projekt-issues

Projekt — issues (bulk create + assign/move)

Create issues in bulk from a file, and assign+move existing issues, with dry-run safety and idempotent resume. Part of the projekt pipeline (phases CREATE + ASSIGN).

SK="${CLAUDE_SKILL_DIR}/scripts" — use it for every command below.

Prerequisite (connect once)

Run the projekt skill's setup first; these scripts read .projekt-run/context.json and never re-query identity:

bash "${CLAUDE_SKILL_DIR}/../projekt/scripts/auth_check.sh"
bash "${CLAUDE_SKILL_DIR}/../projekt/scripts/context_sync.sh"

If there's no token, see skills/projekt/references/auth-setup.md.

1. Bulk create — bulk_issue_create.py

Reads the columns of skills/projekt/assets/import_template.csv (title,description,status,assignee,estimated_hours,priority,type,labels,external_ref) or a JSON list. Resolves the project by key/name and each assignee (email or name) → user_id from context. Dedupes against a live GET /issues sweep (by title and external_ref) and the Ledger.

# DRY-RUN: prints a create/skip table, writes nothing
python3 "$SK/bulk_issue_create.py" --project WEB --file backlog.csv

# APPLY: sequential POST /issues, ≤3 in flight, every create logged for resume
python3 "$SK/bulk_issue_create.py" --project WEB --file backlog.csv --apply

There is no bulk-create endpoint — /issues/bulk only mutates existing issues. So creation is one POST per row at concurrency ≤3 (--concurrency, capped at 3). Re-running creates 0 (idempotent).

Flags: --strict-status skips (instead of demoting) rows that target a working column with no owner.

2. Assign + move — assign_and_move.py

Assigns an owner then moves issues to a target column via POST /issues/bulk — two actions in order: {action:"assignee",value} then {action:"status",value}.

# DRY-RUN
python3 "$SK/assign_and_move.py" --issues WEB-12,WEB-13 --assignee [email protected] --status "In Progress"

# APPLY
python3 "$SK/assign_and_move.py" --issues WEB-12,WEB-13 --assignee [email protected] --status "In Progress" --apply

Issue tokens are keys (WEB-12) or ids. --assignee is optional (omit it when the issues already have owners and you only need to move them).

The assignee-required rule (critical)

An issue cannot leave Backlog/To Do for a working column (In Progress / In Review / Done) without an assignee_id — the API returns 422 (blocked_unassigned). See skills/projekt/references/errors.md.

• create: a row targeting a working column with no resolvable assignee is demoted to To Do and flagged "needs owner" (never dropped). --strict-status skips it instead.
• assign/move: any issue still unassigned after the optional assign step is pre-filtered out of the move and reported as "needs owner"; the rest of the batch still moves. Creating/parking in Backlog/To Do without an assignee is always fine.

Gotchas

• /issues/bulk does NOT create — it only mutates existing issues (assignee/status/priority/labels). Use bulk_issue_create.py for creation.
• Dedupe keys: (project_id,title) + external_ref. Give every import row a stable external_ref so re-imports are safe even if a title is edited.
• Status names are per-project (project.columns); the server normalizes localized inputs ("En revisión"). Defaults: Backlog, To Do, In Progress, In Review, Done (errors.md).
• 403 cross-org is fatal and not retried — a PAT is bound to one org; fix the token/org. 429 is auto-backed-off by the client; keep concurrency ≤3. (errors.md)
• Re-run after fixing owners — both scripts dedupe via the Ledger and resume cleanly.

What it does NOT do

No estimation (→ projekt-estimate), no time logging (→ projekt-time), no docs (→ projekt-docs), no workload reports (→ projekt-workload). No hard delete (use /issues/:iid/archive). No CSV editing — it only reads the import file. It never touches the 1.3 MB spec.