Plugin Claude Code
Orchestrates an evaluate -> plan -> calibrate -> re-evaluate loop over your Claude Code setup (CLAUDE.md, rules, settings, skills, subagents, hooks, MCP, plugins). Three-layer architecture: per-feature calibration skill bundles (templates/examples/scripts/reference) -> three worker agents (planner, evaluator…
Agent Claude Code
Author-only maintenance worker for the claude-calibration repo. Fetches ONE official Claude Code documentation page (or agents.md) over the network and returns its current content as clean text. Use only when the /docs-update skill needs a single official doc page refreshed — one page per call, so the parent can diff…
Hook Claude Code
Runs before the agent uses a tool for Edit, Write and MultiEdit tool calls, executing plugin-dev-guard.sh. From odere-pro/claude-calibration.
Settings file Claude Code
Agent settings declaring 1 hook event (PreToolUse) and 13 allowed tools.
Skill Claude CodeCodex
Author-only maintenance skill for the claude-calibration repo. Read-only staleness report for the docs/ rubric: for every page with a ## Sources block it lists the official source URLs, the git last-touched date, and a staleness flag. Use before /docs-update to see which rubric pages may have drifted from the upstream…
Skill Claude CodeCodex
Author-only maintenance skill for the claude-calibration repo. Use when the docs/ rubric may have drifted from the official Claude Code docs. Refreshes the docs/ rubric against the official Claude Code docs: spawns the docs-fetcher subagent (one page at a time), diffs the live facts against each local page, and…
Skill Claude CodeCodex
Author-only maintenance skill for the claude-calibration repo. Realigns the plugin's own components with the now-current docs/ rubric: walks each skills/calibrate-/reference.md, its templates/, and scripts/lint.sh constants, and updates any limits / required fields / signatures that drifted from the matching…
Instructions file
Claude Code instructions for odere-pro/claude-calibration, covering claude-calibration — plugin development, what ships, what doesn't, pointers and agent routing.
Agent
The five worker subagents the skill layer dispatches. They ship with the plugin and are invoked only by the skill layer — never by the user directly, never auto-fired.
Agent
Applies an approved calibration improvement plan to a Claude Code setup — project-scope changes only; user-scope changes are written up as recommendations, not applied. Surgical edits, a change report, and a record of what was touched. Invoked only by the /calibrate orchestrator after the user approves the plan. Never…
Agent
Use this agent when auditing a Claude Code setup against the calibration rubric. Two passes — 1 (baseline) writes per-feature, interactions, and intent-flow reports under the run folder; 2 (delta) re-audits the same scope and writes a delta report comparing against the baseline. Dispatches per-feature work to the…
Agent
Per-feature worker for the calibration evaluator's parallel fan-out. Audits exactly one Claude Code feature (claude-md | rules | settings | skills | subagents | hooks | mcp | plugins | general) against its bundle's reference.md + scripts/, writes a slim draft section to /.drafts/feat- .md, and returns one summary…
Agent
Behavioural worker for the /calibration-flow flow. Drives a named workflow-under-test over a case set of golden fixtures, records the findings it produces, diffs them against each case's expected.md oracle via the deterministic score-flow.sh, and writes eval-flow- .md (node recall/precision, edge handoff contracts…
Agent
Writes and updates plan.md for a calibration run. Two modes: init (creates the run folder, writes the skeleton plan.md, records the intent) and improve (reads the evaluator's baseline reports, groups findings by pattern signature for the recurrence detector, and emits a prioritised improvement plan with kind: edit…
Hook
Runs when a session starts, executing install-workflows.sh. From odere-pro/claude-calibration.
Skill Claude CodeCodex
Audits CLAUDE.md, CLAUDE.local.md, and every nested /CLAUDE.md across user / project scopes. Flags committed secrets, files over 200 / 400 effective lines, aspirational ("must" / "always" / "never") wording that should be a hook, vague rules without specifics, @-import chains > 5 hops, contradictions between nested…
Skill Claude CodeCodex
Use this bundle when cross-cutting calibration findings span more than one Claude Code feature. Rolls up estimated always-on context cost (CLAUDE.md + unconditional rules), nested-CLAUDE.md conflict risk, settings precedence surprises, missing .gitignore coverage for CLAUDE.local.md / .claude/settings.local.json /…
Skill Claude CodeCodex
Audits and tunes hooks across user / project / plugin-shipped layers — both the hooks blocks inside any settings.json and standalone hook scripts under .claude/hooks/ or a plugin's hooks/. Flags bare- matchers on hot events (PreToolUse/PostToolUse/UserPromptSubmit), enforcement scripts that use exit 1 (non-blocking…
Skill Claude CodeCodex
Audits MCP server configuration across every layer Claude Code reads — project .mcp.json, user /.claude.json's mcpServers block, and per-agent mcpServers: frontmatter inside .claude/agents/.md. Flags literal tokens in committed JSON (the high-impact security finding), servers with no paired wrapper skill (always-on…
Skill Claude CodeCodex
Audits installed plugins, registered marketplaces, and (when the project itself is a plugin) the local manifest at .claude-plugin/plugin.json. Flags manifests missing name or version, manifests that don't parse, components misplaced under .claude-plugin/ instead of at the plugin root, plugins that use legacy commands/…
Skill Claude CodeCodex
Audits and tunes every .claude/rules//.md across user / project / plugin-self. Flags oversized files (> 200 effective lines), language- or area-specific rules missing paths: frontmatter (always-on context cost), plugin-shipped rules without paths: (load for every user who enables the plugin), broken globs…
Skill Claude CodeCodex
Audits and tunes every settings.json Claude Code reads — user (/.claude/settings.json), user-local (/.claude/settings.local.json), project (.claude/settings.json), project-local (.claude/settings.local.json), and plugin-self when the project is itself a plugin. Flags committed secrets, --dangerously-skip-permissions…
Skill Claude CodeCodex
Audits every SKILL.md across user / project / plugin-self / plugin-cache. Flags missing or vague frontmatter (name, description), descriptions that bust the 1,536-char routing budget, bodies over 500 lines, side-effecting verbs (deploy / commit / push / publish / release / delete / post) without…
Skill Claude CodeCodex
Audits every subagent .md across user / project / plugin-self / plugin-cache. Flags missing frontmatter (name, description, tools), the big footgun (no tools: → inherits ALL tools including every MCP server), bodies over 200 lines, omitted model: (defaults to inherit, silently inflates cost), vague descriptions Claude…