From 8a97cb988471a7a2d8a04c525d77ebd6ac2a171f Mon Sep 17 00:00:00 2001 From: yifu zuo Date: Thu, 29 Oct 2026 12:31:32 -0500 Subject: [PATCH] feat(prove-it-works): add prove-it-works guardrail skill plugin Behavioral guardrail: verify every task output against the real artifact (run the feature, read the actual value, inspect the diff) before declaring done, not proxies, self-reports, or "it compiles." Adapted from the pstack skills (MIT, (c) Lauren Tan, https://github.com/cursor/plugins), Cursor-specific tooling removed. --- .factory-plugin/marketplace.json | 6 ++++++ README.md | 8 +++++++ .../.factory-plugin/plugin.json | 8 +++++++ plugins/prove-it-works/README.md | 9 ++++++++ .../skills/prove-it-works/SKILL.md | 21 +++++++++++++++++++ skills/prove-it-works | 1 + 6 files changed, 53 insertions(+) create mode 100644 plugins/prove-it-works/.factory-plugin/plugin.json create mode 100644 plugins/prove-it-works/README.md create mode 100644 plugins/prove-it-works/skills/prove-it-works/SKILL.md create mode 120000 skills/prove-it-works diff --git a/.factory-plugin/marketplace.json b/.factory-plugin/marketplace.json index 4282159..85ea6ce 100644 --- a/.factory-plugin/marketplace.json +++ b/.factory-plugin/marketplace.json @@ -53,6 +53,12 @@ "description": "Pull request lifecycle skills: create PRs with consistent conventions and follow up on them until merge-ready", "source": "./plugins/code-review", "category": "productivity" + }, + { + "name": "prove-it-works", + "description": "Guardrail that verifies task output against the real artifact before declaring done, not proxies or self-reports", + "source": "./plugins/prove-it-works", + "category": "quality" } ] } diff --git a/README.md b/README.md index 9793470..90db247 100644 --- a/README.md +++ b/README.md @@ -96,6 +96,14 @@ Skills for continuous learning and improvement. Autonomous experiment loop for optimization research. Try an idea, measure it, keep what works, discard what doesn't, repeat. Works standalone or as a mission worker. +### prove-it-works + +Guardrail skill: verify every task output against the real artifact, not proxies or self-reports. + +**Skills:** + +- `prove-it-works` - Verify against the real artifact (run the feature, read the actual value, inspect the diff), not a proxy, self-report, or "it compiles" + ## Plugin Structure Each plugin follows the Factory plugin format: diff --git a/plugins/prove-it-works/.factory-plugin/plugin.json b/plugins/prove-it-works/.factory-plugin/plugin.json new file mode 100644 index 0000000..0074542 --- /dev/null +++ b/plugins/prove-it-works/.factory-plugin/plugin.json @@ -0,0 +1,8 @@ +{ + "name": "prove-it-works", + "description": "Guardrail that verifies task output against the real artifact before declaring done, not proxies or self-reports", + "author": { + "name": "yifu zuo", + "email": "yzuo.2019@gmail.com" + } +} diff --git a/plugins/prove-it-works/README.md b/plugins/prove-it-works/README.md new file mode 100644 index 0000000..5fa952c --- /dev/null +++ b/plugins/prove-it-works/README.md @@ -0,0 +1,9 @@ +# prove-it-works + +Behavioral guardrail: verify every task output against the real artifact before declaring done. + +**Skills:** + +- `prove-it-works` - Verify against the real artifact (run the feature, read the actual value, inspect the diff), not a proxy, self-report, or "it compiles" + +Adapted from the pstack skills by Lauren Tan ([cursor/plugins](https://github.com/cursor/plugins), MIT License), with Cursor-specific tooling removed and Droid conventions applied. diff --git a/plugins/prove-it-works/skills/prove-it-works/SKILL.md b/plugins/prove-it-works/skills/prove-it-works/SKILL.md new file mode 100644 index 0000000..746ae4e --- /dev/null +++ b/plugins/prove-it-works/skills/prove-it-works/SKILL.md @@ -0,0 +1,21 @@ +--- +name: prove-it-works +description: "Apply after completing a task, before declaring done. Verify against the real artifact (run the feature, read the actual value, inspect the diff), not a proxy, self-report, or 'it compiles.'" +--- + +# Prove It Works + +Verify every task output by checking the real thing directly. Do not infer from proxies, self-reports, or "it compiles." + +**Why:** Unverified work has unknown correctness. Indirect verification (file mtimes, output freshness, agent self-reports, cached screenshots) feels cheaper than direct observation. Acting on a wrong inference costs far more than checking the source. + +Check the real thing, not a proxy: +- Check process liveness directly, not indirectly through derived state +- Read the actual value, not a cached or derived representation +- When verification fails, suspect the observation method before suspecting the system + +## Script the check when you can + +The strongest proof is a deterministic script that re-runs the same comparison, not a one-time eyeball. Write the script, run it, and keep its output as an artifact a reviewer can re-run instead of trusting your word. + +Keep the artifact visible for the human. Commit it only for large or complex work where the trail has to be auditable later, like a big port or migration (the show-me-your-work skill). diff --git a/skills/prove-it-works b/skills/prove-it-works new file mode 120000 index 0000000..d155aa8 --- /dev/null +++ b/skills/prove-it-works @@ -0,0 +1 @@ +../plugins/prove-it-works/skills/prove-it-works \ No newline at end of file