Score repository targets against rubrics and execute isolated improvement loops.
npx skills add https://github.com/mferris77/SpringBoard --skill improve-mferris77
Please help me install this Agent Skill. Skill: improve Source: https://github.com/mferris77/SpringBoard/tree/main/vscode-citadel-harness/skills/improve Command: npx skills add https://github.com/mferris77/SpringBoard --skill improve-mferris77
/improve automates the discovery, prioritization, and remediation of quality issues across a product repository by scoring against a human-authored rubric and executing targeted improvements until progress plateaus. It removes manual iteration overhead by orchestrating programmatic checks, structural analysis, perceptual evaluators, isolated experiments, and verification steps in a repeatable loop. The Skill enforces human approval for rubric changes and strong regression gates to protect security and stability.
Invoke the Skill with a target rubric slug, for example: /improve citadel --n=3 to run three improvement loops against the citadel target.
High-intent search queries and answers about installing and using this skill.
Automated quality improvement scores a repository target against a human-authored rubric, selects a high-leverage axis, executes an isolated improvement, and verifies results. It runs programmatic checks, structural analysis, and perceptual evaluators in a repeatable loop until progress plateaus.
Isolated improvement experiments use worktree-based branches to execute targeted attacks on a single quality axis. This ensures changes are contained, verified against programmatic and structural checks, and automatically reverted if regressions or security failures occur.
Campaign mode raises product quality across multiple loops using resumable state files. It supports campaign continuation, level-up snapshots, and human approval workflows for rubric re-anchoring when progress plateaus.
Prioritizing quality remediation uses a transparent formula to select the single highest-leverage axis from an objective scorecard. It evaluates programmatic, structural, and perceptual metrics to target the most impactful improvement rather than addressing issues manually.
Behavioral simulation verifies user-facing quality improvements by testing how changes affect user-facing axes. It acts as a safety gate alongside programmatic, structural, and perceptual verifications to abort and revert any changes that introduce regressions.
Automated quality campaigns require human approval for rubric changes to ensure re-anchoring remains safe. The system enforces this workflow alongside strong regression guards to protect repository security and stability during autonomous loops.
wukangcheng2944Evaluate and improve Claude-style SKILL.md entries with Git-backed optimization.
Driving-TeacherFetch Hackle MCP experiment data, compare with snapshots, and share insights via Slack.
ZESCODEValidate task completion via multi-agent tests, linting, and requirement checks.
iabheejitScan codebases for technical debt and quality issues.
