Cost and Telemetry
See what a session cost, what it did, and what the kit contains.
Read-only reporting over logs the harness already writes. Nothing here changes project state.
budget
View token usage and cost tracking.
npx mewkit budget [--monthly] [--session [id]] [--day [YYYY-MM-DD]]| Flag | Description |
|---|---|
--monthly | Aggregate by month instead of showing the last 10 entries |
--session | Filter to the current Claude Code session id |
--session <id> | Filter to a specific session id from cost-log.json |
--day | Filter to today using YYYY-MM-DD |
--day <date> | Filter to a specific day in YYYY-MM-DD format |
Behavior:
- Without filters,
budgetshows the latest 10 rows and the all-time token total. --sessionresolves the current session fromHOOK_SESSION_IDfirst, thensession-state/last-session-id.--daywithout a value uses today.- Filters compose, so
--session --dayreports the intersection.
Examples:
npx mewkit budget
npx mewkit budget --monthly
npx mewkit budget --session
npx mewkit budget --session 9adf6a6b-9fbf-45ca-b36b-b0dd55356ac1
npx mewkit budget --day
npx mewkit budget --day 2026-05-01
npx mewkit budget --session --day 2026-05-01budget context
Estimate the loadable .claude/ context size for a profile (the context-budget guardrail; also wired into CI).
npx mewkit budget context --profile <core|developer|full> [--fail-over N] [--json]--fail-over N exits non-zero when the estimated loadable context exceeds N tokens.
trace
Inspect the append-only trace log for entropy and recurring friction, and record friction inline so future sessions can recall it. Advisory — always exits 0.
npx mewkit trace audit # orphaned / stale / unverified / repeated-friction signals
npx mewkit trace propose # group repeated friction (≥2) into advisory backlog items
npx mewkit trace score # entropy score
npx mewkit trace --friction "<note>" [--responsibility <r>] [--commit]index
Build or refresh the derived SQLite index (.meowkit/cache/wiki-index.db) from the canonical append logs (.meowkit/memory/*.jsonl) and the wiki tree. Disposable and deterministic — delete it and re-run to get an identical index.
npx mewkit index [--json]query
Run read-only aggregate queries over the derived index (events by type, runs, friction by responsibility, cost by model). Requires a prior mewkit index.
npx mewkit query [--json]inventory
Report harness-artifact coverage. With --substrate, map artifacts onto the responsibility taxonomy (covered / partial / missing); --emit rewrites the generated harness-substrate.md view.
npx mewkit inventory [--json]
npx mewkit inventory --substrate [--emit]
npx mewkit inventory --stale | --critical | --portable-missing | --check | --emit-countscapabilities
Inspect and resolve the capability manifest — the semantic map of installed skills/agents/commands/hooks plus authored tool and context/state services. The manifest is a resolution surface; it is never injected into a model session. The CLI reads the live .claude/ install when present; a Codex-migrated project without it falls back to the validated .codex/capabilities.json projection written by mewkit migrate codex.
npx mewkit capabilities list [--json]
npx mewkit capabilities explain <id> [--json]
npx mewkit capabilities resolve --intent "implement this feature" [--provider <id>] [--json]
npx mewkit capabilities view # render the trigger-registry table from the manifest
npx mewkit capabilities bootstrap [--provider <id>] # the always-visible discovery bootstrap
npx mewkit capabilities projections [--json] # per-provider discovery + 4 support levels| Subcommand | Purpose |
|---|---|
list | All capabilities with kind, owner, install state, and intents. |
explain <id> | One capability's requirements, support levels, verification, and provenance. |
resolve | Rank capabilities for an --intent and, with --provider, report host availability, the safe invocation, any repo-context requirement + how that provider acquires it, and the adapter + evidence behind the claim. In a Codex-only project, resolve uses the validated migration snapshot. |
view | Render the capabilities table (the generated portion of the trigger registry). |
bootstrap | Print the bounded, brand-neutral discovery bootstrap for a provider (--write regenerates the committed file). |
projections | Per-provider discovery projection + the four support levels (discoverable/selectable/invocable/enforceable). |