seldonframe/Reelier replay
Replay a Reelier skill in CI — deterministic, 0 LLM tokens, exit 1 on drift — and post the run receipt as a job summary.
View on GitHubTrust Signals
- Scorecard Score
- not yet scored
- Maintenance Recency
- Stale
- License
- None
Inputs
| name | description | required | default |
|---|---|---|---|
| skill | Path (relative to the checked-out repo) to the .skill.md file to replay. A single path only — for multiple skills, call this action once per skill (see README.md "Run it in CI" for the matrix pattern). | yes | — |
| max-level | Escalation ceiling passed to `reelier run --max-level`. "0" (the default) never calls an LLM — pure deterministic replay, 0 tokens. "1" or "2" allow self-healing escalation and require LLM credentials to be present in the environment (see README.md "Works with any model"). | no | 0 |
| cloud-url | Base URL of the receipt-ledger endpoint to push the receipt to. Only used when `cloud-key` is also set. | no | https://www.reelier.com |
| cloud-key | Receipt-ledger API key. When set, `reelier push` runs after a successful replay so the receipt accrues in your ledger. When unset, the replay is local-only — the job summary is still written, nothing is pushed anywhere. Pass this as a GitHub secret, never inline. | no | "" |
| vars | Newline-separated `name=value` pairs, forwarded to `reelier run` as `--var name=value` for each line. Optional — most skills declare their own inputs and need none. | no | "" |
Outputs
| name | description |
|---|---|
| passed | "true" or "false" — the run record's own `passed` field. |
| run-record-path | Path to the .reelier/runs/<skill-name>.jsonl file written by this run, if any. |