How do I install the Jj Hunk skill?

Run npx skillmds@latest add knoopx/jj-hunk in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Jj Hunk safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Jj Hunk?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Jj Hunk free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Jj Hunk?

knoopx (@knoopx) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Jj Hunk by knoopx — Agent Skill

Name: Jj Hunk
Author: knoopx

File contents

jj-hunk

Programmatic hunk selection for Jujutsu. Split, commit, or squash specific hunks without interactive prompts.

Basic Workflow

List available hunks, create a spec file, then apply:

# See what hunks exist
jj-hunk list
jj-hunk list --format text    # Human-readable output
jj-hunk list --files          # Files with hunk counts only

# Split changes using a spec file
jj-hunk split --spec-file /tmp/spec.yaml "feat: add feature A"

# Commit specific files, leave rest uncommitted
jj-hunk commit --spec-file /tmp/commit-spec.yaml "fix: handle edge case"

# Squash into parent
jj-hunk squash --spec-file /tmp/squash-spec.yaml

Targets any revision with -r <rev> (default is @ for working copy). Read specs from stdin with -.

Spec Format

YAML files control which hunks to keep or reset:

files:
  src/main.rs:
    hunks: [0, "hunk-7c3d..."] # Select by index or stable id
  path/to/skip:
    action: reset # Discard all changes
default: reset # Unlisted files get this action

hunks: [indices|ids] — select by 0-based index or stable id from jj-hunk list
ids: ["hunk-..."] — select by id string (sha256-based, stable across runs)
action: keep / action: reset — keep or discard all changes in a file
default — action for unlisted files

Use jj-hunk list --spec-template to generate an ID-based starting spec. IDs are more reliable than indices since they're stable across runs.

Example: Split by Hunk

# 1. List hunks
jj-hunk list --format text
# Output: M src/main.rs — hunk 0 insert hunk-abc123, hunk 1 insert hunk-def456

# 2. Create spec for first hunk only
cat > /tmp/split-spec.yaml << 'EOF'
files:
  src/main.rs:
    hunks: [0]
default: reset
EOF

# 3. Split — first hunk becomes commit A, rest stays in working copy
jj-hunk split --spec-file /tmp/split-spec.yaml "feat: add feature A"

Example: Commit Subset of Changes

cat > /tmp/commit-spec.yaml << 'EOF'
files:
  src/fix.rs:
    action: keep
default: reset
EOF

jj-hunk commit --spec-file /tmp/commit-spec.yaml "fix: handle edge case"

Tips

IDs are stable (sha256-based), indices are not — prefer ids for reproducibility
Use --files for a quick summary without full hunk details
Read spec from stdin: cat spec.yaml | jj-hunk commit - "message"
Use -r <rev> with split/squash to target any revision
When verifying after operations, always use the original change ID — see jj-core skill for template syntax (concat(), description field)

---name: jj-hunkdescription: Programmatic hunk selection for Jujutsu — split, commit, or squash specific hunks without interactive prompts. Use when making partial commits or selective squashes.---# jj-hunkProgrammatic hunk selection for Jujutsu. Split, commit, or squash specific hunks without interactive prompts.## Basic WorkflowList available hunks, create a spec file, then apply:```bash# See what hunks existjj-hunk listjj-hunk list --format text # Human-readable outputjj-hunk list --files # Files with hunk counts only# Split changes using a spec filejj-hunk split --spec-file /tmp/spec.yaml "feat: add feature A"# Commit specific files, leave rest uncommittedjj-hunk commit --spec-file /tmp/commit-spec.yaml "fix: handle edge case"# Squash into parentjj-hunk squash --spec-file /tmp/squash-spec.yaml```Targets any revision with `-r <rev>` (default is `@` for working copy). Read specs from stdin with `-`.## Spec FormatYAML files control which hunks to keep or reset:```yamlfiles: src/main.rs: hunks: [0, "hunk-7c3d..."] # Select by index or stable id path/to/skip: action: reset # Discard all changesdefault: reset # Unlisted files get this action```- `hunks: [indices|ids]` — select by 0-based index or stable id from `jj-hunk list`- `ids: ["hunk-..."]` — select by id string (sha256-based, stable across runs)- `action: keep` / `action: reset` — keep or discard all changes in a file- `default` — action for unlisted filesUse `jj-hunk list --spec-template` to generate an ID-based starting spec. IDs are more reliable than indices since they're stable across runs.## Example: Split by Hunk```bash# 1. List hunksjj-hunk list --format text# Output: M src/main.rs — hunk 0 insert hunk-abc123, hunk 1 insert hunk-def456# 2. Create spec for first hunk onlycat > /tmp/split-spec.yaml << 'EOF'files: src/main.rs: hunks: [0]default: resetEOF# 3. Split — first hunk becomes commit A, rest stays in working copyjj-hunk split --spec-file /tmp/split-spec.yaml "feat: add feature A"```## Example: Commit Subset of Changes```bashcat > /tmp/commit-spec.yaml << 'EOF'files: src/fix.rs: action: keepdefault: resetEOFjj-hunk commit --spec-file /tmp/commit-spec.yaml "fix: handle edge case"```## Tips- IDs are stable (sha256-based), indices are not — prefer ids for reproducibility- Use `--files` for a quick summary without full hunk details- Read spec from stdin: `cat spec.yaml | jj-hunk commit - "message"`- Use `-r <rev>` with split/squash to target any revision- When verifying after operations, always use the original change ID — see **jj-core** skill for template syntax (`concat()`, `description` field)

Frequently asked questions

Jj Hunk