skills/delta-logger/SKILL.md
Logs every state change — file edits, exec results, config changes, API calls — into VAULT999 with before/after state and reversibility flag. Activates after every exec, write, edit, or delete operation. Makes rollback reasoning concrete and creates an auditable delta chain for the session.
npx skillsauth add ariffazil/openclaw-workspace delta-loggerInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Every time the agent changes something in the world (file, config, external state), log it. This creates a VAULT999-attached delta chain that makes rollback concrete, not philosophical.
Log ALL of these after any state-changing operation:
write tool — new file created or overwrittenedit tool — file modifiedexec with side effects — cp, mv, rm, tee, curl to external systemsconfig.patch or config.apply — configuration changesgateway restart — runtime state changeDELTA #[N] | YYYY-MM-DD HH:MM UTC
Action: [write | edit | exec | config | api | delete]
Target: [file path or API endpoint]
Before: [state before — can be "NEW" or "NONE" for first creation]
After: [state after — what changed]
Reversible: [YES | PARTIAL | NO]
Rollback: [how to undo — command or procedure]
Risk: [TRIVIAL | REVERSIBLE | CRITICAL | IRREVERSIBLE]
Arif notified: [YES | NO]
---
| Reversible | How |
|---|---|
| YES | Single command can undo (e.g., cp backup original) |
| PARTIAL | Can mostly undo but some state may be lost |
| NO | Cannot undo — external system, permanent deletion, or multi-step cascade |
For fast logging during a session, append to memory/vault999-triage.md:
HH:MM UTC | DELTA | [action] | [target] | rev:[YES/PARTIAL/NO] | [brief description]
At session end, consolidate all triage deltas into memory/vault999.md with full entries:
## SESSION DELTA LOG — [session-id] — [date]
### Files Created
| Target | Reversible | Rollback |
|---|---|---|
| [path] | YES | [cmd] |
### Files Modified
| Target | Before-SHA | After-SHA | Reversible | Rollback |
|---|---|---|---|---|
| [path] | [hash] | [hash] | PARTIAL | [cmd] |
### External Actions
| Action | Target | Risk | Arif Notified |
|---|---|---|---|
| [api call] | [endpoint] | CRITICAL | NO — used mmx-quota-guard |
### Open Rollback Items
- [item] — [rollback procedure pending]
❌ "I deleted some old memory files to clean up" ✅ Actually logged: "DELETE | /root/.openclaw/workspace/memory/old.md | was 2.3KB | Reversible: NO | Rollback: git checkout"
❌ "Updated the config" ✅ Actually logged: "CONFIG | openclaw.json | {field: before} → {field: after} | Reversible: YES | Rollback: config.patch with old value"
❌ "Made an API call" ✅ Actually logged: "API | https://api.example.com/resource | cost: ~$0.02 | Risk: CRITICAL | Arif notified: NO — within YELLOW band"
At session start, announce:
Session delta tracking active. All state changes will be logged to VAULT999.
development
Check every skill’s “use when” and “do not use when” clauses for collisions, missing negatives, and vague verbs like “help,” “assist,” or “improve.” Load when linting, reviewing, or validating trigger boundaries.
development
Bootstrap, design, and package new skills. Load when capturing user intent for a new skill or drafting its initial instruction framework.
content-media
Diagnose which federation services are up, down, or drifting. Produce a prioritized remediation plan.
business
Scan a repo or workspace for exposed secrets, tokens, keys, and credentials. Produce a findings report with remediation steps.