Curate a good-first-issue / help-wanted starter backlog

[omridevk]

Motivation

A healthy good first issue / help wanted backlog is one of the clearest signals to a
prospective contributor that a project is approachable, and it gives maintainers a ready set of
small, well-scoped tasks to hand out instead of improvising one on request. Right now the backlog
of labeled starter issues is thin.

Acceptance criteria

• At least 10 open issues carry good first issue or help wanted.
  Verify (labels combine as AND in a single --label flag, so query separately and union):
  { gh issue list --repo conciv-dev/conciv --label "good first issue" --state open --json number; gh issue list --repo conciv-dev/conciv --label "help wanted" --state open --json number; } \ | jq -s 'add | unique_by(.number) | length'
  returns ≥ 10.
• Every labeled issue is a real, honestly-scoped small task — if fewer than 10 genuine
  candidates exist in the current tracker, write new, real starter issues to fill the gap
  rather than mislabeling larger issues.

Code pointers

• gh issue list --repo conciv-dev/conciv --state open (current backlog to triage)

🤖 Generated with Claude Code

[omridevk]

Grooming: implementation-ready spec

The acceptance criteria's count is already met — but quality isn't

Ran the issue's own verification command today:

{ gh issue list --repo conciv-dev/conciv --label "good first issue" --state open --json number;
  gh issue list --repo conciv-dev/conciv --label "help wanted" --state open --json number; } \
  | jq -s 'add | unique_by(.number) | length'

Result: 12 (no overlap between the two label sets), which already clears the ≥10 bar:

good first issue: #487, #398, #344, #307, #306, #214, #174
help wanted: #448, #416, #337, #224, #193

So the literal AC is satisfied today and this issue's premise ("the backlog is thin") is stale.
The real gap is quality, which the issue also asks for ("every labeled issue is a real, honestly
scoped small task"). Checked each of the 12 against that bar:

Mislabeled — too large/cross-cutting for a first-time contributor, recommend removing the label
(or splitting into a tracking issue + small sub-issues) rather than counting them:

• lint: zero-warning policy — scope Solid rules per-framework, then promote all oxlint warnings to errors #416 "lint: zero-warning policy — scope Solid rules per-framework, then promote all oxlint
  warnings to errors" — a repo-wide lint-policy rollout, not a bounded task.
• Lint does not enforce three AGENTS.md code laws: no-IIFE, no type assertions, no abbreviated identifiers #337 "Lint does not enforce three AGENTS.md code laws: no-IIFE, no type assertions, no
  abbreviated identifiers" — needs new oxlint rule authorship (packages/oxlint-plugin) plus a
  full-repo sweep to land clean; not first-issue scope.
• Re-land no-timers-in-tests with the actual sweep, zero exemptions #214 "Re-land no-timers-in-tests with the actual sweep, zero exemptions" — "the actual sweep,
  zero exemptions" is explicitly repo-wide by its own title.

Removing those three drops the genuinely-small count to 9 (#487, #398, #344, #307, #306,
#174, #448, #224, #193) — just under the bar again once the mislabeled ones are corrected, which
is the honest reason to still add new candidates.

New candidates (verified against real code, each independently small)

Package-level gap, repeatable and mechanical — cross-checked against .fallowrc.json's
publicPackages and packages/publish/src/guards.ts's PUBLIC_PACKAGES so these are confirmed
published packages, not internal ones:

1. Add packages/try/README.md — @conciv/try is public (PUBLIC_PACKAGES in
   packages/publish/src/guards.ts:47) and ships to npm with no README; every sibling published
   package that has one (packages/cli/README.md is the model to follow).
2. Add packages/client/README.md — public (guards.ts:56), no README.
3. Add packages/db/README.md — public (guards.ts:54), no README.
4. Add packages/storage-history/README.md — public (guards.ts:55), no README.
5. Add packages/page/README.md — public (guards.ts:58), no README.
6. Add packages/ui-kit-terminal/README.md — public (guards.ts:69), no README.

(Also missing: packages/contract, packages/extension-compiler, packages/serve — didn't list
individually to avoid an oversized batch of near-duplicates; pick 2-3 of the six above plus these
if more volume is wanted later.)

Docs/repo-hygiene, each independently real and already observed during this grooming pass:

7. apps/site/content/docs/cli.mdx has no init section — already tracked as docs: cli.mdx has no init section #307
   (good first issue), just confirmed still true: grep -n "init" apps/site/content/docs/cli.mdx
   returns nothing, while packages/cli/README.md:5-16 and
   apps/site/content/docs/quick-start/vite.mdx:9-23 both document init in depth. No new issue
   needed — this is corroborating evidence docs: cli.mdx has no init section #307 is a good, real starter task.
8. Add CONTRIBUTING.md — already tracked as Add a CONTRIBUTING.md #495 (currently unlabeled). Genuinely small
   (one file, no code), self-contained, and immediately useful for exactly the audience this issue
   is trying to serve. Recommend adding good first issue to Add a CONTRIBUTING.md #495 rather than opening a duplicate.
9. Improve repo discoverability metadata: topics + social preview image #493 "Improve repo discoverability metadata: topics + social preview image" — already open,
   unlabeled, dx + improvement. Small (GitHub repo settings + one image asset), self-contained.
   Recommend adding good first issue.

Label-scheme note

No changes needed to the label scheme itself — good first issue / help wanted already exist
and are distinct enough (newcomer-friendly vs. needs-more-hands); the gap was curation, not
taxonomy.

Acceptance criteria

• #416, #337, #214 have good first issue/help wanted removed (or are split into a
  parent tracking issue + genuinely small sub-issues that keep the label).
• good first issue added to #495 (CONTRIBUTING.md) and #493 (repo metadata).
• At least 2-3 of the missing-README package issues (candidates 1-6 above) are filed with
  good first issue.
• Re-running the issue's own verification jq command afterward returns ≥ 10 and every
  counted issue is independently small (spot-check: no remaining issue whose title contains
  "sweep", "zero exemptions", "policy", or a repo-wide scope word).

Size

S — no code changes; this is label/issue triage. Filing the 2-3 new README issues is
mechanical (title + the file:line evidence above, copy-pasted per package).