/itp:release

Delegate to the repository's mise release tasks. Every repo should define its own release DAG in .mise/tasks/release/.

Self-Evolving Skill: This skill improves through use. If instructions are wrong, parameters drifted, or a workaround was needed — fix this file immediately, don't defer. Only update for real, reproducible issues.

First Action: Detect Repo's Release Tasks

# Check if the repo has mise release tasks
mise tasks ls 2>/dev/null | grep -i release

If tasks exist: Delegate to mise run release:full (or the repo's equivalent)
If no tasks: Fall back to reading the semantic-release skill for guidance

Arguments

Flag	Short	Description
`--dry`	`-d`	Dry-run mode (preview, no modifications)
`--status`	`-s`	Show current version and release state

Execution

# Full release (preflight → version → sync → verify)
mise run release:full

# Dry-run (preview what would be released)
mise run release:dry

# Check current state
mise run release:status

Why mise Tasks Over Prescriptive Skills

Concern	Prescriptive Skill	mise Task Delegation
Repo-specific logic	Duplicated in skill	Lives in `.mise/tasks/`
DAG enforcement	Manual ordering	`depends` array enforced
Maintainability	Update skill + tasks	Single source in tasks
Portability	Assumes npm/bun	Uses whatever the repo configures
Secrets	Hardcoded patterns	`[env]` in `.mise.toml`

Expected mise Release Task Structure

Repos should follow the hub-and-spoke pattern:

.mise.toml                      # Hub: [env] + [tools] + task docs
.mise/tasks/
  └── release/
      ├── _default              # Help / navigation
      ├── preflight             # Phase 1: Validate prerequisites
      ├── version               # Phase 2: Bump version (semantic-release)
      ├── sync                  # Phase 3: Sync artifacts (marketplace, cache)
      ├── verify                # Phase 4: Verify release artifacts
      ├── full                  # Orchestrator: depends on all phases
      ├── dry                   # Dry-run preview
      └── status                # Current version info

Task DAG

            ┌──────────┐
            │ preflight│
            └─────┬────┘
                  │ depends
            ┌─────▼────┐
            │ version  │
            └─────┬────┘
                  │ sequential
            ┌─────▼────┐
            │   sync   │
            └─────┬────┘
                  │ sequential
            ┌─────▼────┐
            │  verify  │
            └──────────┘

Key Patterns

# .mise/tasks/release/full (orchestrator)
depends = ["release:preflight"]
# Chains: preflight → version → sync → verify

# .mise/tasks/release/preflight (guard)
# Checks: clean dir, auth, plugins, releasable commits

# .mise/tasks/release/version (core)
depends = ["release:preflight"]
# Runs: semantic-release (or language-specific versioning)

Fallback: No mise Tasks

If the repo has no mise release tasks, read the semantic-release skill:

Read: ${CLAUDE_PLUGIN_ROOT}/skills/semantic-release/SKILL.md
Read: ${CLAUDE_PLUGIN_ROOT}/skills/semantic-release/references/local-release-workflow.md

Then follow the 4-phase workflow documented there.

Error Recovery

Error	Resolution
`mise tasks` not found	Install mise: `curl https://mise.run \| sh`
No release tasks	Create `.mise/tasks/release/` or use fallback skill
Working dir not clean	`git stash` or commit changes
Not on main branch	`git checkout main`
No releasable commits	Create a `feat:` or `fix:` commit first
Wrong account	Check `GH_TOKEN` / `GH_ACCOUNT` in `.mise.toml [env]`

Reference

Post-Execution Reflection

After this skill completes, check before closing:

Did the command succeed? — If not, fix the instruction or error table that caused the failure.
Did parameters or output change? — If the underlying tool's interface drifted, update Usage examples and Parameters table to match.
Was a workaround needed? — If you had to improvise (different flags, extra steps), update this SKILL.md so the next invocation doesn't need the same workaround.

Only update if the issue is real and reproducible — not speculative.