plutonicsauce/Escrow
Verify coding-agent instructions against deterministic repository evidence.
View on GitHubTrust Signals
- Scorecard Score
- not yet scored
- Maintenance Recency
- Stale
- License
- None
Inputs
| name | description | required | default |
|---|---|---|---|
| openai-api-key | OpenAI API key used only by the hosted Codex claim-extraction process. Leave empty for local Ollama or LM Studio mode. | no | — |
| local-provider | Optional local provider: ollama or lmstudio. Requires a self-hosted runner that can reach it. | no | "" |
| model | Codex model used to interpret repository instructions. | no | gpt-5.6-luna |
| target | Optional repository-relative directory whose instruction scope should be checked. | no | "" |
| execute | Run documented commands in temporary Git worktrees. | no | true |
| allow-network | Allow documented commands that need network access. | no | false |
| report-directory | Relative directory in the checked-out repository for generated reports. | no | .escrow-artifacts |
| codex-version | Version of @openai/codex to install (latest is the default). | no | latest |
Outputs
| name | description |
|---|---|
| status | Escrow CLI exit code (0 passes, 1 stale instructions, 2-4 operational errors). |
| json-report | Repository-relative JSON report path. |
| markdown-report | Repository-relative Markdown report path. |
| html-report | Repository-relative HTML report path. |
| summary-report | Repository-relative concise CI summary path. |