Strava

Load and analyze Strava activities, stats, and workouts using the Strava API

This skill appears to do what it says (call the Strava API and refresh tokens) but there are metadata inconsistencies and sensitive credential handling to consider before installing: - Metadata mismatch: the registry claims no required env vars but the SKILL.md and included script clearly need STRAVA_ACCESS_TOKEN, STRAVA_REFRESH_TOKEN, STRAVA_CLIENT_ID and STRAVA_CLIENT_SECRET. Ask the publisher to correct the manifest so required permissions are explicit. - Secrets handling: the README/SKILL.md suggests storing tokens in ~/.clawdbot/clawdbot.json (plaintext). The included refresh_token.sh prints new tokens to stdout rather than securely storing them. Prefer setting secrets via environment variables or a secure secrets store; if you must use the config file, restrict its filesystem permissions (chmod 600). - Review the script before running: the refresh script is short and readable, but you should inspect it locally to confirm it only calls Strava and does not send data elsewhere. - Trust and provenance: the source is listed as unknown. Only provide your Strava client secret and refresh token to skills from publishers you trust. If you have doubts, create a Strava app with limited scope or a throwaway account for testing. If the publisher updates the registry metadata to accurately list required env vars and documents secure handling of tokens (or implements a safe automatic config update mechanism), the concerns would be reduced.

Type: OpenClaw Skill Name: strava Version: 1.0.0 The skill bundle is benign. It provides a legitimate integration with the Strava API, requiring standard OAuth credentials (STRAVA_ACCESS_TOKEN, STRAVA_REFRESH_TOKEN, STRAVA_CLIENT_ID, STRAVA_CLIENT_SECRET) for its stated purpose. All `curl` commands in SKILL.md and scripts/refresh_token.sh are directed at official Strava API endpoints. There is no evidence of data exfiltration beyond necessary API interaction, malicious execution (e.g., `curl|bash` from external sources), persistence mechanisms, obfuscation, or prompt injection attempts against the agent.