Skill434 repo starsupdated 3d ago
fitness
Measure declared project fitness goals Triggers: "fitness", "check project fitness", "measure goals".
Install in Claude Code
Copygit clone --depth 1 https://github.com/boshu2/agentops /tmp/fitness && cp -r /tmp/fitness/skills-codex/fitness ~/.claude/skills/fitnessThen start a new Claude Code session; the skill loads automatically.
Definition
SKILL.md
# Fitness — read-only goal measurement Inspect the active goals document and run only the caller-selected measurement, validation, drift, history, export, or meta-goal command. Renamed from `goals` (2026-07-29): the semantic skill is `fitness`; the `ao goals` CLI command family is a separate product surface and keeps its name. A thin `goals` compatibility alias resolves to this skill. Measurement stays trustworthy only because it cannot mutate what it measures; the moment a fitness report edits a goal, the next report measures the editor, not the project. Named failure mode — **advice creep**: a measurement report that ends with "you should…" has silently become work selection. Anti-pattern: padding the report with recommendations to look helpful. Corrective: return the numbers, the evidence gaps, and checked/not-checked scope, and let the caller decide. ## Boundary - Prefer `GOALS.md` when both Markdown and legacy YAML exist. - Preserve stable directive and gate identities in the report. - Every measured gate must name its executable check and observed outcome. - Do not add, remove, prioritize, recommend, apply, prune, migrate, or otherwise mutate goals. - Do not translate a fitness gap into work selection or a next action. - No subcommand edits the goals source through its own logic. `measure`, `drift`, and `export` persist a best-effort JSON snapshot under the fixed derived path `.agents/ao/goals/baselines/`. `render --out <file>` writes a Gherkin spec to whatever path the caller names — the CLI does not constrain it, so never point `--out` at the goals source or any non-derived file. ## Commands All eight subcommands read the goals source without mutating it. The snapshot write (fixed derived path) and the `render --out` write (caller-chosen path, caller's responsibility) are the only side effects. ```bash ao goals measure --json ao goals validate --json ao goals drift ao goals history ao goals export ao goals meta --json ao goals scenarios ao goals render # append --out <file> to write the spec instead of stdout ``` Run the requested command once. Return the command, exit code, goal-level results, aggregate measurement, missing evidence, and checked/not-checked scope. Then stop.
More from this repository
agent-mailSkill
Use Agent Mail as an optional messaging and Triggers: "coordinate writers", "reserve files".
beads-brSkill
>-
beads-bvSkill
>-
beads-workflowSkill
Use when converting markdown plans into br beads with dependencies for implementation or swarm execution.
caamSkill
Use when switching AI coding CLI accounts quickly to recover from subscription rate limits or OAuth friction.
casrSkill
>-
cass-memorySkill
Use when starting non-trivial work, mining lessons, or preventing repeated mistakes with cm procedural memory.
cassSkill
Mine past agent sessions for working Triggers: "cass", "mine past agent sessions for", "cass skill".