TELOS

TELOS (Telic Evolution and Life Operating System) — Reads, updates, and backs up the user's personal life OS stored in ~/clawd/telos/. Provides missions, goa...

TELOS appears to do what it says: it reads/writes local TELOS files, creates backups/snapshots, and can optionally install a hook to auto-inject user context at session start. Before installing, consider the following: (1) Privacy: the skill reads deeply personal files (missions, traumas, etc.). Installing the hook will make those files available automatically at each session — opt in only if you want that. (2) Paths: the code respects OPENCLAW_WORKSPACE/C LAWD_WORKSPACE and falls back to common defaults; however the docs mix references to ~/clawd/telos/ and $OPENCLAW_WORKSPACE/telos/ — verify which path will be used on your system so data ends up where you expect. (3) Inspect code before copying: the hook and scripts are included in the package and run locally; review them (they are pure fs operations) before copying to ~/.openclaw/hooks or executing with bun. (4) Runtime: you need bun (for the scripts) and node (for the hook) if you plan to run scripts or install the hook. (5) Trauma file behavior: TRAUMAS.md is protected and requires explicit triggers per the changelog/hook. (6) No network exfiltration was found, and no secrets are requested — if you see any later changes that add network calls or credential requests, treat them as high risk. If you want minimal risk, do not install the hook and run update/init/backup scripts manually when you want them to run.

Type: OpenClaw Skill Name: telos Version: 1.0.2 The 'telos' skill is a personal life-management framework (Life OS) that allows the AI to maintain context on a user's goals, beliefs, and missions. It manages Markdown files locally in the user's workspace and includes utility scripts for initialization, updates, and backups (scripts/init-telos.ts, update-telos.ts, backup-telos.ts). While the skill handles sensitive personal data and uses a hook (hooks/telos-context.js) to inject this data into the system prompt, its behavior is transparently documented, entirely local, and strictly aligned with its stated purpose. The presence of security-focused updates in the CHANGELOG.md, such as restricting the automatic injection of sensitive 'trauma' data, further supports a benign classification.