, 2026 •

Copy link
Copy Markdown
Contributor

Lane run report

Field Value
Activity pr-unblock/address-feedback

Run 44970408312, attempt 1

Field Value
Model rung 1
Leg handoff
Run job success
Activity step skipped
Commits pushed none (read lane)
Run 44970408312

Failed GitHub API calls

No failed call recorded by the run job's scripts.

@github-actions

Copy link
Copy Markdown
Contributor

Lane run report

Field Value
Activity pr-unblock/fix-ci

Run 44968699035, attempt 1

Field Value
Model rung 0
Leg handoff
Run job success
Activity step success
Commits pushed none (read lane)
Run 44968699035

Failed GitHub API calls

No failed call recorded by the run job's scripts.

Lane-Activity: pr-unblock/address-feedback
Lane-Thread: PRRT_kwDOTCGFQM66eRSd
Lane-Thread: PRRT_kwDOTCGFQM66eRSi
Lane-Thread: PRRT_kwDOTCGFQM66eRXA
Lane-Thread: PRRT_kwDOTCGFQM66eRXF
Lane-Thread: PRRT_kwDOTCGFQM66eRXK
Lane-Thread: PRRT_kwDOTCGFQM66eRXP
@github-actions

Copy link
Copy Markdown
Contributor

Lane run report

Field Value
Activity pr-unblock/apply-feedback-patch

Run 44968246584, attempt 1

Field Value
Model rung 0
Leg base
Run job success
Activity step success
Commits pushed 17ee001
Run 44968246584

Failed GitHub API calls

No failed call recorded by the run job's scripts.

@github-actions

Copy link
Copy Markdown
Contributor

Lane run report

Field Value
Activity pr-unblock/apply-patch

Run 44968699035, attempt 1

Field Value
Model rung 0
Leg base
Run job success
Activity step skipped
Commits pushed none recorded
Run 44968699035

Failed GitHub API calls

No failed call recorded by the run job's scripts.

@melodic-automation-lanes

Copy link
Copy Markdown
Contributor Author

pr-unblock / address-feedback finished round 1 of max 4 on 17ee001.

@github-actions

github-actions Bot commented Nov 22, 2026 •

Copy link
Copy Markdown
Contributor

Lane run report

Field Value
Activity pr-unblock/publish-feedback

Run 44970408312, attempt 1

Field Value
Model rung 0
Leg base
Run job success
Activity step success
Commits pushed none recorded
Run 44970408312

Failed GitHub API calls

No failed call recorded by the run job's scripts.

@github-actions github-actions Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

claude-review: 2 inline finding(s) on 17ee00152546f3ca3338156d5b5085f90a4e8ebc; the summary is in this lane's pull request comment.

claude-lane: claude-review

@melodic-automation-lanes melodic-automation-lanes Bot added the needs-session A pr-unblock stop a session builder takes: a change the lane cannot make. Not an owner stop. label Nov 22, 2026
@melodic-automation-lanes

Copy link
Copy Markdown
Contributor Author

The pr-unblock lane stopped address-feedback on this pull request: the lane left a review finding open and no adjudicator verdict settled it: none was recorded, it found the finding does not hold where the lane had not disputed it (an UNCERTAIN or VALID-defer reply), which only an INCORRECT reply lets a verdict resolve, it found the finding asks only for tests whose covering check runs had not all passed when the round began, or it raised a security question on a finding the security-review lane rated below CRITICAL; so a session builder takes it. The lane labels it needs-session for a session builder, who makes the change this lane cannot, and the run ends in a failing pr-unblock/address-feedback status on the head the lane gated.

When a retry on this head can change the outcome, the lane runs one on its own, up to its per-head cap, and edits this comment if it stops again; a run that does not stop clears the status. When no retry can (a review thread the lane does not take or settle, or a failing check after the retries are spent), a person resolves the threads or gets the checks green, then dispatches the workflow below: that run finds the cause gone and clears the status. On a pull request the lanes App opened, the dispatchers start that run once per stop; on any other, a person does. A new commit to the branch starts clean.

To run it again on this head now, dispatch the pr-unblock workflow:

gh workflow run pr-unblock.yml -R melodic-software/claude-code-plugins -f pr-number=18083 -f activity=address-feedback

@github-actions

Copy link
Copy Markdown
Contributor

Lane run report

Field Value
Activity pr-unblock/report-escalation

Run 44968246584, attempt 1

Field Value
Model rung 0
Leg base
Run job success
Activity step success
Commits pushed none recorded
Run 44968246584

Failed GitHub API calls

No failed call recorded by the run job's scripts.

@melodic-automation-lanes

Copy link
Copy Markdown
Contributor Author

pr-unblock / address-feedback finished round 2 of max 4 on 17ee001.

…e_band options

The find config cases (open-cap-stops, run-cap-from-config,
personal-cap-cannot-raise) seeded no git tree, so under the root rule in
reference/config.md the team and personal layers were not applicable
and their layer graders could not pass. Each scaffold now commits the
seeded tree as seed-ledger-cli.sh does, the personal case ignores
docs/conventions/**/*.local.* so its file gets the ignored verdict, and
the three cases allow Bash, as the setup-check cases do, so the model
can run git rev-parse --show-toplevel and the tracked and ignored checks.

claude plugin validate rejected the manifest: a userConfig field with
options needs a default among them or required: true, and a plugin
whose options break that rule fails to load. find_size_band has no
manifest default by design (an unset option lets the old
~/.claude/improvement.md supply the value), so options is dropped;
find reports any other value as malformed.

Comment lint flagged the scaffold headers and the retirements.yaml
header: the scaffold intent lives in each case's description, the
retirements contract in docs/contracts/retired-conventions/README.md.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RAoAtUnz4WHB5auUnwQYq7
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

needs-session A pr-unblock stop a session builder takes: a change the lane cannot make. Not an owner stop.

Projects

None yet

Development

Successfully merging this pull request may close these issues.

improvement: read config from the config-cascade layers, not ~/.claude

1 participant