feat(plugin): ship coder_eval as a Claude Code plugin + marketplace - #82

Open

uipreliga wants to merge 24 commits into

mainfrom

feat/claude-code-plugin

Commits on Aug 4, 2026

feat(plugin): 1/6 — marketplace + coder-eval plugin skeleton

uipreliga
and
claude
committed
1492496
feat(plugin): 2/6 — generate the bundled criteria reference, guard it with CE032

uipreliga
and
claude
committed
428895f
feat(plugin): 3/6 — skill-check skill and the canonical activation suite

uipreliga
and
claude
committed
c574eb8
feat(plugin): 4/6 — init and task skills

uipreliga
and
claude
committed
af01be4
feat(plugin): 5/6 — analyze and ci skills

uipreliga
and
claude
committed
ed33b3c
feat(plugin): 6/6 — validate the plugin in CI, extend CE026, document it

uipreliga
and
claude
committed
850cc70

Commits on Aug 5, 2026

fix: code review fixes for the Claude Code plugin marketplace

uipreliga
and
claude
committed
4cdfedc
test: defer the plugin-skill repo-file containment guard

uipreliga
and
claude
committed
50b4e34
fix(plugin): name the tautological-criterion trap in init and task

uipreliga
and
claude
committed
55eea70
fix(plugin): make analyze compute its numbers, and weight smoke criteria honestly

uipreliga
and
claude
committed
7f8c4e7
feat(lint): CE026 clause 4 — snippet `with:` keys must be real action inputs

uipreliga
and
claude
committed
6892fed
feat(plugin): 1/5 — the bundled criteria reference explains optional fields

uipreliga
and
claude
committed
93e92df
feat(plugin): 2/5 — shared adversarial task rubric, applied by `task`

uipreliga
and
claude
committed
1955458
feat(plugin): 3/5 — a real run becomes part of done in `task`

uipreliga
and
claude
committed
9a82b6e
feat(plugin): 4/5 — `/coder-eval:lint-tasks`, a read-only reviewer of existing tasks

uipreliga
and
claude
committed
4f2376b
feat(plugin): 5/5 — activation budgets in `skill-check`, layer routing in `analyze`

uipreliga
and
claude
committed
d52d762
fix: make lint-tasks' read-only rule outlive the frontmatter deny

uipreliga
and
claude
committed
2489081
docs: defer one harness candidate from the plugin-audit run

uipreliga
and
claude
committed
c36f8ea
fix: code review fixes for the plugin-audit P0/P1 plan

uipreliga
and
claude
committed
fc429e1
fix(tasks): armed positives must require success, guarded by CE034

uipreliga
and
claude
committed
e706bb1
fix(plugin): address PR #82 review — reachable activation suites, least-privilege CI

uipreliga
and
claude
committed
0d3c5a0
docs: add Tutorial 07 for the plugin, and fix two gaps in PLUGIN.md

uipreliga
and
claude
committed
84dd3b2
docs: rework Tutorial 07 after review — accuracy fixes and far less narration

uipreliga
and
claude
committed
6c354af
feat(plugin): CLI-driving skills offer to install coder-eval, asking first

uipreliga
and
claude
committed
8846ff5

Navigation Menu

Uh oh!

feat(plugin): ship coder_eval as a Claude Code plugin + marketplace - #82

feat(plugin): ship coder_eval as a Claude Code plugin + marketplace#82
uipreliga wants to merge 24 commits into
mainfrom
feat/claude-code-plugin

Commits on Aug 4, 2026

feat(plugin): 1/6 — marketplace + coder-eval plugin skeleton

feat(plugin): 2/6 — generate the bundled criteria reference, guard it with CE032

feat(plugin): 3/6 — skill-check skill and the canonical activation suite

feat(plugin): 4/6 — init and task skills

feat(plugin): 5/6 — analyze and ci skills

feat(plugin): 6/6 — validate the plugin in CI, extend CE026, document it

Commits on Aug 5, 2026

fix: code review fixes for the Claude Code plugin marketplace

test: defer the plugin-skill repo-file containment guard

fix(plugin): name the tautological-criterion trap in init and task

fix(plugin): make analyze compute its numbers, and weight smoke criteria honestly

feat(lint): CE026 clause 4 — snippet `with:` keys must be real action inputs

feat(plugin): 1/5 — the bundled criteria reference explains optional fields

feat(plugin): 2/5 — shared adversarial task rubric, applied by `task`

feat(plugin): 3/5 — a real run becomes part of done in `task`

feat(plugin): 4/5 — `/coder-eval:lint-tasks`, a read-only reviewer of existing tasks

feat(plugin): 5/5 — activation budgets in `skill-check`, layer routing in `analyze`

fix: make lint-tasks' read-only rule outlive the frontmatter deny

docs: defer one harness candidate from the plugin-audit run

fix: code review fixes for the plugin-audit P0/P1 plan

fix(tasks): armed positives must require success, guarded by CE034

fix(plugin): address PR #82 review — reachable activation suites, least-privilege CI

docs: add Tutorial 07 for the plugin, and fix two gaps in PLUGIN.md

docs: rework Tutorial 07 after review — accuracy fixes and far less narration

feat(plugin): CLI-driving skills offer to install coder-eval, asking first