Fable Review

Iterative plan review with Fable by default

Run a bounded review loop between the active agent and an independent reviewer. The skill name records the default reviewer; it does not lock the workflow to Fable, Claude Code, Anthropic, or a particular host application.

Reviewer selection

Resolve the reviewer once before round 1, in this order:

1. Use the reviewer tool, CLI, agent, or model explicitly named by the user.
2. Otherwise use a reviewer configured by the current project or session.
3. Otherwise default to Claude Code CLI with the fable model alias, --effort high, read-only plan mode, and only read/search tools enabled.

Use the fable alias rather than pinning a version when the user asks for the latest Fable model. At skill creation time the alias resolves to Claude Fable 5, but the installed provider or harness is the authority at runtime.

Treat host and reviewer as separate choices. Claude Code, Codex, Cursor, or another Agent Skills client may host this workflow while Fable, Codex, Claude, Cursor Agent, or another isolated coding agent performs the review.

Never silently replace an explicitly requested reviewer. If it is unavailable, report the missing executable, authentication, model access, or harness capability and stop. If only the default is unavailable, declare the fallback before using the host's standard isolated read-only reviewer.

Record both the requested reviewer and the actual reviewer when the provider exposes that information. If a provider safeguard, overload fallback, router, or gateway substitutes another model, disclose it and do not label the result as a Fable review.

Harness-neutral review protocol

Use the host's native isolated-agent capability when it can select the requested reviewer and enforce read-only access. Otherwise invoke the selected reviewer through its authenticated non-interactive CLI.

The reviewer must be able to:

• read the plan and relevant repository context;
• return durable text;
• remain read-only;
• end with exactly VERDICT: APPROVED or VERDICT: REVISE.

Persistent reviewer sessions are preferred, not required. Resume the exact session when supported. For stateless reviewers, include the previous review, revised plan, and revision summary in the next request so no context is lost.

Workflow

1. Create isolated review artifacts

Create one session-scoped temporary directory using the platform's safe temporary-directory mechanism and assign its absolute path to REVIEW_DIR. Keep these files inside it:

• plan.md: the current full implementation plan;
• request-N.md: the review request for round N;
• response-N.json: structured provider output when available;
• review-N.md: the reviewer's durable text for round N.

If no plan exists in the conversation or repository, ask the user what should be reviewed.

2. Run round 1

Ask the selected reviewer to inspect plan.md plus relevant read-only repository context and evaluate:

1. correctness and goal coverage;
2. risks, edge cases, and regression potential;
3. missing steps or verification;
4. simpler or safer alternatives;
5. security and data-loss concerns.

Require specific, actionable feedback and one exact final verdict.

The default Claude Code command is:

claude -p \
  --model fable \
  --effort high \
  --permission-mode plan \
  --tools "Read,Glob,Grep" \
  --add-dir "$REVIEW_DIR" \
  --output-format json \
  "Read $REVIEW_DIR/request-1.md and review the plan it references. End with exactly VERDICT: APPROVED or VERDICT: REVISE." \
  > "$REVIEW_DIR/response-1.json"

Extract the response text to review-1.md and capture the exact session identifier from the structured output. For example, when jq is available:

jq -r '.result' "$REVIEW_DIR/response-1.json" > "$REVIEW_DIR/review-1.md"
REVIEW_SESSION_ID=$(jq -r '.session_id' "$REVIEW_DIR/response-1.json")

Use the user's requested model or command instead when supplied. Never resume a global “last session” during concurrent work.

3. Present and classify the review

Show each round with reviewer identity:

## Plan review - Round N
Reviewer requested: <tool/model>
Reviewer reported: <actual tool/model when available>

<feedback>

• APPROVED: finish.
• REVISE: continue.
• Missing verdict with no actionable concern: treat as approved and note the inference.
• Missing verdict with actionable concerns: treat as revise.
• Five rounds reached: stop and report unresolved concerns.

4. Revise the plan

The active agent—not a hard-coded Claude, Fable, or Codex role—must verify each finding and revise plan.md:

• apply confirmed improvements;
• reject stale, incorrect, or requirement-conflicting suggestions with reasons;
• preserve the user's explicit constraints;
• summarize each change for the next reviewer round.

5. Re-review

If the reviewer supports session resume, resume the captured session and point it to the revised plan and change summary. For default Fable through Claude Code:

claude -p \
  --resume "$REVIEW_SESSION_ID" \
  --model fable \
  --effort high \
  --permission-mode plan \
  --tools "Read,Glob,Grep" \
  --add-dir "$REVIEW_DIR" \
  --output-format json \
  "Read the revised plan and revision summary in $REVIEW_DIR/request-N.md. Re-review and end with exactly VERDICT: APPROVED or VERDICT: REVISE." \
  > "$REVIEW_DIR/response-N.json"

Extract the new response to review-N.md. If resume is unavailable or fails, start a fresh read-only review using a request packet containing the prior response, the revised plan path, and the revision summary.

6. Report and clean up

Report requested and actual reviewer tool/model, rounds, final verdict, accepted revisions, rebutted findings, unresolved concerns, and any provider reroute. Remove only the exact temporary directory created for this review after its path has been verified.

Rules

• Maximum five rounds.
• Keep every reviewer read-only.
• Never claim approval when unresolved findings remain.
• Do not let reviewer feedback override explicit user requirements.
• Do not require slash-command syntax, a specific subagent API, or a particular host application.
• Default Fable review requires an installed and authenticated Claude Code CLI with Fable access; alternative reviewers require their corresponding tool or native harness access.