Harness audit
File-native project continuity and engineering standards for coding agents.
npx -y skills add cekrauseee/harness --skill harness-auditAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- 23 days oldThe repository was created 23 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Audit and safely migrate Harness identity, bindings, semantic records, catalogs, sessions, locks, budgets, and global storage. Use when project continuity is stale, noisy, missing, interrupted, legacy, or due for maintenance.
SKILL.md
2.7 KB, 532 tokens by cl100k_base, as published. Nobody here has run it
Harness Audit
Inspect the global Harness without changing project code or documentation.
Run an audit
Audit the current project:
python3 scripts/audit.py --project <project-path> --json
Audit every global project:
python3 scripts/audit.py --all --json
Use --project <project-path> for the canonical project-native form; --repo remains a compatibility alias. The audit reports:
- malformed or mismatched project identities and bindings;
- missing required global container paths;
- invalid memory, candidate, archive, or session JSON;
- catalog drift from active topic memory;
- empty, stale, or excessive active sessions and locks;
- semantic records missing useful titles, summaries, read rules, or tags;
- oversized or unrecallable memory and recall-card accounting overhead;
- registered project paths that no longer exist;
- any Harness-owned
.harnessor.project-harnessdirectory found at a registered project root.
Use --repair-catalog only to atomically rebuild the derived catalog.jsonl. It does not classify memory, close sessions, remove locks, or edit repositories.
Migrate legacy state
Preview every planned change first:
python3 scripts/audit.py --all --migrate --dry-run --json
Apply the deterministic schema migration:
python3 scripts/audit.py --all --migrate --json
Migration creates one timestamped backup below ~/.harness/backups/ before changing any selected project. It adds project-native bindings and compact semantic fields, rebuilds the derived session-and-memory catalog, and moves empty or stale active sessions to sessions/dormant/. It preserves full memory content and every legacy session. If a live and closed session share an old ID, it preserves the live ID, deterministically rekeys the closed copy, and records legacy_id. Non-UUID legacy IDs are converted the same way so core recall can hydrate them by canonical ID.
Run migration only when the user authorized state changes. Prefer --project <path> or --project-id <uuid> for one project. A repeated migration with no pending changes creates no backup and changes nothing.
Interpret severity
error: integrity is broken; do not trust affected recall until repaired.warning: state may be stale or require human classification.info: maintenance counts or healthy observations.
The command exits nonzero only when errors exist. Read references/audit-contract.md before adding new repair behavior.
What ships with it: 3 files
45.1 KB alongside SKILL.md, 1 of them executable
agents/
- openai.yaml303 B
references/
- audit-contract.md1.8 KB
scripts/
- audit.pyruns43.0 KB
Gives 0 of the 12 instructions most audit compliance skills give in 532 tokens
Counted across 937 of the 1,487 authors here whose files we hold, read 2026-08-07
- Fetch latest guidelines before each reviewin 43 of 937, across 3 files
- Group findings by severityin 43 of 937
- Check files against all fetched rulesin 42 of 937, across 2 files
- Output findings in terse file:line formatin 41 of 937, across 3 files
- Ask user which files to review if none specifiedin 41 of 937, across 3 files
- Read specified files or prompt user for filesin 39 of 937, across 1 file
- Generate the audit reportin 33 of 937, across 30 files
- Assign a severity to every findingin 25 of 937
- Run automated accessibility scansin 23 of 937, across 13 files
- Output a markdown audit reportin 22 of 937
- Map findings to WCAG criteriain 20 of 937, across 10 files
- Confirm audit scopein 19 of 937, across 9 files
Said here and by no other author read
- inspect global harness without changing project code
- audit the current project using the audit script
- audit all global projects when needed
- use the project path for canonical project-native form
- use repair catalog only to atomically rebuild catalog.jsonl
- preview every planned migration change first
Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.