Delegate Fable

Delegate to Fable

Delegate through Herdr with one explicit mode:

• review: Keep Fable read-only. Use for plans, diffs, bugs, architecture, tests, or independent verification.
• implement: Make Fable the sole writer. Stop editing until Fable finishes, then inspect and verify its work.

Default to review when the user's authorization is ambiguous. Never infer permission to edit from phrases such as "take a look" or "check this".

Prepare the handoff

Create a self-contained task containing:

• the objective and relevant project context;
• the exact scope and files when known;
• current evidence, errors, or competing hypotheses;
• constraints and explicit non-goals;
• required verification and return format.

Do not forward the entire conversation. Preserve only information needed to complete the delegated task.

Before implement, inspect the current workspace state and preserve unrelated or pre-existing changes. Tell Fable not to commit, check in, shelve, push, switch branches, or alter workspace configuration unless the user explicitly requested it.

Run the delegation

Resolve this skill's directory and invoke its script. Pass the task with --task, --task-file, or stdin.

bash <skill-dir>/scripts/delegate-fable.sh review \
  --target-title "Fable 5 Reviewer" \
  --task "Review the current change for correctness, regressions, and missing tests."

bash <skill-dir>/scripts/delegate-fable.sh implement \
  --task "Implement the agreed fix within the stated scope and run the relevant verification."

The script must:

1. Honor --target first, then --target-title. A requested target is mandatory: if it is missing or busy, stop instead of falling back to another Cursor pane.
2. Without an explicit target, prefer a settled pane titled Fable 5 Reviewer, then reuse another idle Fable Cursor Agent in the current Herdr workspace. Split only when no Fable pane exists.
3. Submit the task, handle the known Cursor Enter fallback, wait for settlement, and return visible output.
4. Leave the Fable pane open for follow-up work.

Do not manually duplicate these steps unless the script reports a concrete failure.

Default to one complete Fable turn per hypothesis. If the response lacks the required handoff sections, report an incomplete handoff; do not start a new pane or automatically send an open-ended continuation. On timeout, send Escape to cancel the active turn, collect its last output, and stop.

Handle the result

For review, report Fable's findings with your own assessment. Do not implement findings unless the user separately authorizes changes.

For implement:

1. Resume ownership only after Fable settles.
2. Inspect the actual changed files and version-control state.
3. Run proportionate verification independently.
4. Report what Fable changed, what you verified, and any remaining risk.

If Fable is blocked, show the blocking output to the user. Never auto-approve permissions, destructive actions, commits, check-ins, shelvesets, pushes, or branch changes.

If the script reports that no Herdr environment is available, stop and explain that the skill must run from an agent pane inside Herdr.