tale-mode
01Plugin Claude Code
Tale Mode — a discipline plugin for complex/high-stakes work, plus a self-armed autonomous goal loop (and an optional governor companion).
Plugin Claude Code
Tale Mode — a discipline plugin for complex/high-stakes work, plus a self-armed autonomous goal loop (and an optional governor companion).
Plugin Claude Code
Optional companion to the tale-mode plugin: a read-only "governor" that breaks anchoring when the autonomous goal loop gets stuck. v2 is a single command hook on both hosts — a free bash gate at every turn-end, escalating to ONE read-only model call exactly when a goal first fails twice (Claude Code: claude -p restri.
Hook
Runs when the agent finishes a response, executing governor.sh via bash. From alicicek/tale-mode.
Plugin Claude Code
A discipline mode for complex/high-stakes work: staged planning, decisions-with-receipts, verify-against-source, parallel delegation, adversarial review. Always-on core disciplines via a SessionStart hook, plus a self-armed autonomous loop (a deterministic Stop gate) that can also arm itself from a committed…
Agent
Adversarial reviewer for plans and substantial diffs. Reads the actual code/sources and tries to break the work — finds breakage, security/privacy leaks, races, stale assumptions, ordering bugs, and untested gaps. Returns numbered, code-grounded findings. Use before shipping anything high-stakes.
Command
Start a scoped chunk of a larger plan in a fresh session — interview the user first, then implement only that chunk.
Command
Plan a task to a high bar — verify against the code, receipts on every decision, adversarial review, runnable verification gates.
Skill Claude CodeCodex
Explicitly end a tale-mode build phase by clearing the phase marker(s) under /.claude/, which stops the Stop-hook's committed-gate enforcement for this repo. Invoke ONLY when the user explicitly asks to end the phase, stop or disarm the tale-mode phase loop, or clean up phase markers — never proactively, and never as…
Skill Claude CodeCodex
Build ONE scoped phase of an existing multi-phase plan in a fresh session — re-verify the plan against the live code, interview the user sharply on real forks, then implement only that phase behind plan-mode approval with runnable verification and a fresh-eyes review. Invoke explicitly when starting a phase (the…
Skill Claude CodeCodex
Plan a large, multi-phase feature to a high bar — verify against the code, receipts on every decision, an adversarial fresh-eyes review, and runnable verification gates, decomposed into independently-shippable phases. Invoke explicitly when starting a big multi-session build (the cross-platform form of the…
Skill Claude CodeCodex
Scan a repo's package.json scripts, CI workflows, and equivalent manifests, then SUGGEST a committed .claude/tale-mode.json gate list for tale-mode's Stop-hook loop — suggestions only; this skill never writes the config unasked, never trusts it, and never arms anything. Invoke ONLY when the user explicitly asks to…
Skill Claude CodeCodex
Make Claude work rigorously on complex or high-stakes tasks instead of one-shotting them. Triggers: "tale mode", "tale on" — or self-activates on hard multi-step work. Enforces staged planning, decisions-with-receipts, verifying claims against the real source (not memory), parallel delegation, an independent…
Skill Claude CodeCodex
Reference for tale-mode's two user-only opt-in files — the committed-gate trust store (/.claude/tale-mode-trust) and the Codex cwd-root opt-in (/.tale-mode-allow-cwd-root) — and their security model. Invoke ONLY when the user explicitly asks how to trust a repo's committed tale-mode gates, about the "content-hash is…