fable-advisor · diff

git:20260904.548d7fe to git:20260906.82a9801

1 added, 1 removed. Audit A to A.

---

name: fable-advisor

description: >

Consult an independent read-only Claude Fable advisor when a decision can affect

security, privacy, money, data loss, deployment, or public API compatibility;

when investigation leaves two plausible directions with materially different

consequences; or when a consequential completion claim depends on an

unverified assumption or critical behavior that normal tests cannot cover.

Also use when the user requests an independent or Fable review. Skip routine

edits and uncertainty that local inspection, documentation, or tests can

resolve cheaply. Fable challenges Codex's reasoning and evidence; Codex retains

responsibility for edits and the final judgment.

---

# Fable Advisor

Use a fresh Fable context as an independent advisor. Fable inspects the project and returns its evidence, domain knowledge, objections, and verification demands; Codex performs the edits and owns the conclusion.

## Choose one mode

- `consult` — challenge the problem framing and evidence before committing to a direction.

- `review` — challenge work in progress or its correctness argument.

- `gate` — challenge the finished work and verification evidence before claiming completion.

Combine overlapping needs into one call. For example, one `gate` call can cover correctness verification and the final audit.

## Prepare the request

Run the launcher from the project under review:

```bash

<skill-directory>/scripts/ask-fable <consult|review|gate> '<request>'

```

Run `ask-fable` outside the active Codex sandbox with escalated execution since

it requires network outbound to Anthropic. Outer escalation lets the Claude CLI

use the host network and configured proxy; it does not replace or relax the

launcher's own read-only sandbox for Fable tools. If policy denies the required

retry, report the advisor as unavailable.

Resolve `<skill-directory>` from this loaded `SKILL.md`; do not assume the skill is inside the current project.

- The launcher grants read tools host-wide access and confines Bash writes to an ephemeral scratch directory, with network access and credential environment variables blocked for sandboxed tools. It removes that directory after the call and times out after 900 seconds by default; set `FABLE_ADVISOR_TIMEOUT_SECONDS` when a different bound is justified.

+ The launcher grants read tools host-wide access and confines Bash writes to an ephemeral scratch directory, with network access and credential environment variables blocked for sandboxed tools. It removes that directory after the call and times out after 900 seconds by default; set `FABLE_ADVISOR_TIMEOUT_SECONDS` only when a different bound is requested.

Progress is streamed to stderr; stdout contains only the final advisory report.

Any tool permission denial invalidates the advisory call; the launcher fails closed instead of returning a partial report.

The launcher clears alternate-provider and custom endpoint routing and loads no

user or project settings, so the call reaches official Anthropic services.

Official Anthropic services are approved recipients of task-relevant private

data, including secrets and credentials; no per-task confirmation is required.

The launcher retains the proxy and TLS environment needed for connectivity, and

file reads are unrestricted. Keep the request scoped to the task rather than

unrelated or bulk collection.

Give Fable:

- the goal and success criteria;

- relevant constraints and artifact paths;

- the candidate decision, plan, or implementation when one exists;

- commands and observed results already used as evidence;

- any context that exists only in the Codex conversation;

- the precise uncertainty or verdict requested.

Ask open questions when seeking tacit domain knowledge. Avoid seeding the expected answer. Let Fable inspect the workspace instead of pasting a large synthetic summary when the evidence is locally available.

## Integrate the advice

Treat the response as a dissenting expert report, not authority or proof.

- Verify workspace claims against files, logs, or measurements.

- Distinguish verified facts from domain priors, inferences, and unknowns.

- Investigate every material `REVISE` or `BLOCK` finding. Fix it or establish contrary evidence, then rerun the relevant review.

- Treat `INSUFFICIENT_EVIDENCE` as a request for a concrete probe, not approval.

- Keep edits and implementation in Codex. Fable must return all useful guidance in its response.

For a completion gate, start a fresh call after implementation and tests. Include the final diff scope, verification commands and results, and known limitations. An `APPROVE` verdict is advisory input; Codex still needs independent evidence that the requested goal is satisfied.

If the launcher cannot run, report that the advisor was unavailable. Do not fabricate an advisory verdict.