COOL IDEA™: make hello-echo an executable index of licensed claims

[flyingrobots]

Observed

The repository is 31 lines of Edict application against 2,759 lines of tests and 1,652 lines of docs. That ratio is correct for a proof harness. But the claims it licenses are currently discoverable only by reading shell scripts and a CHANGELOG that has grown past 100 lines of prose.

Answering "what exactly does this repository prove, against which producers, and how strongly" requires reading everything.

Proposal

A machine-readable claim ledger — one entry per licensed claim:

- id: OBSERVATION-EVIDENCE-001
  claim: >
    A stale-basis observation settlement projects evidence derived from the
    bytes actually observed, rather than echoing the requested basis.
  witness:
    file: tests/effect-runtime.sh
    case: stale-basis-actual
  independence: I2          # see #30
  mutation:
    change: external_evidence_digest <- basis_digest
    expected: witness-fails
  producers:
    echo: c354d531...
    edict: df80f92a...

What it buys

1. Downstream consumers can cite it. Graft's Roadmap B work needs to know precisely which observation guarantees it may lean on. Right now that answer lives in a person's memory.
2. The mutation column is the interesting one. A claim whose mutation field is empty has never been falsified. That is visible at a glance instead of requiring archaeology.
3. It pairs with COOL IDEA™: name the independence level of every witness claim #30. An independence field is only meaningful once the levels are named.
4. Producer coordinates travel with the claim. A claim proven against c354d531 is not automatically a claim about any later Echo — currently nothing records that scoping.

Non-goals

Not a test framework. Not generated from the tests. Hand-maintained, reviewed like documentation, and permitted to be incomplete — an entry-less witness is not an error, it is an unindexed claim.

Acceptance

• claims.yaml (or similar) exists with entries for the A.1 witnesses.
• New witness work adds an entry in the same PR.
• The roadmap links to it instead of restating guarantees in prose.

Provenance

Surfaced by an external review of the #26/#27 session.

[coderabbitai]

🔗 Related PRs

#17 - Prove durable Hello Echo execution [merged]
#21 - Prove typed Hello Echo result recovery [merged]
#22 - Prove bounded workspace observation [merged]
#24 - Prove basis-bound validated patch application [merged]
#27 - Bind observation settlements to their basis evidence [open]

---

📝 Issue Planner

_{Check the box below or use the @coderabbitai plan command to generate an implementation plan and prompts that you can use with your favorite coding assistant.}

• Create Plan

---

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord!