Parable
Working-style rules that move agent deliberation out of chat into a notes file
Test report
- Verdict
- Works with setup
- Score
- Tested
- Jul 29, 2026
- Environment
- Claude Code 2.x (agent harness)
- Upstream re-checked
- Jul 30, 2026 · 0996aca
⚠ This skill is no longer available upstream. Our re-check on Aug 10, 2026 couldn't find it any more (repo unreachable/deleted). The test below is what we measured on Jul 29, 2026 and we're leaving it up as a record — but there is nothing left to install, so we've removed the command.
Tested it on a verification job whose answer was unknown up front - does the repo's own rule-refresh hook behave as the README claims - and the difference showed up in evidence, not in the conclusion. The unaided baseline answered from reading the script; the skill's Gate 3 and Gate 4 rules made the same run execute five payload variants, confirming the hook fires once then stays silent for 300 seconds, writes a per-session stamp file, falls back to a shared "global" stamp when no session id is present, and is not hijacked by a session-id lookalike inside a tool response. The real friction is setup: three installed files (SKILL.md, director.md, PROMPTING.md) still contain unfilled {{NOTES_DIR}}, {{BUILD_CHECK}} and {{PROJECT_ROOT}} placeholders you must edit by hand, and the mechanism the README calls most important - periodic rule re-injection - requires manually merging a hooks block into .claude/settings.json. Documentation is unusually honest for this genre: the README and EXPERIMENTS.md both state up front that the numbers come from one two-day study on one codebase by one person and should not be treated as a benchmark.
Scored on four weighted criteria — install, triggering, output vs. baseline, docs. How scoring works
- Installs cleanly 5/5
- Triggers reliably 4/5
- Output vs. baseline 6/10
- Docs & honesty 5/5
What Parable does
A working-style skill set for Claude Code agents: short status-only visible messages, deliberation appended to a scratch notes file, a five-gate loop (scope, evidence, adversarial, verify, report) and per-task addenda for debugging, features, research and planning. It also ships optional hooks that re-inject a rule card every prompt and every five minutes so long runs do not drift. Load it by name when you want an agent to plan visibly, verify by execution and report with numbers instead of narrating.
How to install Parable
Nothing to install: the source repository no longer has this skill. If the author brings it back, our daily re-check will pick it up and the command will reappear here.
Commands — how to trigger Parable
-
/parableWorking-style rules that move agent deliberation out of chat into a notes file
It also activates on plain-language prompts like these:
-
Use parable mode so you plan visibly and verify everything by execution -
Work like a director agent here: plan, delegate, then run a review gate -
Debug this flaky test and keep your deliberation in the scratch notes file
Frequently asked questions
- Is the Parable skill free?
- Yes. The skill itself is free from SamBWarren/parable. SkillProof publishes the install command and an independent test verdict at no cost.
- Does Parable work with Claude Code?
- We tested it with Claude Code 2.x (agent harness) on Jul 29, 2026. Verdict: Works with setup. Tested it on a verification job whose answer was unknown up front - does the repo's own rule-refresh hook behave as the README claims - and the difference showed up in evidence, not in the conclusion. The unaided baseline answered from reading the script; the skill's Gate 3 and Gate 4 rules made the same run execute five payload variants, confirming the hook fires once then stays silent for 300 seconds, writes a per-session stamp file, falls back to a shared "global" stamp when no session id is present, and is not hijacked by a session-id lookalike inside a tool response. The real friction is setup: three installed files (SKILL.md, director.md, PROMPTING.md) still contain unfilled {{NOTES_DIR}}, {{BUILD_CHECK}} and {{PROJECT_ROOT}} placeholders you must edit by hand, and the mechanism the README calls most important - periodic rule re-injection - requires manually merging a hooks block into .claude/settings.json. Documentation is unusually honest for this genre: the README and EXPERIMENTS.md both state up front that the numbers come from one two-day study on one codebase by one person and should not be treated as a benchmark.
- What is the Parable SkillProof Score?
- 8.0/10 — installs cleanly 5/5, triggers reliably 4/5, output vs. baseline 6/10, docs & honesty 5/5.
- How do I install Parable?
- Copy the install command from this page, run it in your terminal, and restart Claude Code. Skills live in ~/.claude/skills/ (global) or .claude/skills/ inside a project.
- Can I use Parable with Cursor, Copilot, Gemini CLI, Codex or other AI tools?
- The SKILL.md format is native to Claude (Claude Code, Desktop, claude.ai). The instructions inside adapt to other assistants: Cursor rules, GitHub Copilot instructions, Windsurf rules, Custom GPTs, AGENTS.md for OpenAI Codex, and GEMINI.md for Google Gemini CLI — our conversion guides cover each, and the free converter on the tools page does the wrapping for you.